fix: PR review — DB reload, market-local date, market-scoped scan_candidates

Address PR #110 review findings:

1. High — Realtime mode now loads playbook from DB before calling Gemini,
   preventing duplicate API calls on process restart (4/day budget).
2. Medium — Pass market-local date (via market.timezone) to
   generate_playbook() and _empty_playbook() instead of date.today().
3. Medium — scan_candidates restructured from {stock_code: candidate}
   to {market_code: {stock_code: candidate}} to prevent KR/US symbol
   collision.

New test: test_scan_candidates_market_scoped verifies cross-market
isolation.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.env.example

@@ -16,8 +16,9 @@ CONFIDENCE_THRESHOLD=80
 # Database
 DB_PATH=data/trade_logs.db
 
-# Rate Limiting
-RATE_LIMIT_RPS=10.0
+# Rate Limiting (requests per second for KIS API)
+# Reduced to 5.0 to avoid "초당 거래건수 초과" errors (EGW00201)
+RATE_LIMIT_RPS=5.0
 
 # Trading Mode (paper / live)
 MODE=paper
@@ -26,3 +27,10 @@ MODE=paper
 # NEWS_API_KEY=your_news_api_key_here
 # NEWS_API_PROVIDER=alphavantage
 # MARKET_DATA_API_KEY=your_market_data_key_here
+
+# Telegram Notifications (optional)
+# Get bot token from @BotFather on Telegram
+# Get chat ID from @userinfobot or your chat
+# TELEGRAM_BOT_TOKEN=1234567890:ABCdefGHIjklMNOpqrsTUVwxyz
+# TELEGRAM_CHAT_ID=123456789
+# TELEGRAM_ENABLED=true

CLAUDE.md

@@ -17,6 +17,67 @@ pytest -v --cov=src
 python -m src.main --mode=paper
 ```
 
+## Telegram Notifications (Optional)
+
+Get real-time alerts for trades, circuit breakers, and system events via Telegram.
+
+### Quick Setup
+
+1. **Create bot**: Message [@BotFather](https://t.me/BotFather) on Telegram → `/newbot`
+2. **Get chat ID**: Message [@userinfobot](https://t.me/userinfobot) → `/start`
+3. **Configure**: Add to `.env`:
+   ```bash
+   TELEGRAM_BOT_TOKEN=1234567890:ABCdefGHIjklMNOpqrsTUVwxyz
+   TELEGRAM_CHAT_ID=123456789
+   TELEGRAM_ENABLED=true
+   ```
+4. **Test**: Start bot conversation (`/start`), then run the agent
+
+**Full documentation**: [src/notifications/README.md](src/notifications/README.md)
+
+### What You'll Get
+
+- 🟢 Trade execution alerts (BUY/SELL with confidence)
+- 🚨 Circuit breaker trips (automatic trading halt)
+- ⚠️ Fat-finger rejections (oversized orders blocked)
+- ℹ️ Market open/close notifications
+- 📝 System startup/shutdown status
+
+**Fail-safe**: Notifications never crash the trading system. Missing credentials or API errors are logged but trading continues normally.
+
+## Smart Volatility Scanner (Optional)
+
+Python-first filtering pipeline that reduces Gemini API calls by pre-filtering stocks using technical indicators.
+
+### How It Works
+
+1. **Fetch Rankings** — KIS API volume surge rankings (top 30 stocks)
+2. **Python Filter** — RSI + volume ratio calculations (no AI)
+   - Volume > 200% of previous day
+   - RSI(14) < 30 (oversold) OR RSI(14) > 70 (momentum)
+3. **AI Judgment** — Only qualified candidates (1-3 stocks) sent to Gemini
+
+### Configuration
+
+Add to `.env` (optional, has sensible defaults):
+```bash
+RSI_OVERSOLD_THRESHOLD=30    # 0-50, default 30
+RSI_MOMENTUM_THRESHOLD=70    # 50-100, default 70
+VOL_MULTIPLIER=2.0           # Volume threshold (2.0 = 200%)
+SCANNER_TOP_N=3              # Max candidates per scan
+```
+
+### Benefits
+
+- **Reduces API costs** — Process 1-3 stocks instead of 20-30
+- **Python-based filtering** — Fast technical analysis before AI
+- **Evolution-ready** — Selection context logged for strategy optimization
+- **Fault-tolerant** — Falls back to static watchlist on API failure
+
+### Realtime Mode Only
+
+Smart Scanner runs in `TRADE_MODE=realtime` only. Daily mode uses static watchlists for batch efficiency.
+
 ## Documentation
 
 - **[Workflow Guide](docs/workflow.md)** — Git workflow policy and agent-based development
@@ -25,6 +86,7 @@ python -m src.main --mode=paper
 - **[Context Tree](docs/context-tree.md)** — L1-L7 hierarchical memory system
 - **[Testing](docs/testing.md)** — Test structure, coverage requirements, writing tests
 - **[Agent Policies](docs/agents.md)** — Prime directives, constraints, prohibited actions
+- **[Requirements Log](docs/requirements-log.md)** — User requirements and feedback tracking
 
 ## Core Principles
 
@@ -33,20 +95,31 @@ python -m src.main --mode=paper
 3. **Issue-Driven Development** — All work goes through Gitea issues → feature branches → PRs
 4. **Agent Specialization** — Use dedicated agents for design, coding, testing, docs, review
 
+## Requirements Management
+
+User requirements and feedback are tracked in [docs/requirements-log.md](docs/requirements-log.md):
+
+- New requirements are added chronologically with dates
+- Code changes should reference related requirements
+- Helps maintain project evolution aligned with user needs
+- Preserves context across conversations and development cycles
+
 ## Project Structure
 
 ```
 src/
+├── analysis/        # Technical analysis (RSI, volatility, smart scanner)
 ├── broker/          # KIS API client (domestic + overseas)
 ├── brain/           # Gemini AI decision engine
 ├── core/            # Risk manager (READ-ONLY)
 ├── evolution/       # Self-improvement optimizer
 ├── markets/         # Market schedules and timezone handling
+├── notifications/   # Telegram real-time alerts
 ├── db.py            # SQLite trade logging
 ├── main.py          # Trading loop orchestrator
 └── config.py        # Settings (from .env)
 
-tests/               # 54 tests across 4 files
+tests/               # 343 tests across 14 files
 docs/                # Extended documentation
 ```
 

README.md

@@ -29,6 +29,7 @@ KIS(한국투자증권) API로 매매하고, Google Gemini로 판단하며, 자
 | 브로커 | `src/broker/kis_api.py` | KIS API 비동기 래퍼 (토큰 갱신, 레이트 리미터, 해시키) |
 | 두뇌 | `src/brain/gemini_client.py` | Gemini 프롬프트 구성 및 JSON 응답 파싱 |
 | 방패 | `src/core/risk_manager.py` | 서킷 브레이커 + 팻 핑거 체크 |
+| 알림 | `src/notifications/telegram_client.py` | 텔레그램 실시간 거래 알림 (선택사항) |
 | 진화 | `src/evolution/optimizer.py` | 실패 패턴 분석 → 새 전략 생성 → 테스트 → PR |
 | DB | `src/db.py` | SQLite 거래 로그 기록 |
 
@@ -75,6 +76,34 @@ python -m src.main --mode=paper
 docker compose up -d ouroboros
 ```
 
+## 텔레그램 알림 (선택사항)
+
+거래 실행, 서킷 브레이커 발동, 시스템 상태 등을 텔레그램으로 실시간 알림 받을 수 있습니다.
+
+### 빠른 설정
+
+1. **봇 생성**: 텔레그램에서 [@BotFather](https://t.me/BotFather) 메시지 → `/newbot` 명령
+2. **채팅 ID 확인**: [@userinfobot](https://t.me/userinfobot) 메시지 → `/start` 명령
+3. **환경변수 설정**: `.env` 파일에 추가
+   ```bash
+   TELEGRAM_BOT_TOKEN=1234567890:ABCdefGHIjklMNOpqrsTUVwxyz
+   TELEGRAM_CHAT_ID=123456789
+   TELEGRAM_ENABLED=true
+   ```
+4. **테스트**: 봇과 대화 시작 (`/start` 전송) 후 에이전트 실행
+
+**상세 문서**: [src/notifications/README.md](src/notifications/README.md)
+
+### 알림 종류
+
+- 🟢 거래 체결 알림 (BUY/SELL + 신뢰도)
+- 🚨 서킷 브레이커 발동 (자동 거래 중단)
+- ⚠️ 팻 핑거 차단 (과도한 주문 차단)
+- ℹ️ 장 시작/종료 알림
+- 📝 시스템 시작/종료 상태
+
+**안전장치**: 알림 실패해도 거래는 계속 진행됩니다. 텔레그램 API 오류나 설정 누락이 있어도 거래 시스템은 정상 작동합니다.
+
 ## 테스트
 
 35개 테스트가 TDD 방식으로 구현 전에 먼저 작성되었습니다.
@@ -104,15 +133,16 @@ The-Ouroboros/
 │   ├── agents.md                # AI 에이전트 페르소나 정의
 │   └── skills.md                # 사용 가능한 도구 목록
 ├── src/
-│   ├── config.py                # Pydantic 설정
-│   ├── logging_config.py        # JSON 구조화 로깅
-│   ├── db.py                    # SQLite 거래 기록
-│   ├── main.py                  # 비동기 거래 루프
-│   ├── broker/kis_api.py        # KIS API 클라이언트
-│   ├── brain/gemini_client.py   # Gemini 의사결정 엔진
-│   ├── core/risk_manager.py     # 리스크 관리
-│   ├── evolution/optimizer.py   # 전략 진화 엔진
-│   └── strategies/base.py       # 전략 베이스 클래스
+│   ├── config.py                      # Pydantic 설정
+│   ├── logging_config.py              # JSON 구조화 로깅
+│   ├── db.py                          # SQLite 거래 기록
+│   ├── main.py                        # 비동기 거래 루프
+│   ├── broker/kis_api.py              # KIS API 클라이언트
+│   ├── brain/gemini_client.py         # Gemini 의사결정 엔진
+│   ├── core/risk_manager.py           # 리스크 관리
+│   ├── notifications/telegram_client.py # 텔레그램 알림
+│   ├── evolution/optimizer.py         # 전략 진화 엔진
+│   └── strategies/base.py             # 전략 베이스 클래스
 ├── tests/                       # TDD 테스트 스위트
 ├── Dockerfile                   # 멀티스테이지 빌드
 ├── docker-compose.yml           # 서비스 오케스트레이션

docs/agent-constraints.md

@@ -0,0 +1,45 @@
+# Agent Constraints
+
+This document records **persistent behavioral constraints** for agents working on this repository.
+It is distinct from `docs/requirements-log.md`, which records **project/product requirements**.
+
+## Scope
+
+- Applies to all AI agents and automation that modify this repo.
+- Supplements (does not replace) `docs/agents.md` and `docs/workflow.md`.
+
+## Persistent Rules
+
+1. **Workflow enforcement**
+   - Follow `docs/workflow.md` for all changes.
+   - Create a Gitea issue before any code or documentation change.
+   - Work on a feature branch `feature/issue-{N}-{short-description}` and open a PR.
+   - Never commit directly to `main`.
+
+2. **Document-first routing**
+   - When performing work, consult relevant `docs/` files *before* making changes.
+   - Route decisions to the documented policy whenever applicable.
+   - If guidance conflicts, prefer the stricter/safety-first rule and note it in the PR.
+
+3. **Docs with code**
+   - Any code change must be accompanied by relevant documentation updates.
+   - If no doc update is needed, state the reason explicitly in the PR.
+
+4. **Session-persistent user constraints**
+   - If the user requests that a behavior should persist across sessions, record it here
+     (or in a dedicated policy doc) and reference it when working.
+   - Keep entries short and concrete, with dates.
+
+## Change Control
+
+- Changes to this file follow the same workflow as code changes.
+- Keep the history chronological and minimize rewording of existing entries.
+
+## History
+
+### 2026-02-08
+
+- Always enforce Gitea workflow: issue -> feature branch -> PR before changes.
+- When work requires guidance, consult the relevant `docs/` policies first.
+- Any code change must be accompanied by relevant documentation updates.
+- Persist user constraints across sessions by recording them in this document.

docs/architecture.md

@@ -2,7 +2,42 @@
 
 ## Overview
 
-Self-evolving AI trading agent for global stock markets via KIS (Korea Investment & Securities) API. The main loop in `src/main.py` orchestrates four components in a 60-second cycle per stock across multiple markets.
+Self-evolving AI trading agent for global stock markets via KIS (Korea Investment & Securities) API. The main loop in `src/main.py` orchestrates four components across multiple markets with two trading modes: daily (batch API calls) or realtime (per-stock decisions).
+
+## Trading Modes
+
+The system supports two trading frequency modes controlled by the `TRADE_MODE` environment variable:
+
+### Daily Mode (default)
+
+Optimized for Gemini Free tier API limits (20 calls/day):
+
+- **Batch decisions**: 1 API call per market per session
+- **Fixed schedule**: 4 sessions per day at 6-hour intervals (configurable)
+- **API efficiency**: Processes all stocks in a market simultaneously
+- **Use case**: Free tier users, cost-conscious deployments
+- **Configuration**:
+  ```bash
+  TRADE_MODE=daily
+  DAILY_SESSIONS=4        # Sessions per day (1-10)
+  SESSION_INTERVAL_HOURS=6  # Hours between sessions (1-24)
+  ```
+
+**Example**: With 2 markets (US, KR) and 4 sessions/day = 8 API calls/day (within 20 call limit)
+
+### Realtime Mode
+
+High-frequency trading with individual stock analysis:
+
+- **Per-stock decisions**: 1 API call per stock per cycle
+- **60-second interval**: Continuous monitoring
+- **Use case**: Production deployments with Gemini paid tier
+- **Configuration**:
+  ```bash
+  TRADE_MODE=realtime
+  ```
+
+**Note**: Realtime mode requires Gemini API subscription due to high call volume.
 
 ## Core Components
 
@@ -29,7 +64,39 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
 - `get_open_markets()` returns currently active markets
 - `get_next_market_open()` finds next market to open and when
 
-### 2. Brain (`src/brain/gemini_client.py`)
+**New API Methods** (added in v0.9.0):
+- `fetch_market_rankings()` — Fetch volume surge rankings from KIS API
+- `get_daily_prices()` — Fetch OHLCV history for technical analysis
+
+### 2. Analysis (`src/analysis/`)
+
+**VolatilityAnalyzer** (`volatility.py`) — Technical indicator calculations
+
+- ATR (Average True Range) for volatility measurement
+- RSI (Relative Strength Index) using Wilder's smoothing method
+- Price change percentages across multiple timeframes
+- Volume surge ratios and price-volume divergence
+- Momentum scoring (0-100 scale)
+- Breakout/breakdown pattern detection
+
+**SmartVolatilityScanner** (`smart_scanner.py`) — Python-first filtering pipeline
+
+- **Step 1**: Fetch volume rankings from KIS API (top 30 stocks)
+- **Step 2**: Calculate RSI and volume ratio for each stock
+- **Step 3**: Apply filters:
+  - Volume ratio >= `VOL_MULTIPLIER` (default 2.0x previous day)
+  - RSI < `RSI_OVERSOLD_THRESHOLD` (30) OR RSI > `RSI_MOMENTUM_THRESHOLD` (70)
+- **Step 4**: Score candidates by RSI extremity (60%) + volume surge (40%)
+- **Step 5**: Return top N candidates (default 3) for AI analysis
+- **Fallback**: Uses static watchlist if ranking API unavailable
+- **Realtime mode only**: Daily mode uses batch processing for API efficiency
+
+**Benefits:**
+- Reduces Gemini API calls from 20-30 stocks to 1-3 qualified candidates
+- Fast Python-based filtering before expensive AI judgment
+- Logs selection context (RSI, volume_ratio, signal, score) for Evolution system
+
+### 3. Brain (`src/brain/gemini_client.py`)
 
 **GeminiClient** — AI decision engine powered by Google Gemini
 
@@ -39,7 +106,7 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
 - Falls back to safe HOLD on any parse/API error
 - Handles markdown-wrapped JSON, malformed responses, invalid actions
 
-### 3. Risk Manager (`src/core/risk_manager.py`)
+### 4. Risk Manager (`src/core/risk_manager.py`)
 
 **RiskManager** — Safety circuit breaker and order validation
 
@@ -51,7 +118,26 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
 - **Fat-Finger Protection**: Rejects orders exceeding 30% of available cash
   - Must always be enforced, cannot be disabled
 
-### 4. Evolution (`src/evolution/optimizer.py`)
+### 5. Notifications (`src/notifications/telegram_client.py`)
+
+**TelegramClient** — Real-time event notifications via Telegram Bot API
+
+- Sends alerts for trades, circuit breakers, fat-finger rejections, system events
+- Non-blocking: failures are logged but never crash trading
+- Rate-limited: 1 message/second default to respect Telegram API limits
+- Auto-disabled when credentials missing
+- Gracefully handles API errors, network timeouts, invalid tokens
+
+**Notification Types:**
+- Trade execution (BUY/SELL with confidence)
+- Circuit breaker trips (critical alert)
+- Fat-finger protection triggers (order rejection)
+- Market open/close events
+- System startup/shutdown status
+
+**Setup:** See [src/notifications/README.md](../src/notifications/README.md) for bot creation and configuration.
+
+### 6. Evolution (`src/evolution/optimizer.py`)
 
 **StrategyOptimizer** — Self-improvement loop
 
@@ -63,9 +149,11 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
 
 ## Data Flow
 
+### Realtime Mode (with Smart Scanner)
+
 ```
 ┌─────────────────────────────────────────────────────────────┐
-│ Main Loop (60s cycle per stock, per market)                │
+│ Main Loop (60s cycle per market)                           │
 └─────────────────────────────────────────────────────────────┘
                            │
                            ▼
@@ -78,6 +166,21 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
                            │
                            ▼
         ┌──────────────────────────────────┐
+        │ Smart Scanner (Python-first)      │
+        │ - Fetch volume rankings (KIS)    │
+        │ - Get 20d price history per stock│
+        │ - Calculate RSI(14) + vol ratio  │
+        │ - Filter: vol>2x AND RSI extreme │
+        │ - Return top 3 qualified stocks  │
+        └──────────────────┬────────────────┘
+                           │
+                           ▼
+        ┌──────────────────────────────────┐
+        │ For Each Qualified Candidate      │
+        └──────────────────┬────────────────┘
+                           │
+                           ▼
+        ┌──────────────────────────────────┐
         │ Broker: Fetch Market Data        │
         │ - Domestic: orderbook + balance  │
         │ - Overseas: price + balance      │
@@ -91,7 +194,7 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
                            │
                            ▼
         ┌──────────────────────────────────┐
-        │ Brain: Get Decision               │
+        │ Brain: Get Decision (AI)          │
         │ - Build prompt with market data   │
         │ - Call Gemini API                 │
         │ - Parse JSON response             │
@@ -115,10 +218,21 @@ Self-evolving AI trading agent for global stock markets via KIS (Korea Investmen
                            │
                            ▼
         ┌──────────────────────────────────┐
+        │ Notifications: Send Alert         │
+        │ - Trade execution notification    │
+        │ - Non-blocking (errors logged)    │
+        │ - Rate-limited to 1/sec           │
+        └──────────────────┬────────────────┘
+                           │
+                           ▼
+        ┌──────────────────────────────────┐
         │ Database: Log Trade               │
         │ - SQLite (data/trades.db)         │
         │ - Track: action, confidence,      │
         │   rationale, market, exchange     │
+        │ - NEW: selection_context (JSON)   │
+        │   - RSI, volume_ratio, signal     │
+        │   - For Evolution optimization    │
         └───────────────────────────────────┘
 ```
 
@@ -138,11 +252,24 @@ CREATE TABLE trades (
     price REAL,
     pnl REAL DEFAULT 0.0,
     market TEXT DEFAULT 'KR',       -- KR | US_NASDAQ | JP | etc.
-    exchange_code TEXT DEFAULT 'KRX' -- KRX | NASD | NYSE | etc.
+    exchange_code TEXT DEFAULT 'KRX', -- KRX | NASD | NYSE | etc.
+    selection_context TEXT          -- JSON: {rsi, volume_ratio, signal, score}
 );
 ```
 
-Auto-migration: Adds `market` and `exchange_code` columns if missing for backward compatibility.
+**Selection Context** (new in v0.9.0): Stores scanner selection criteria as JSON:
+```json
+{
+  "rsi": 28.5,
+  "volume_ratio": 2.7,
+  "signal": "oversold",
+  "score": 85.2
+}
+```
+
+Enables Evolution system to analyze correlation between selection criteria and trade outcomes.
+
+Auto-migration: Adds `market`, `exchange_code`, and `selection_context` columns if missing for backward compatibility.
 
 ## Configuration
 
@@ -164,6 +291,22 @@ CONFIDENCE_THRESHOLD=80
 MAX_LOSS_PCT=3.0
 MAX_ORDER_PCT=30.0
 ENABLED_MARKETS=KR,US_NASDAQ  # Comma-separated market codes
+
+# Trading Mode (API efficiency)
+TRADE_MODE=daily              # daily | realtime
+DAILY_SESSIONS=4              # Sessions per day (daily mode only)
+SESSION_INTERVAL_HOURS=6      # Hours between sessions (daily mode only)
+
+# Telegram Notifications (optional)
+TELEGRAM_BOT_TOKEN=1234567890:ABCdefGHIjklMNOpqrsTUVwxyz
+TELEGRAM_CHAT_ID=123456789
+TELEGRAM_ENABLED=true
+
+# Smart Scanner (optional, realtime mode only)
+RSI_OVERSOLD_THRESHOLD=30    # 0-50, oversold threshold
+RSI_MOMENTUM_THRESHOLD=70    # 50-100, momentum threshold
+VOL_MULTIPLIER=2.0           # Minimum volume ratio (2.0 = 200%)
+SCANNER_TOP_N=3              # Max qualified candidates per scan
 ```
 
 Tests use in-memory SQLite (`DB_PATH=":memory:"`) and dummy credentials via `tests/conftest.py`.
@@ -189,3 +332,12 @@ Tests use in-memory SQLite (`DB_PATH=":memory:"`) and dummy credentials via `tes
 - Wait until next market opens
 - Use `get_next_market_open()` to calculate wait time
 - Sleep until market open time
+
+### Telegram API Errors
+- Log warning but continue trading
+- Missing credentials → auto-disable notifications
+- Network timeout → skip notification, no retry
+- Invalid token → log error, trading unaffected
+- Rate limit exceeded → queued via rate limiter
+
+**Guarantee**: Notification failures never interrupt trading operations.

docs/disaster_recovery.md

@@ -0,0 +1,348 @@
+# Disaster Recovery Guide
+
+Complete guide for backing up and restoring The Ouroboros trading system.
+
+## Table of Contents
+
+- [Backup Strategy](#backup-strategy)
+- [Creating Backups](#creating-backups)
+- [Restoring from Backup](#restoring-from-backup)
+- [Health Monitoring](#health-monitoring)
+- [Export Formats](#export-formats)
+- [RTO/RPO](#rtorpo)
+- [Testing Recovery](#testing-recovery)
+
+## Backup Strategy
+
+The system implements a 3-tier backup retention policy:
+
+| Policy | Frequency | Retention | Purpose |
+|--------|-----------|-----------|---------|
+| **Daily** | Every day | 30 days | Quick recovery from recent issues |
+| **Weekly** | Sunday | 1 year | Medium-term historical analysis |
+| **Monthly** | 1st of month | Forever | Long-term archival |
+
+### Storage Structure
+
+```
+data/backups/
+├── daily/          # Last 30 days
+├── weekly/         # Last 52 weeks
+└── monthly/        # Forever (cold storage)
+```
+
+## Creating Backups
+
+### Automated Backups (Recommended)
+
+Set up a cron job to run daily:
+
+```bash
+# Edit crontab
+crontab -e
+
+# Run backup at 2 AM every day
+0 2 * * * cd /path/to/The-Ouroboros && ./scripts/backup.sh >> logs/backup.log 2>&1
+```
+
+### Manual Backups
+
+```bash
+# Run backup script
+./scripts/backup.sh
+
+# Or use Python directly
+python3 -c "
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler, BackupPolicy
+
+scheduler = BackupScheduler('data/trade_logs.db', Path('data/backups'))
+metadata = scheduler.create_backup(BackupPolicy.DAILY, verify=True)
+print(f'Backup created: {metadata.file_path}')
+"
+```
+
+### Export to Other Formats
+
+```bash
+python3 -c "
+from pathlib import Path
+from src.backup.exporter import BackupExporter, ExportFormat
+
+exporter = BackupExporter('data/trade_logs.db')
+results = exporter.export_all(
+    Path('exports'),
+    formats=[ExportFormat.JSON, ExportFormat.CSV],
+    compress=True
+)
+"
+```
+
+## Restoring from Backup
+
+### Interactive Restoration
+
+```bash
+./scripts/restore.sh
+```
+
+The script will:
+1. List available backups
+2. Ask you to select one
+3. Create a safety backup of current database
+4. Restore the selected backup
+5. Verify database integrity
+
+### Manual Restoration
+
+```python
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler
+
+scheduler = BackupScheduler('data/trade_logs.db', Path('data/backups'))
+
+# List backups
+backups = scheduler.list_backups()
+for backup in backups:
+    print(f"{backup.timestamp}: {backup.file_path}")
+
+# Restore specific backup
+scheduler.restore_backup(backups[0], verify=True)
+```
+
+## Health Monitoring
+
+### Check System Health
+
+```python
+from pathlib import Path
+from src.backup.health_monitor import HealthMonitor
+
+monitor = HealthMonitor('data/trade_logs.db', Path('data/backups'))
+
+# Run all checks
+report = monitor.get_health_report()
+print(f"Overall status: {report['overall_status']}")
+
+# Individual checks
+checks = monitor.run_all_checks()
+for name, result in checks.items():
+    print(f"{name}: {result.status.value} - {result.message}")
+```
+
+### Health Checks
+
+The system monitors:
+
+- **Database Health**: Accessibility, integrity, size
+- **Disk Space**: Available storage (alerts if < 10 GB)
+- **Backup Recency**: Ensures backups are < 25 hours old
+
+### Health Status Levels
+
+- **HEALTHY**: All systems operational
+- **DEGRADED**: Warning condition (e.g., low disk space)
+- **UNHEALTHY**: Critical issue (e.g., database corrupted, no backups)
+
+## Export Formats
+
+### JSON (Human-Readable)
+
+```json
+{
+  "export_timestamp": "2024-01-15T10:30:00Z",
+  "record_count": 150,
+  "trades": [
+    {
+      "timestamp": "2024-01-15T09:00:00Z",
+      "stock_code": "005930",
+      "action": "BUY",
+      "quantity": 10,
+      "price": 70000.0,
+      "confidence": 85,
+      "rationale": "Strong momentum",
+      "pnl": 0.0
+    }
+  ]
+}
+```
+
+### CSV (Analysis Tools)
+
+Compatible with Excel, pandas, R:
+
+```csv
+timestamp,stock_code,action,quantity,price,confidence,rationale,pnl
+2024-01-15T09:00:00Z,005930,BUY,10,70000.0,85,Strong momentum,0.0
+```
+
+### Parquet (Big Data)
+
+Columnar format for Spark, DuckDB:
+
+```python
+import pandas as pd
+df = pd.read_parquet('exports/trades_20240115.parquet')
+```
+
+## RTO/RPO
+
+### Recovery Time Objective (RTO)
+
+**Target: < 5 minutes**
+
+Time to restore trading operations:
+1. Identify backup to restore (1 min)
+2. Run restore script (2 min)
+3. Verify database integrity (1 min)
+4. Restart trading system (1 min)
+
+### Recovery Point Objective (RPO)
+
+**Target: < 24 hours**
+
+Maximum acceptable data loss:
+- Daily backups ensure ≤ 24-hour data loss
+- For critical periods, run backups more frequently
+
+## Testing Recovery
+
+### Quarterly Recovery Test
+
+Perform full disaster recovery test every quarter:
+
+1. **Create test backup**
+   ```bash
+   ./scripts/backup.sh
+   ```
+
+2. **Simulate disaster** (use test database)
+   ```bash
+   cp data/trade_logs.db data/trade_logs_test.db
+   rm data/trade_logs_test.db  # Simulate data loss
+   ```
+
+3. **Restore from backup**
+   ```bash
+   DB_PATH=data/trade_logs_test.db ./scripts/restore.sh
+   ```
+
+4. **Verify data integrity**
+   ```python
+   import sqlite3
+   conn = sqlite3.connect('data/trade_logs_test.db')
+   cursor = conn.execute('SELECT COUNT(*) FROM trades')
+   print(f"Restored {cursor.fetchone()[0]} trades")
+   ```
+
+5. **Document results** in `logs/recovery_test_YYYYMMDD.md`
+
+### Backup Verification
+
+Always verify backups after creation:
+
+```python
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler
+
+scheduler = BackupScheduler('data/trade_logs.db', Path('data/backups'))
+
+# Create and verify
+metadata = scheduler.create_backup(BackupPolicy.DAILY, verify=True)
+print(f"Checksum: {metadata.checksum}")  # Should not be None
+```
+
+## Emergency Procedures
+
+### Database Corrupted
+
+1. Stop trading system immediately
+2. Check most recent backup age: `ls -lht data/backups/daily/`
+3. Restore: `./scripts/restore.sh`
+4. Verify: Run health check
+5. Resume trading
+
+### Disk Full
+
+1. Check disk space: `df -h`
+2. Clean old backups: Run cleanup manually
+   ```python
+   from pathlib import Path
+   from src.backup.scheduler import BackupScheduler
+   scheduler = BackupScheduler('data/trade_logs.db', Path('data/backups'))
+   scheduler.cleanup_old_backups()
+   ```
+3. Consider archiving old monthly backups to external storage
+4. Increase disk space if needed
+
+### Lost All Backups
+
+If local backups are lost:
+1. Check if exports exist in `exports/` directory
+2. Reconstruct database from CSV/JSON exports
+3. If no exports: Check broker API for trade history
+4. Manual reconstruction as last resort
+
+## Best Practices
+
+1. **Test Restores Regularly**: Don't wait for disaster
+2. **Monitor Disk Space**: Set up alerts at 80% usage
+3. **Keep Multiple Generations**: Never delete all backups at once
+4. **Verify Checksums**: Always verify backup integrity
+5. **Document Changes**: Update this guide when backup strategy changes
+6. **Off-Site Storage**: Consider external backup for monthly archives
+
+## Troubleshooting
+
+### Backup Script Fails
+
+```bash
+# Check database file permissions
+ls -l data/trade_logs.db
+
+# Check disk space
+df -h data/
+
+# Run backup manually with debug
+python3 -c "
+import logging
+logging.basicConfig(level=logging.DEBUG)
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler, BackupPolicy
+scheduler = BackupScheduler('data/trade_logs.db', Path('data/backups'))
+scheduler.create_backup(BackupPolicy.DAILY, verify=True)
+"
+```
+
+### Restore Fails Verification
+
+```bash
+# Check backup file integrity
+python3 -c "
+import sqlite3
+conn = sqlite3.connect('data/backups/daily/trade_logs_daily_20240115.db')
+cursor = conn.execute('PRAGMA integrity_check')
+print(cursor.fetchone()[0])
+"
+```
+
+### Health Check Fails
+
+```python
+from pathlib import Path
+from src.backup.health_monitor import HealthMonitor
+
+monitor = HealthMonitor('data/trade_logs.db', Path('data/backups'))
+
+# Check each component individually
+print("Database:", monitor.check_database_health())
+print("Disk Space:", monitor.check_disk_space())
+print("Backup Recency:", monitor.check_backup_recency())
+```
+
+## Contact
+
+For backup/recovery issues:
+- Check logs: `logs/backup.log`
+- Review health status: Run health monitor
+- Raise issue on GitHub if automated recovery fails

docs/requirements-log.md

@@ -0,0 +1,66 @@
+# Requirements Log
+
+프로젝트 진화를 위한 사용자 요구사항 기록.
+
+이 문서는 시간순으로 사용자와의 대화에서 나온 요구사항과 피드백을 기록합니다.
+새로운 요구사항이 있으면 날짜와 함께 추가하세요.
+
+---
+
+## 2026-02-05
+
+### API 효율화
+- Gemini API는 귀중한 자원. 종목별 개별 호출 대신 배치 호출 필요
+- Free tier 한도(20 calls/day) 고려하여 일일 몇 차례 거래 모드로 전환
+- 배치 API 호출로 여러 종목을 한 번에 분석
+
+### 거래 모드
+- **Daily Mode**: 하루 4회 거래 세션 (6시간 간격) - Free tier 호환
+- **Realtime Mode**: 60초 간격 실시간 거래 - 유료 구독 필요
+- `TRADE_MODE` 환경변수로 모드 선택
+
+### 진화 시스템
+- 사용자 대화 내용을 문서로 기록하여 향후에도 의도 반영
+- 프롬프트 품질 검증은 별도 이슈로 다룰 예정
+
+### 문서화
+- 시스템 구조, 기능별 설명 등 코드 문서화 항상 신경쓸 것
+- 새로운 기능 추가 시 관련 문서 업데이트 필수
+
+---
+
+## 2026-02-06
+
+### Smart Volatility Scanner (Python-First, AI-Last 파이프라인)
+
+**배경:**
+- 정적 종목 리스트를 순회하는 방식은 비효율적
+- KIS API 거래량 순위를 통해 시장 주도주를 자동 탐지해야 함
+- Gemini API 호출 전에 Python 기반 기술적 분석으로 필터링 필요
+
+**요구사항:**
+1. KIS API 거래량 순위 API 통합 (`fetch_market_rankings`)
+2. 일별 가격 히스토리 API 추가 (`get_daily_prices`)
+3. RSI(14) 계산 기능 구현 (Wilder's smoothing method)
+4. 필터 조건:
+   - 거래량 > 전일 대비 200% (VOL_MULTIPLIER)
+   - RSI < 30 (과매도) OR RSI > 70 (모멘텀)
+5. 상위 1-3개 적격 종목만 Gemini에 전달
+6. 종목 선정 배경(RSI, volume_ratio, signal, score) 데이터베이스 기록
+
+**구현 결과:**
+- `src/analysis/smart_scanner.py`: SmartVolatilityScanner 클래스
+- `src/analysis/volatility.py`: calculate_rsi() 메서드 추가
+- `src/broker/kis_api.py`: 2개 신규 API 메서드
+- `src/db.py`: selection_context 컬럼 추가
+- 설정 가능한 임계값: RSI_OVERSOLD_THRESHOLD, RSI_MOMENTUM_THRESHOLD, VOL_MULTIPLIER, SCANNER_TOP_N
+
+**효과:**
+- Gemini API 호출 20-30개 → 1-3개로 감소
+- Python 기반 빠른 필터링 → 비용 절감
+- 선정 기준 추적 → Evolution 시스템 최적화 가능
+- API 장애 시 정적 watchlist로 자동 전환
+
+**참고:** Realtime 모드 전용. Daily 모드는 배치 효율성을 위해 정적 watchlist 사용.
+
+**이슈/PR:** #76, #77

docs/workflow.md

@@ -6,6 +6,7 @@
 
 1. **Create Gitea Issue First** — All features, bug fixes, and policy changes require a Gitea issue before any code is written
 2. **Create Feature Branch** — Branch from `main` using format `feature/issue-{N}-{short-description}`
+   - After creating the branch, run `git pull origin main` and rebase to ensure the branch is up to date
 3. **Implement Changes** — Write code, tests, and documentation on the feature branch
 4. **Create Pull Request** — Submit PR to `main` branch referencing the issue number
 5. **Review & Merge** — After approval, merge via PR (squash or merge commit)

scripts/backup.sh

@@ -0,0 +1,96 @@
+#!/usr/bin/env bash
+# Automated backup script for The Ouroboros trading system
+# Runs daily/weekly/monthly backups
+
+set -euo pipefail
+
+# Configuration
+DB_PATH="${DB_PATH:-data/trade_logs.db}"
+BACKUP_DIR="${BACKUP_DIR:-data/backups}"
+PYTHON="${PYTHON:-python3}"
+
+# Colors for output
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+RED='\033[0;31m'
+NC='\033[0m' # No Color
+
+log_info() {
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+log_warn() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+log_error() {
+    echo -e "${RED}[ERROR]${NC} $1"
+}
+
+# Check if database exists
+if [ ! -f "$DB_PATH" ]; then
+    log_error "Database not found: $DB_PATH"
+    exit 1
+fi
+
+# Create backup directory
+mkdir -p "$BACKUP_DIR"
+
+log_info "Starting backup process..."
+log_info "Database: $DB_PATH"
+log_info "Backup directory: $BACKUP_DIR"
+
+# Determine backup policy based on day of week and month
+DAY_OF_WEEK=$(date +%u)  # 1-7 (Monday-Sunday)
+DAY_OF_MONTH=$(date +%d)
+
+if [ "$DAY_OF_MONTH" == "01" ]; then
+    POLICY="monthly"
+    log_info "Running MONTHLY backup (first day of month)"
+elif [ "$DAY_OF_WEEK" == "7" ]; then
+    POLICY="weekly"
+    log_info "Running WEEKLY backup (Sunday)"
+else
+    POLICY="daily"
+    log_info "Running DAILY backup"
+fi
+
+# Run Python backup script
+$PYTHON -c "
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler, BackupPolicy
+from src.backup.health_monitor import HealthMonitor
+
+# Create scheduler
+scheduler = BackupScheduler(
+    db_path='$DB_PATH',
+    backup_dir=Path('$BACKUP_DIR')
+)
+
+# Create backup
+policy = BackupPolicy.$POLICY.upper()
+metadata = scheduler.create_backup(policy, verify=True)
+print(f'Backup created: {metadata.file_path}')
+print(f'Size: {metadata.size_bytes / 1024 / 1024:.2f} MB')
+print(f'Checksum: {metadata.checksum}')
+
+# Cleanup old backups
+removed = scheduler.cleanup_old_backups()
+total_removed = sum(removed.values())
+if total_removed > 0:
+    print(f'Removed {total_removed} old backup(s)')
+
+# Health check
+monitor = HealthMonitor('$DB_PATH', Path('$BACKUP_DIR'))
+status = monitor.get_overall_status()
+print(f'System health: {status.value}')
+"
+
+if [ $? -eq 0 ]; then
+    log_info "Backup completed successfully"
+else
+    log_error "Backup failed"
+    exit 1
+fi
+
+log_info "Backup process finished"

scripts/restore.sh

@@ -0,0 +1,111 @@
+#!/usr/bin/env bash
+# Restore script for The Ouroboros trading system
+# Restores database from a backup file
+
+set -euo pipefail
+
+# Configuration
+DB_PATH="${DB_PATH:-data/trade_logs.db}"
+BACKUP_DIR="${BACKUP_DIR:-data/backups}"
+PYTHON="${PYTHON:-python3}"
+
+# Colors for output
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+RED='\033[0;31m'
+NC='\033[0m' # No Color
+
+log_info() {
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+log_warn() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+log_error() {
+    echo -e "${RED}[ERROR]${NC} $1"
+}
+
+# Check if backup directory exists
+if [ ! -d "$BACKUP_DIR" ]; then
+    log_error "Backup directory not found: $BACKUP_DIR"
+    exit 1
+fi
+
+log_info "Available backups:"
+log_info "=================="
+
+# List available backups
+$PYTHON -c "
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler
+
+scheduler = BackupScheduler(
+    db_path='$DB_PATH',
+    backup_dir=Path('$BACKUP_DIR')
+)
+
+backups = scheduler.list_backups()
+
+if not backups:
+    print('No backups found.')
+    exit(1)
+
+for i, backup in enumerate(backups, 1):
+    size_mb = backup.size_bytes / 1024 / 1024
+    print(f'{i}. [{backup.policy.value.upper()}] {backup.file_path.name}')
+    print(f'   Date: {backup.timestamp.strftime(\"%Y-%m-%d %H:%M:%S UTC\")}')
+    print(f'   Size: {size_mb:.2f} MB')
+    print()
+"
+
+# Ask user to select backup
+echo ""
+read -p "Enter backup number to restore (or 'q' to quit): " BACKUP_NUM
+
+if [ "$BACKUP_NUM" == "q" ]; then
+    log_info "Restore cancelled"
+    exit 0
+fi
+
+# Confirm restoration
+log_warn "WARNING: This will replace the current database!"
+log_warn "Current database will be backed up to: ${DB_PATH}.before_restore"
+read -p "Are you sure you want to continue? (yes/no): " CONFIRM
+
+if [ "$CONFIRM" != "yes" ]; then
+    log_info "Restore cancelled"
+    exit 0
+fi
+
+# Perform restoration
+$PYTHON -c "
+from pathlib import Path
+from src.backup.scheduler import BackupScheduler
+
+scheduler = BackupScheduler(
+    db_path='$DB_PATH',
+    backup_dir=Path('$BACKUP_DIR')
+)
+
+backups = scheduler.list_backups()
+backup_index = int('$BACKUP_NUM') - 1
+
+if backup_index < 0 or backup_index >= len(backups):
+    print('Invalid backup number')
+    exit(1)
+
+selected = backups[backup_index]
+print(f'Restoring: {selected.file_path.name}')
+
+scheduler.restore_backup(selected, verify=True)
+print('Restore completed successfully')
+"
+
+if [ $? -eq 0 ]; then
+    log_info "Database restored successfully"
+else
+    log_error "Restore failed"
+    exit 1
+fi

src/analysis/__init__.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 from src.analysis.scanner import MarketScanner
+from src.analysis.smart_scanner import ScanCandidate, SmartVolatilityScanner
 from src.analysis.volatility import VolatilityAnalyzer
 
-__all__ = ["VolatilityAnalyzer", "MarketScanner"]
+__all__ = ["VolatilityAnalyzer", "MarketScanner", "SmartVolatilityScanner", "ScanCandidate"]

src/analysis/scanner.py

@@ -42,6 +42,7 @@ class MarketScanner:
         volatility_analyzer: VolatilityAnalyzer,
         context_store: ContextStore,
         top_n: int = 5,
+        max_concurrent_scans: int = 1,
     ) -> None:
         """Initialize the market scanner.
 
@@ -51,12 +52,14 @@ class MarketScanner:
             volatility_analyzer: Volatility analyzer instance
             context_store: Context store for L7 real-time data
             top_n: Number of top movers to return per market (default 5)
+            max_concurrent_scans: Max concurrent stock scans (default 1, fully serialized)
         """
         self.broker = broker
         self.overseas_broker = overseas_broker
         self.analyzer = volatility_analyzer
         self.context_store = context_store
         self.top_n = top_n
+        self._scan_semaphore = asyncio.Semaphore(max_concurrent_scans)
 
     async def scan_stock(
         self,
@@ -83,8 +86,8 @@ class MarketScanner:
                 # Convert to orderbook-like structure
                 orderbook = {
                     "output1": {
-                        "stck_prpr": price_data.get("output", {}).get("last", "0"),
-                        "acml_vol": price_data.get("output", {}).get("tvol", "0"),
+                        "stck_prpr": price_data.get("output", {}).get("last", "0") or "0",
+                        "acml_vol": price_data.get("output", {}).get("tvol", "0") or "0",
                     }
                 }
 
@@ -139,8 +142,12 @@ class MarketScanner:
 
         logger.info("Scanning %s market (%d stocks)", market.name, len(stock_codes))
 
-        # Scan all stocks concurrently (with rate limiting handled by broker)
-        tasks = [self.scan_stock(code, market) for code in stock_codes]
+        # Scan stocks with bounded concurrency to prevent API rate limit burst
+        async def _bounded_scan(code: str) -> VolatilityMetrics | None:
+            async with self._scan_semaphore:
+                return await self.scan_stock(code, market)
+
+        tasks = [_bounded_scan(code) for code in stock_codes]
         results = await asyncio.gather(*tasks)
 
         # Filter out failures and sort by momentum score

src/analysis/smart_scanner.py

@@ -0,0 +1,192 @@
+"""Smart Volatility Scanner with RSI and volume filters.
+
+Fetches market rankings from KIS API and applies technical filters
+to identify high-probability trading candidates.
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from typing import Any
+
+from src.analysis.volatility import VolatilityAnalyzer
+from src.broker.kis_api import KISBroker
+from src.config import Settings
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class ScanCandidate:
+    """A qualified candidate from the smart scanner."""
+
+    stock_code: str
+    name: str
+    price: float
+    volume: float
+    volume_ratio: float  # Current volume / previous day volume
+    rsi: float
+    signal: str  # "oversold" or "momentum"
+    score: float  # Composite score for ranking
+
+
+class SmartVolatilityScanner:
+    """Scans market rankings and applies RSI/volume filters.
+
+    Flow:
+    1. Fetch volume rankings from KIS API
+    2. For each ranked stock, fetch daily prices
+    3. Calculate RSI and volume ratio
+    4. Apply filters: volume > VOL_MULTIPLIER AND (RSI < 30 OR RSI > 70)
+    5. Return top N qualified candidates
+    """
+
+    def __init__(
+        self,
+        broker: KISBroker,
+        volatility_analyzer: VolatilityAnalyzer,
+        settings: Settings,
+    ) -> None:
+        """Initialize the smart scanner.
+
+        Args:
+            broker: KIS broker for API calls
+            volatility_analyzer: Analyzer for RSI calculation
+            settings: Application settings
+        """
+        self.broker = broker
+        self.analyzer = volatility_analyzer
+        self.settings = settings
+
+        # Extract scanner settings
+        self.rsi_oversold = settings.RSI_OVERSOLD_THRESHOLD
+        self.rsi_momentum = settings.RSI_MOMENTUM_THRESHOLD
+        self.vol_multiplier = settings.VOL_MULTIPLIER
+        self.top_n = settings.SCANNER_TOP_N
+
+    async def scan(
+        self,
+        fallback_stocks: list[str] | None = None,
+    ) -> list[ScanCandidate]:
+        """Execute smart scan and return qualified candidates.
+
+        Args:
+            fallback_stocks: Stock codes to use if ranking API fails
+
+        Returns:
+            List of ScanCandidate, sorted by score, up to top_n items
+        """
+        # Step 1: Fetch rankings
+        try:
+            rankings = await self.broker.fetch_market_rankings(
+                ranking_type="volume",
+                limit=30,  # Fetch more than needed for filtering
+            )
+            logger.info("Fetched %d stocks from volume rankings", len(rankings))
+        except ConnectionError as exc:
+            logger.warning("Ranking API failed, using fallback: %s", exc)
+            if fallback_stocks:
+                # Create minimal ranking data for fallback
+                rankings = [
+                    {
+                        "stock_code": code,
+                        "name": code,
+                        "price": 0,
+                        "volume": 0,
+                        "change_rate": 0,
+                        "volume_increase_rate": 0,
+                    }
+                    for code in fallback_stocks
+                ]
+            else:
+                return []
+
+        # Step 2: Analyze each stock
+        candidates: list[ScanCandidate] = []
+
+        for stock in rankings:
+            stock_code = stock["stock_code"]
+            if not stock_code:
+                continue
+
+            try:
+                # Fetch daily prices for RSI calculation
+                daily_prices = await self.broker.get_daily_prices(stock_code, days=20)
+
+                if len(daily_prices) < 15:  # Need at least 14+1 for RSI
+                    logger.debug("Insufficient price history for %s", stock_code)
+                    continue
+
+                # Calculate RSI
+                close_prices = [p["close"] for p in daily_prices]
+                rsi = self.analyzer.calculate_rsi(close_prices, period=14)
+
+                # Calculate volume ratio (today vs previous day avg)
+                if len(daily_prices) >= 2:
+                    prev_day_volume = daily_prices[-2]["volume"]
+                    current_volume = stock.get("volume", 0) or daily_prices[-1]["volume"]
+                    volume_ratio = (
+                        current_volume / prev_day_volume if prev_day_volume > 0 else 1.0
+                    )
+                else:
+                    volume_ratio = stock.get("volume_increase_rate", 0) / 100 + 1  # Fallback
+
+                # Apply filters
+                volume_qualified = volume_ratio >= self.vol_multiplier
+                rsi_oversold = rsi < self.rsi_oversold
+                rsi_momentum = rsi > self.rsi_momentum
+
+                if volume_qualified and (rsi_oversold or rsi_momentum):
+                    signal = "oversold" if rsi_oversold else "momentum"
+
+                    # Calculate composite score
+                    # Higher score for: extreme RSI + high volume
+                    rsi_extremity = abs(rsi - 50) / 50  # 0-1 scale
+                    volume_score = min(volume_ratio / 5, 1.0)  # Cap at 5x
+                    score = (rsi_extremity * 0.6 + volume_score * 0.4) * 100
+
+                    candidates.append(
+                        ScanCandidate(
+                            stock_code=stock_code,
+                            name=stock.get("name", stock_code),
+                            price=stock.get("price", daily_prices[-1]["close"]),
+                            volume=current_volume,
+                            volume_ratio=volume_ratio,
+                            rsi=rsi,
+                            signal=signal,
+                            score=score,
+                        )
+                    )
+
+                    logger.info(
+                        "Qualified: %s (%s) RSI=%.1f vol=%.1fx signal=%s score=%.1f",
+                        stock_code,
+                        stock.get("name", ""),
+                        rsi,
+                        volume_ratio,
+                        signal,
+                        score,
+                    )
+
+            except ConnectionError as exc:
+                logger.warning("Failed to analyze %s: %s", stock_code, exc)
+                continue
+            except Exception as exc:
+                logger.error("Unexpected error analyzing %s: %s", stock_code, exc)
+                continue
+
+        # Sort by score and return top N
+        candidates.sort(key=lambda c: c.score, reverse=True)
+        return candidates[: self.top_n]
+
+    def get_stock_codes(self, candidates: list[ScanCandidate]) -> list[str]:
+        """Extract stock codes from candidates for watchlist update.
+
+        Args:
+            candidates: List of scan candidates
+
+        Returns:
+            List of stock codes
+        """
+        return [c.stock_code for c in candidates]

src/analysis/volatility.py

@@ -124,6 +124,54 @@ class VolatilityAnalyzer:
             return 1.0
         return current_volume / avg_volume
 
+    def calculate_rsi(
+        self,
+        close_prices: list[float],
+        period: int = 14,
+    ) -> float:
+        """Calculate Relative Strength Index (RSI) using Wilder's smoothing.
+
+        Args:
+            close_prices: List of closing prices (oldest to newest, minimum period+1 values)
+            period: RSI period (default 14)
+
+        Returns:
+            RSI value between 0 and 100, or 50.0 (neutral) if insufficient data
+
+        Examples:
+            >>> analyzer = VolatilityAnalyzer()
+            >>> prices = [100 - i * 0.5 for i in range(20)]  # Downtrend
+            >>> rsi = analyzer.calculate_rsi(prices)
+            >>> assert rsi < 50  # Oversold territory
+        """
+        if len(close_prices) < period + 1:
+            return 50.0  # Neutral RSI if insufficient data
+
+        # Calculate price changes
+        changes = [close_prices[i] - close_prices[i - 1] for i in range(1, len(close_prices))]
+
+        # Separate gains and losses
+        gains = [max(0.0, change) for change in changes]
+        losses = [max(0.0, -change) for change in changes]
+
+        # Calculate initial average gain/loss (simple average for first period)
+        avg_gain = sum(gains[:period]) / period
+        avg_loss = sum(losses[:period]) / period
+
+        # Apply Wilder's smoothing for remaining periods
+        for i in range(period, len(changes)):
+            avg_gain = (avg_gain * (period - 1) + gains[i]) / period
+            avg_loss = (avg_loss * (period - 1) + losses[i]) / period
+
+        # Calculate RS and RSI
+        if avg_loss == 0:
+            return 100.0  # All gains, maximum RSI
+
+        rs = avg_gain / avg_loss
+        rsi = 100 - (100 / (1 + rs))
+
+        return rsi
+
     def calculate_pv_divergence(
         self,
         price_change: float,

src/backup/__init__.py

@@ -0,0 +1,21 @@
+"""Backup and disaster recovery system for long-term sustainability.
+
+This module provides:
+- Automated database backups (daily, weekly, monthly)
+- Multi-format exports (JSON, CSV, Parquet)
+- Cloud storage integration (S3-compatible)
+- Health monitoring and alerts
+"""
+
+from src.backup.exporter import BackupExporter, ExportFormat
+from src.backup.scheduler import BackupScheduler, BackupPolicy
+from src.backup.cloud_storage import CloudStorage, S3Config
+
+__all__ = [
+    "BackupExporter",
+    "ExportFormat",
+    "BackupScheduler",
+    "BackupPolicy",
+    "CloudStorage",
+    "S3Config",
+]

src/backup/cloud_storage.py

@@ -0,0 +1,274 @@
+"""Cloud storage integration for off-site backups.
+
+Supports S3-compatible storage providers:
+- AWS S3
+- MinIO
+- Backblaze B2
+- DigitalOcean Spaces
+- Cloudflare R2
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class S3Config:
+    """Configuration for S3-compatible storage."""
+
+    endpoint_url: str | None  # None for AWS S3, custom URL for others
+    access_key: str
+    secret_key: str
+    bucket_name: str
+    region: str = "us-east-1"
+    use_ssl: bool = True
+
+
+class CloudStorage:
+    """Upload backups to S3-compatible cloud storage."""
+
+    def __init__(self, config: S3Config) -> None:
+        """Initialize cloud storage client.
+
+        Args:
+            config: S3 configuration
+
+        Raises:
+            ImportError: If boto3 is not installed
+        """
+        try:
+            import boto3
+        except ImportError:
+            raise ImportError(
+                "boto3 is required for cloud storage. Install with: pip install boto3"
+            )
+
+        self.config = config
+        self.client = boto3.client(
+            "s3",
+            endpoint_url=config.endpoint_url,
+            aws_access_key_id=config.access_key,
+            aws_secret_access_key=config.secret_key,
+            region_name=config.region,
+            use_ssl=config.use_ssl,
+        )
+
+    def upload_file(
+        self,
+        file_path: Path,
+        object_key: str | None = None,
+        metadata: dict[str, str] | None = None,
+    ) -> str:
+        """Upload a file to cloud storage.
+
+        Args:
+            file_path: Local file to upload
+            object_key: S3 object key (default: filename)
+            metadata: Optional metadata to attach
+
+        Returns:
+            S3 object key
+
+        Raises:
+            FileNotFoundError: If file doesn't exist
+            Exception: If upload fails
+        """
+        if not file_path.exists():
+            raise FileNotFoundError(f"File not found: {file_path}")
+
+        if object_key is None:
+            object_key = file_path.name
+
+        extra_args: dict[str, Any] = {}
+
+        # Add server-side encryption
+        extra_args["ServerSideEncryption"] = "AES256"
+
+        # Add metadata if provided
+        if metadata:
+            extra_args["Metadata"] = metadata
+
+        logger.info("Uploading %s to s3://%s/%s", file_path.name, self.config.bucket_name, object_key)
+
+        try:
+            self.client.upload_file(
+                str(file_path),
+                self.config.bucket_name,
+                object_key,
+                ExtraArgs=extra_args,
+            )
+            logger.info("Upload successful: %s", object_key)
+            return object_key
+        except Exception as exc:
+            logger.error("Upload failed: %s", exc)
+            raise
+
+    def download_file(self, object_key: str, local_path: Path) -> Path:
+        """Download a file from cloud storage.
+
+        Args:
+            object_key: S3 object key
+            local_path: Local destination path
+
+        Returns:
+            Path to downloaded file
+
+        Raises:
+            Exception: If download fails
+        """
+        local_path.parent.mkdir(parents=True, exist_ok=True)
+
+        logger.info("Downloading s3://%s/%s to %s", self.config.bucket_name, object_key, local_path)
+
+        try:
+            self.client.download_file(
+                self.config.bucket_name,
+                object_key,
+                str(local_path),
+            )
+            logger.info("Download successful: %s", local_path)
+            return local_path
+        except Exception as exc:
+            logger.error("Download failed: %s", exc)
+            raise
+
+    def list_files(self, prefix: str = "") -> list[dict[str, Any]]:
+        """List files in cloud storage.
+
+        Args:
+            prefix: Filter by object key prefix
+
+        Returns:
+            List of file metadata dictionaries
+        """
+        try:
+            response = self.client.list_objects_v2(
+                Bucket=self.config.bucket_name,
+                Prefix=prefix,
+            )
+
+            if "Contents" not in response:
+                return []
+
+            files = []
+            for obj in response["Contents"]:
+                files.append(
+                    {
+                        "key": obj["Key"],
+                        "size_bytes": obj["Size"],
+                        "last_modified": obj["LastModified"],
+                        "etag": obj["ETag"],
+                    }
+                )
+
+            return files
+        except Exception as exc:
+            logger.error("Failed to list files: %s", exc)
+            raise
+
+    def delete_file(self, object_key: str) -> None:
+        """Delete a file from cloud storage.
+
+        Args:
+            object_key: S3 object key
+
+        Raises:
+            Exception: If deletion fails
+        """
+        logger.info("Deleting s3://%s/%s", self.config.bucket_name, object_key)
+
+        try:
+            self.client.delete_object(
+                Bucket=self.config.bucket_name,
+                Key=object_key,
+            )
+            logger.info("Deletion successful: %s", object_key)
+        except Exception as exc:
+            logger.error("Deletion failed: %s", exc)
+            raise
+
+    def get_storage_stats(self) -> dict[str, Any]:
+        """Get cloud storage statistics.
+
+        Returns:
+            Dictionary with storage stats
+        """
+        try:
+            files = self.list_files()
+
+            total_size = sum(f["size_bytes"] for f in files)
+            total_count = len(files)
+
+            return {
+                "total_files": total_count,
+                "total_size_bytes": total_size,
+                "total_size_mb": total_size / 1024 / 1024,
+                "total_size_gb": total_size / 1024 / 1024 / 1024,
+            }
+        except Exception as exc:
+            logger.error("Failed to get storage stats: %s", exc)
+            return {
+                "error": str(exc),
+                "total_files": 0,
+                "total_size_bytes": 0,
+            }
+
+    def verify_connection(self) -> bool:
+        """Verify connection to cloud storage.
+
+        Returns:
+            True if connection is successful
+        """
+        try:
+            self.client.head_bucket(Bucket=self.config.bucket_name)
+            logger.info("Cloud storage connection verified")
+            return True
+        except Exception as exc:
+            logger.error("Cloud storage connection failed: %s", exc)
+            return False
+
+    def create_bucket_if_not_exists(self) -> None:
+        """Create storage bucket if it doesn't exist.
+
+        Raises:
+            Exception: If bucket creation fails
+        """
+        try:
+            self.client.head_bucket(Bucket=self.config.bucket_name)
+            logger.info("Bucket already exists: %s", self.config.bucket_name)
+        except self.client.exceptions.NoSuchBucket:
+            logger.info("Creating bucket: %s", self.config.bucket_name)
+            if self.config.region == "us-east-1":
+                # us-east-1 requires special handling
+                self.client.create_bucket(Bucket=self.config.bucket_name)
+            else:
+                self.client.create_bucket(
+                    Bucket=self.config.bucket_name,
+                    CreateBucketConfiguration={"LocationConstraint": self.config.region},
+                )
+            logger.info("Bucket created successfully")
+        except Exception as exc:
+            logger.error("Failed to verify/create bucket: %s", exc)
+            raise
+
+    def enable_versioning(self) -> None:
+        """Enable versioning on the bucket.
+
+        Raises:
+            Exception: If versioning enablement fails
+        """
+        try:
+            self.client.put_bucket_versioning(
+                Bucket=self.config.bucket_name,
+                VersioningConfiguration={"Status": "Enabled"},
+            )
+            logger.info("Versioning enabled for bucket: %s", self.config.bucket_name)
+        except Exception as exc:
+            logger.error("Failed to enable versioning: %s", exc)
+            raise

src/backup/exporter.py

@@ -0,0 +1,326 @@
+"""Multi-format database exporter for backups.
+
+Supports JSON, CSV, and Parquet formats for different use cases:
+- JSON: Human-readable, easy to inspect
+- CSV: Analysis tools (Excel, pandas)
+- Parquet: Big data tools (Spark, DuckDB)
+"""
+
+from __future__ import annotations
+
+import csv
+import gzip
+import json
+import logging
+import sqlite3
+from datetime import UTC, datetime
+from enum import Enum
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+class ExportFormat(str, Enum):
+    """Supported export formats."""
+
+    JSON = "json"
+    CSV = "csv"
+    PARQUET = "parquet"
+
+
+class BackupExporter:
+    """Export database to multiple formats."""
+
+    def __init__(self, db_path: str) -> None:
+        """Initialize the exporter.
+
+        Args:
+            db_path: Path to SQLite database
+        """
+        self.db_path = db_path
+
+    def export_all(
+        self,
+        output_dir: Path,
+        formats: list[ExportFormat] | None = None,
+        compress: bool = True,
+        incremental_since: datetime | None = None,
+    ) -> dict[ExportFormat, Path]:
+        """Export database to multiple formats.
+
+        Args:
+            output_dir: Directory to write export files
+            formats: List of formats to export (default: all)
+            compress: Whether to gzip compress exports
+            incremental_since: Only export records after this timestamp
+
+        Returns:
+            Dictionary mapping format to output file path
+        """
+        if formats is None:
+            formats = [ExportFormat.JSON, ExportFormat.CSV, ExportFormat.PARQUET]
+
+        output_dir.mkdir(parents=True, exist_ok=True)
+        timestamp = datetime.now(UTC).strftime("%Y%m%d_%H%M%S")
+
+        results: dict[ExportFormat, Path] = {}
+
+        for fmt in formats:
+            try:
+                output_file = self._export_format(
+                    fmt, output_dir, timestamp, compress, incremental_since
+                )
+                results[fmt] = output_file
+                logger.info("Exported to %s: %s", fmt.value, output_file)
+            except Exception as exc:
+                logger.error("Failed to export to %s: %s", fmt.value, exc)
+
+        return results
+
+    def _export_format(
+        self,
+        fmt: ExportFormat,
+        output_dir: Path,
+        timestamp: str,
+        compress: bool,
+        incremental_since: datetime | None,
+    ) -> Path:
+        """Export to a specific format.
+
+        Args:
+            fmt: Export format
+            output_dir: Output directory
+            timestamp: Timestamp string for filename
+            compress: Whether to compress
+            incremental_since: Incremental export cutoff
+
+        Returns:
+            Path to output file
+        """
+        if fmt == ExportFormat.JSON:
+            return self._export_json(output_dir, timestamp, compress, incremental_since)
+        elif fmt == ExportFormat.CSV:
+            return self._export_csv(output_dir, timestamp, compress, incremental_since)
+        elif fmt == ExportFormat.PARQUET:
+            return self._export_parquet(
+                output_dir, timestamp, compress, incremental_since
+            )
+        else:
+            raise ValueError(f"Unsupported format: {fmt}")
+
+    def _get_trades(
+        self, incremental_since: datetime | None = None
+    ) -> list[dict[str, Any]]:
+        """Fetch trades from database.
+
+        Args:
+            incremental_since: Only fetch trades after this timestamp
+
+        Returns:
+            List of trade records
+        """
+        conn = sqlite3.connect(self.db_path)
+        conn.row_factory = sqlite3.Row
+
+        if incremental_since:
+            cursor = conn.execute(
+                "SELECT * FROM trades WHERE timestamp > ?",
+                (incremental_since.isoformat(),),
+            )
+        else:
+            cursor = conn.execute("SELECT * FROM trades")
+
+        trades = [dict(row) for row in cursor.fetchall()]
+        conn.close()
+
+        return trades
+
+    def _export_json(
+        self,
+        output_dir: Path,
+        timestamp: str,
+        compress: bool,
+        incremental_since: datetime | None,
+    ) -> Path:
+        """Export to JSON format.
+
+        Args:
+            output_dir: Output directory
+            timestamp: Timestamp for filename
+            compress: Whether to gzip
+            incremental_since: Incremental cutoff
+
+        Returns:
+            Path to output file
+        """
+        trades = self._get_trades(incremental_since)
+
+        filename = f"trades_{timestamp}.json"
+        if compress:
+            filename += ".gz"
+
+        output_file = output_dir / filename
+
+        data = {
+            "export_timestamp": datetime.now(UTC).isoformat(),
+            "incremental_since": (
+                incremental_since.isoformat() if incremental_since else None
+            ),
+            "record_count": len(trades),
+            "trades": trades,
+        }
+
+        if compress:
+            with gzip.open(output_file, "wt", encoding="utf-8") as f:
+                json.dump(data, f, indent=2, ensure_ascii=False)
+        else:
+            with open(output_file, "w", encoding="utf-8") as f:
+                json.dump(data, f, indent=2, ensure_ascii=False)
+
+        return output_file
+
+    def _export_csv(
+        self,
+        output_dir: Path,
+        timestamp: str,
+        compress: bool,
+        incremental_since: datetime | None,
+    ) -> Path:
+        """Export to CSV format.
+
+        Args:
+            output_dir: Output directory
+            timestamp: Timestamp for filename
+            compress: Whether to gzip
+            incremental_since: Incremental cutoff
+
+        Returns:
+            Path to output file
+        """
+        trades = self._get_trades(incremental_since)
+
+        filename = f"trades_{timestamp}.csv"
+        if compress:
+            filename += ".gz"
+
+        output_file = output_dir / filename
+
+        if not trades:
+            # Write empty CSV with headers
+            if compress:
+                with gzip.open(output_file, "wt", encoding="utf-8", newline="") as f:
+                    writer = csv.writer(f)
+                    writer.writerow(
+                        [
+                            "timestamp",
+                            "stock_code",
+                            "action",
+                            "quantity",
+                            "price",
+                            "confidence",
+                            "rationale",
+                            "pnl",
+                        ]
+                    )
+            else:
+                with open(output_file, "w", encoding="utf-8", newline="") as f:
+                    writer = csv.writer(f)
+                    writer.writerow(
+                        [
+                            "timestamp",
+                            "stock_code",
+                            "action",
+                            "quantity",
+                            "price",
+                            "confidence",
+                            "rationale",
+                            "pnl",
+                        ]
+                    )
+            return output_file
+
+        # Get column names from first trade
+        fieldnames = list(trades[0].keys())
+
+        if compress:
+            with gzip.open(output_file, "wt", encoding="utf-8", newline="") as f:
+                writer = csv.DictWriter(f, fieldnames=fieldnames)
+                writer.writeheader()
+                writer.writerows(trades)
+        else:
+            with open(output_file, "w", encoding="utf-8", newline="") as f:
+                writer = csv.DictWriter(f, fieldnames=fieldnames)
+                writer.writeheader()
+                writer.writerows(trades)
+
+        return output_file
+
+    def _export_parquet(
+        self,
+        output_dir: Path,
+        timestamp: str,
+        compress: bool,
+        incremental_since: datetime | None,
+    ) -> Path:
+        """Export to Parquet format.
+
+        Args:
+            output_dir: Output directory
+            timestamp: Timestamp for filename
+            compress: Whether to compress (Parquet has built-in compression)
+            incremental_since: Incremental cutoff
+
+        Returns:
+            Path to output file
+        """
+        trades = self._get_trades(incremental_since)
+
+        filename = f"trades_{timestamp}.parquet"
+        output_file = output_dir / filename
+
+        try:
+            import pyarrow as pa
+            import pyarrow.parquet as pq
+        except ImportError:
+            raise ImportError(
+                "pyarrow is required for Parquet export. "
+                "Install with: pip install pyarrow"
+            )
+
+        # Convert to pyarrow table
+        table = pa.Table.from_pylist(trades)
+
+        # Write with compression
+        compression = "gzip" if compress else "none"
+        pq.write_table(table, output_file, compression=compression)
+
+        return output_file
+
+    def get_export_stats(self) -> dict[str, Any]:
+        """Get statistics about exportable data.
+
+        Returns:
+            Dictionary with data statistics
+        """
+        conn = sqlite3.connect(self.db_path)
+        cursor = conn.cursor()
+
+        stats = {}
+
+        # Total trades
+        cursor.execute("SELECT COUNT(*) FROM trades")
+        stats["total_trades"] = cursor.fetchone()[0]
+
+        # Date range
+        cursor.execute("SELECT MIN(timestamp), MAX(timestamp) FROM trades")
+        min_date, max_date = cursor.fetchone()
+        stats["date_range"] = {"earliest": min_date, "latest": max_date}
+
+        # Database size
+        cursor.execute("SELECT page_count * page_size FROM pragma_page_count(), pragma_page_size()")
+        stats["db_size_bytes"] = cursor.fetchone()[0]
+
+        conn.close()
+
+        return stats

src/backup/health_monitor.py

@@ -0,0 +1,282 @@
+"""Health monitoring for backup system.
+
+Checks:
+- Database accessibility and integrity
+- Disk space availability
+- Backup success/failure tracking
+- Self-healing capabilities
+"""
+
+from __future__ import annotations
+
+import logging
+import shutil
+import sqlite3
+from dataclasses import dataclass
+from datetime import UTC, datetime, timedelta
+from enum import Enum
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+class HealthStatus(str, Enum):
+    """Health check status."""
+
+    HEALTHY = "healthy"
+    DEGRADED = "degraded"
+    UNHEALTHY = "unhealthy"
+
+
+@dataclass
+class HealthCheckResult:
+    """Result of a health check."""
+
+    status: HealthStatus
+    message: str
+    details: dict[str, Any] | None = None
+    timestamp: datetime | None = None
+
+    def __post_init__(self) -> None:
+        if self.timestamp is None:
+            self.timestamp = datetime.now(UTC)
+
+
+class HealthMonitor:
+    """Monitor system health and backup status."""
+
+    def __init__(
+        self,
+        db_path: str,
+        backup_dir: Path,
+        min_disk_space_gb: float = 10.0,
+        max_backup_age_hours: int = 25,  # Daily backups should be < 25 hours old
+    ) -> None:
+        """Initialize health monitor.
+
+        Args:
+            db_path: Path to SQLite database
+            backup_dir: Backup directory
+            min_disk_space_gb: Minimum required disk space in GB
+            max_backup_age_hours: Maximum acceptable backup age in hours
+        """
+        self.db_path = Path(db_path)
+        self.backup_dir = backup_dir
+        self.min_disk_space_bytes = int(min_disk_space_gb * 1024 * 1024 * 1024)
+        self.max_backup_age = timedelta(hours=max_backup_age_hours)
+
+    def check_database_health(self) -> HealthCheckResult:
+        """Check database accessibility and integrity.
+
+        Returns:
+            HealthCheckResult
+        """
+        # Check if database exists
+        if not self.db_path.exists():
+            return HealthCheckResult(
+                status=HealthStatus.UNHEALTHY,
+                message=f"Database not found: {self.db_path}",
+            )
+
+        # Check if database is accessible
+        try:
+            conn = sqlite3.connect(str(self.db_path))
+            cursor = conn.cursor()
+
+            # Run integrity check
+            cursor.execute("PRAGMA integrity_check")
+            result = cursor.fetchone()[0]
+
+            if result != "ok":
+                conn.close()
+                return HealthCheckResult(
+                    status=HealthStatus.UNHEALTHY,
+                    message=f"Database integrity check failed: {result}",
+                )
+
+            # Get database size
+            cursor.execute(
+                "SELECT page_count * page_size FROM pragma_page_count(), pragma_page_size()"
+            )
+            db_size = cursor.fetchone()[0]
+
+            # Get row counts
+            cursor.execute("SELECT COUNT(*) FROM trades")
+            trade_count = cursor.fetchone()[0]
+
+            conn.close()
+
+            return HealthCheckResult(
+                status=HealthStatus.HEALTHY,
+                message="Database is healthy",
+                details={
+                    "size_bytes": db_size,
+                    "size_mb": db_size / 1024 / 1024,
+                    "trade_count": trade_count,
+                },
+            )
+
+        except sqlite3.Error as exc:
+            return HealthCheckResult(
+                status=HealthStatus.UNHEALTHY,
+                message=f"Database access error: {exc}",
+            )
+
+    def check_disk_space(self) -> HealthCheckResult:
+        """Check available disk space.
+
+        Returns:
+            HealthCheckResult
+        """
+        try:
+            stat = shutil.disk_usage(self.backup_dir)
+
+            free_gb = stat.free / 1024 / 1024 / 1024
+            total_gb = stat.total / 1024 / 1024 / 1024
+            used_percent = (stat.used / stat.total) * 100
+
+            if stat.free < self.min_disk_space_bytes:
+                return HealthCheckResult(
+                    status=HealthStatus.UNHEALTHY,
+                    message=f"Low disk space: {free_gb:.2f} GB free (minimum: {self.min_disk_space_bytes / 1024 / 1024 / 1024:.2f} GB)",
+                    details={
+                        "free_gb": free_gb,
+                        "total_gb": total_gb,
+                        "used_percent": used_percent,
+                    },
+                )
+            elif stat.free < self.min_disk_space_bytes * 2:
+                return HealthCheckResult(
+                    status=HealthStatus.DEGRADED,
+                    message=f"Disk space low: {free_gb:.2f} GB free",
+                    details={
+                        "free_gb": free_gb,
+                        "total_gb": total_gb,
+                        "used_percent": used_percent,
+                    },
+                )
+            else:
+                return HealthCheckResult(
+                    status=HealthStatus.HEALTHY,
+                    message=f"Disk space healthy: {free_gb:.2f} GB free",
+                    details={
+                        "free_gb": free_gb,
+                        "total_gb": total_gb,
+                        "used_percent": used_percent,
+                    },
+                )
+
+        except Exception as exc:
+            return HealthCheckResult(
+                status=HealthStatus.UNHEALTHY,
+                message=f"Failed to check disk space: {exc}",
+            )
+
+    def check_backup_recency(self) -> HealthCheckResult:
+        """Check if backups are recent enough.
+
+        Returns:
+            HealthCheckResult
+        """
+        daily_dir = self.backup_dir / "daily"
+
+        if not daily_dir.exists():
+            return HealthCheckResult(
+                status=HealthStatus.DEGRADED,
+                message="Daily backup directory not found",
+            )
+
+        # Find most recent backup
+        backups = sorted(daily_dir.glob("*.db"), key=lambda p: p.stat().st_mtime, reverse=True)
+
+        if not backups:
+            return HealthCheckResult(
+                status=HealthStatus.UNHEALTHY,
+                message="No daily backups found",
+            )
+
+        most_recent = backups[0]
+        mtime = datetime.fromtimestamp(most_recent.stat().st_mtime, tz=UTC)
+        age = datetime.now(UTC) - mtime
+
+        if age > self.max_backup_age:
+            return HealthCheckResult(
+                status=HealthStatus.DEGRADED,
+                message=f"Most recent backup is {age.total_seconds() / 3600:.1f} hours old",
+                details={
+                    "backup_file": most_recent.name,
+                    "age_hours": age.total_seconds() / 3600,
+                    "threshold_hours": self.max_backup_age.total_seconds() / 3600,
+                },
+            )
+        else:
+            return HealthCheckResult(
+                status=HealthStatus.HEALTHY,
+                message=f"Recent backup found ({age.total_seconds() / 3600:.1f} hours old)",
+                details={
+                    "backup_file": most_recent.name,
+                    "age_hours": age.total_seconds() / 3600,
+                },
+            )
+
+    def run_all_checks(self) -> dict[str, HealthCheckResult]:
+        """Run all health checks.
+
+        Returns:
+            Dictionary mapping check name to result
+        """
+        checks = {
+            "database": self.check_database_health(),
+            "disk_space": self.check_disk_space(),
+            "backup_recency": self.check_backup_recency(),
+        }
+
+        # Log results
+        for check_name, result in checks.items():
+            if result.status == HealthStatus.UNHEALTHY:
+                logger.error("[%s] %s: %s", check_name, result.status.value, result.message)
+            elif result.status == HealthStatus.DEGRADED:
+                logger.warning("[%s] %s: %s", check_name, result.status.value, result.message)
+            else:
+                logger.info("[%s] %s: %s", check_name, result.status.value, result.message)
+
+        return checks
+
+    def get_overall_status(self) -> HealthStatus:
+        """Get overall system health status.
+
+        Returns:
+            HealthStatus (worst status from all checks)
+        """
+        checks = self.run_all_checks()
+
+        # Return worst status
+        if any(c.status == HealthStatus.UNHEALTHY for c in checks.values()):
+            return HealthStatus.UNHEALTHY
+        elif any(c.status == HealthStatus.DEGRADED for c in checks.values()):
+            return HealthStatus.DEGRADED
+        else:
+            return HealthStatus.HEALTHY
+
+    def get_health_report(self) -> dict[str, Any]:
+        """Get comprehensive health report.
+
+        Returns:
+            Dictionary with health report
+        """
+        checks = self.run_all_checks()
+        overall = self.get_overall_status()
+
+        return {
+            "overall_status": overall.value,
+            "timestamp": datetime.now(UTC).isoformat(),
+            "checks": {
+                name: {
+                    "status": result.status.value,
+                    "message": result.message,
+                    "details": result.details,
+                }
+                for name, result in checks.items()
+            },
+        }

src/backup/scheduler.py

@@ -0,0 +1,336 @@
+"""Backup scheduler for automated database backups.
+
+Implements backup policies:
+- Daily: Keep for 30 days (hot storage)
+- Weekly: Keep for 1 year (warm storage)
+- Monthly: Keep forever (cold storage)
+"""
+
+from __future__ import annotations
+
+import logging
+import shutil
+from dataclasses import dataclass
+from datetime import UTC, datetime, timedelta
+from enum import Enum
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+class BackupPolicy(str, Enum):
+    """Backup retention policies."""
+
+    DAILY = "daily"
+    WEEKLY = "weekly"
+    MONTHLY = "monthly"
+
+
+@dataclass
+class BackupMetadata:
+    """Metadata for a backup."""
+
+    timestamp: datetime
+    policy: BackupPolicy
+    file_path: Path
+    size_bytes: int
+    checksum: str | None = None
+
+
+class BackupScheduler:
+    """Manage automated database backups with retention policies."""
+
+    def __init__(
+        self,
+        db_path: str,
+        backup_dir: Path,
+        daily_retention_days: int = 30,
+        weekly_retention_days: int = 365,
+    ) -> None:
+        """Initialize the backup scheduler.
+
+        Args:
+            db_path: Path to SQLite database
+            backup_dir: Root directory for backups
+            daily_retention_days: Days to keep daily backups
+            weekly_retention_days: Days to keep weekly backups
+        """
+        self.db_path = Path(db_path)
+        self.backup_dir = backup_dir
+        self.daily_retention = timedelta(days=daily_retention_days)
+        self.weekly_retention = timedelta(days=weekly_retention_days)
+
+        # Create policy-specific directories
+        self.daily_dir = backup_dir / "daily"
+        self.weekly_dir = backup_dir / "weekly"
+        self.monthly_dir = backup_dir / "monthly"
+
+        for d in [self.daily_dir, self.weekly_dir, self.monthly_dir]:
+            d.mkdir(parents=True, exist_ok=True)
+
+    def create_backup(
+        self, policy: BackupPolicy, verify: bool = True
+    ) -> BackupMetadata:
+        """Create a database backup.
+
+        Args:
+            policy: Backup policy (daily/weekly/monthly)
+            verify: Whether to verify backup integrity
+
+        Returns:
+            BackupMetadata object
+
+        Raises:
+            FileNotFoundError: If database doesn't exist
+            OSError: If backup fails
+        """
+        if not self.db_path.exists():
+            raise FileNotFoundError(f"Database not found: {self.db_path}")
+
+        timestamp = datetime.now(UTC)
+        backup_filename = self._get_backup_filename(timestamp, policy)
+
+        # Determine output directory
+        if policy == BackupPolicy.DAILY:
+            output_dir = self.daily_dir
+        elif policy == BackupPolicy.WEEKLY:
+            output_dir = self.weekly_dir
+        else:  # MONTHLY
+            output_dir = self.monthly_dir
+
+        backup_path = output_dir / backup_filename
+
+        # Create backup (copy database file)
+        logger.info("Creating %s backup: %s", policy.value, backup_path)
+        shutil.copy2(self.db_path, backup_path)
+
+        # Get file size
+        size_bytes = backup_path.stat().st_size
+
+        # Verify backup if requested
+        checksum = None
+        if verify:
+            checksum = self._verify_backup(backup_path)
+
+        metadata = BackupMetadata(
+            timestamp=timestamp,
+            policy=policy,
+            file_path=backup_path,
+            size_bytes=size_bytes,
+            checksum=checksum,
+        )
+
+        logger.info(
+            "Backup created: %s (%.2f MB)",
+            backup_path.name,
+            size_bytes / 1024 / 1024,
+        )
+
+        return metadata
+
+    def _get_backup_filename(self, timestamp: datetime, policy: BackupPolicy) -> str:
+        """Generate backup filename.
+
+        Args:
+            timestamp: Backup timestamp
+            policy: Backup policy
+
+        Returns:
+            Filename string
+        """
+        ts_str = timestamp.strftime("%Y%m%d_%H%M%S")
+        return f"trade_logs_{policy.value}_{ts_str}.db"
+
+    def _verify_backup(self, backup_path: Path) -> str:
+        """Verify backup integrity using SQLite integrity check.
+
+        Args:
+            backup_path: Path to backup file
+
+        Returns:
+            Checksum string (MD5 hash)
+
+        Raises:
+            RuntimeError: If integrity check fails
+        """
+        import hashlib
+        import sqlite3
+
+        # Integrity check
+        try:
+            conn = sqlite3.connect(str(backup_path))
+            cursor = conn.cursor()
+            cursor.execute("PRAGMA integrity_check")
+            result = cursor.fetchone()[0]
+            conn.close()
+
+            if result != "ok":
+                raise RuntimeError(f"Integrity check failed: {result}")
+        except sqlite3.Error as exc:
+            raise RuntimeError(f"Failed to verify backup: {exc}")
+
+        # Calculate MD5 checksum
+        md5 = hashlib.md5()
+        with open(backup_path, "rb") as f:
+            for chunk in iter(lambda: f.read(8192), b""):
+                md5.update(chunk)
+
+        return md5.hexdigest()
+
+    def cleanup_old_backups(self) -> dict[BackupPolicy, int]:
+        """Remove backups older than retention policies.
+
+        Returns:
+            Dictionary mapping policy to number of backups removed
+        """
+        now = datetime.now(UTC)
+        removed_counts: dict[BackupPolicy, int] = {}
+
+        # Daily backups: remove older than retention
+        removed_counts[BackupPolicy.DAILY] = self._cleanup_directory(
+            self.daily_dir, now - self.daily_retention
+        )
+
+        # Weekly backups: remove older than retention
+        removed_counts[BackupPolicy.WEEKLY] = self._cleanup_directory(
+            self.weekly_dir, now - self.weekly_retention
+        )
+
+        # Monthly backups: never remove (kept forever)
+        removed_counts[BackupPolicy.MONTHLY] = 0
+
+        total = sum(removed_counts.values())
+        if total > 0:
+            logger.info("Cleaned up %d old backup(s)", total)
+
+        return removed_counts
+
+    def _cleanup_directory(self, directory: Path, cutoff: datetime) -> int:
+        """Remove backups older than cutoff date.
+
+        Args:
+            directory: Directory to clean
+            cutoff: Remove files older than this
+
+        Returns:
+            Number of files removed
+        """
+        removed = 0
+
+        for backup_file in directory.glob("*.db"):
+            # Get file modification time
+            mtime = datetime.fromtimestamp(backup_file.stat().st_mtime, tz=UTC)
+
+            if mtime < cutoff:
+                logger.debug("Removing old backup: %s", backup_file.name)
+                backup_file.unlink()
+                removed += 1
+
+        return removed
+
+    def list_backups(
+        self, policy: BackupPolicy | None = None
+    ) -> list[BackupMetadata]:
+        """List available backups.
+
+        Args:
+            policy: Filter by policy (None for all)
+
+        Returns:
+            List of BackupMetadata objects
+        """
+        backups: list[BackupMetadata] = []
+
+        policies_to_check = (
+            [policy] if policy else [BackupPolicy.DAILY, BackupPolicy.WEEKLY, BackupPolicy.MONTHLY]
+        )
+
+        for pol in policies_to_check:
+            if pol == BackupPolicy.DAILY:
+                directory = self.daily_dir
+            elif pol == BackupPolicy.WEEKLY:
+                directory = self.weekly_dir
+            else:
+                directory = self.monthly_dir
+
+            for backup_file in sorted(directory.glob("*.db")):
+                mtime = datetime.fromtimestamp(backup_file.stat().st_mtime, tz=UTC)
+                size = backup_file.stat().st_size
+
+                backups.append(
+                    BackupMetadata(
+                        timestamp=mtime,
+                        policy=pol,
+                        file_path=backup_file,
+                        size_bytes=size,
+                    )
+                )
+
+        # Sort by timestamp (newest first)
+        backups.sort(key=lambda b: b.timestamp, reverse=True)
+
+        return backups
+
+    def get_backup_stats(self) -> dict[str, Any]:
+        """Get backup statistics.
+
+        Returns:
+            Dictionary with backup stats
+        """
+        stats: dict[str, Any] = {}
+
+        for policy in BackupPolicy:
+            if policy == BackupPolicy.DAILY:
+                directory = self.daily_dir
+            elif policy == BackupPolicy.WEEKLY:
+                directory = self.weekly_dir
+            else:
+                directory = self.monthly_dir
+
+            backups = list(directory.glob("*.db"))
+            total_size = sum(b.stat().st_size for b in backups)
+
+            stats[policy.value] = {
+                "count": len(backups),
+                "total_size_bytes": total_size,
+                "total_size_mb": total_size / 1024 / 1024,
+            }
+
+        return stats
+
+    def restore_backup(self, backup_metadata: BackupMetadata, verify: bool = True) -> None:
+        """Restore database from backup.
+
+        Args:
+            backup_metadata: Backup to restore
+            verify: Whether to verify restored database
+
+        Raises:
+            FileNotFoundError: If backup file doesn't exist
+            RuntimeError: If verification fails
+        """
+        if not backup_metadata.file_path.exists():
+            raise FileNotFoundError(f"Backup not found: {backup_metadata.file_path}")
+
+        # Create backup of current database
+        if self.db_path.exists():
+            backup_current = self.db_path.with_suffix(".db.before_restore")
+            logger.info("Backing up current database to: %s", backup_current)
+            shutil.copy2(self.db_path, backup_current)
+
+        # Restore backup
+        logger.info("Restoring backup: %s", backup_metadata.file_path.name)
+        shutil.copy2(backup_metadata.file_path, self.db_path)
+
+        # Verify restored database
+        if verify:
+            try:
+                self._verify_backup(self.db_path)
+                logger.info("Backup restored and verified successfully")
+            except RuntimeError as exc:
+                # Restore failed, revert to backup
+                if backup_current.exists():
+                    logger.error("Restore verification failed, reverting: %s", exc)
+                    shutil.copy2(backup_current, self.db_path)
+                raise

src/brain/gemini_client.py

@@ -525,3 +525,233 @@ class GeminiClient:
             DecisionCache instance or None if caching disabled
         """
         return self._cache
+
+    # ------------------------------------------------------------------
+    # Batch Decision Making (for daily trading mode)
+    # ------------------------------------------------------------------
+
+    async def decide_batch(
+        self, stocks_data: list[dict[str, Any]]
+    ) -> dict[str, TradeDecision]:
+        """Make decisions for multiple stocks in a single API call.
+
+        This is designed for daily trading mode to minimize API usage
+        when working with Gemini Free tier (20 calls/day limit).
+
+        Args:
+            stocks_data: List of market data dictionaries, each with:
+                - stock_code: Stock ticker
+                - current_price: Current price
+                - market_name: Market name (optional)
+                - foreigner_net: Foreigner net buy/sell (optional)
+
+        Returns:
+            Dictionary mapping stock_code to TradeDecision
+
+        Example:
+            >>> stocks_data = [
+            ...     {"stock_code": "AAPL", "current_price": 185.5},
+            ...     {"stock_code": "MSFT", "current_price": 420.0},
+            ... ]
+            >>> decisions = await client.decide_batch(stocks_data)
+            >>> decisions["AAPL"].action
+            'BUY'
+        """
+        if not stocks_data:
+            return {}
+
+        # Build compressed batch prompt
+        market_name = stocks_data[0].get("market_name", "stock market")
+
+        # Format stock data as compact JSON array
+        compact_stocks = []
+        for stock in stocks_data:
+            compact = {
+                "code": stock["stock_code"],
+                "price": stock["current_price"],
+            }
+            if stock.get("foreigner_net", 0) != 0:
+                compact["frgn"] = stock["foreigner_net"]
+            compact_stocks.append(compact)
+
+        data_str = json.dumps(compact_stocks, ensure_ascii=False)
+
+        prompt = (
+            f"You are a professional {market_name} trading analyst.\n"
+            "Analyze the following stocks and decide whether to BUY, SELL, or HOLD each one.\n\n"
+            f"Stock Data: {data_str}\n\n"
+            "You MUST respond with ONLY a valid JSON array in this format:\n"
+            '[{"code": "AAPL", "action": "BUY", "confidence": 85, "rationale": "..."},\n'
+            ' {"code": "MSFT", "action": "HOLD", "confidence": 50, "rationale": "..."}, ...]\n\n'
+            "Rules:\n"
+            "- Return one decision object per stock\n"
+            "- action must be exactly: BUY, SELL, or HOLD\n"
+            "- confidence must be 0-100\n"
+            "- rationale should be concise (1-2 sentences)\n"
+            "- Do NOT wrap JSON in markdown code blocks\n"
+        )
+
+        # Estimate tokens
+        token_count = self._optimizer.estimate_tokens(prompt)
+        self._total_tokens_used += token_count
+
+        logger.info(
+            "Requesting batch decision for %d stocks from Gemini",
+            len(stocks_data),
+            extra={"estimated_tokens": token_count},
+        )
+
+        try:
+            response = await self._client.aio.models.generate_content(
+                model=self._model_name,
+                contents=prompt,
+            )
+            raw = response.text
+        except Exception as exc:
+            logger.error("Gemini API error in batch decision: %s", exc)
+            # Return HOLD for all stocks on API error
+            return {
+                stock["stock_code"]: TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale=f"API error: {exc}",
+                    token_count=token_count,
+                    cached=False,
+                )
+                for stock in stocks_data
+            }
+
+        # Parse batch response
+        return self._parse_batch_response(raw, stocks_data, token_count)
+
+    def _parse_batch_response(
+        self, raw: str, stocks_data: list[dict[str, Any]], token_count: int
+    ) -> dict[str, TradeDecision]:
+        """Parse batch response into a dictionary of decisions.
+
+        Args:
+            raw: Raw response from Gemini
+            stocks_data: Original stock data list
+            token_count: Token count for the request
+
+        Returns:
+            Dictionary mapping stock_code to TradeDecision
+        """
+        if not raw or not raw.strip():
+            logger.warning("Empty batch response from Gemini — defaulting all to HOLD")
+            return {
+                stock["stock_code"]: TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale="Empty response",
+                    token_count=0,
+                    cached=False,
+                )
+                for stock in stocks_data
+            }
+
+        # Strip markdown code fences if present
+        cleaned = raw.strip()
+        match = re.search(r"```(?:json)?\s*\n?(.*?)\n?```", cleaned, re.DOTALL)
+        if match:
+            cleaned = match.group(1).strip()
+
+        try:
+            data = json.loads(cleaned)
+        except json.JSONDecodeError:
+            logger.warning("Malformed JSON in batch response — defaulting all to HOLD")
+            return {
+                stock["stock_code"]: TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale="Malformed JSON response",
+                    token_count=0,
+                    cached=False,
+                )
+                for stock in stocks_data
+            }
+
+        if not isinstance(data, list):
+            logger.warning("Batch response is not a JSON array — defaulting all to HOLD")
+            return {
+                stock["stock_code"]: TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale="Invalid response format",
+                    token_count=0,
+                    cached=False,
+                )
+                for stock in stocks_data
+            }
+
+        # Build decision map
+        decisions: dict[str, TradeDecision] = {}
+        stock_codes = {stock["stock_code"] for stock in stocks_data}
+
+        for item in data:
+            if not isinstance(item, dict):
+                continue
+
+            code = item.get("code")
+            if not code or code not in stock_codes:
+                continue
+
+            # Validate required fields
+            if not all(k in item for k in ("action", "confidence", "rationale")):
+                logger.warning("Missing fields for %s — using HOLD", code)
+                decisions[code] = TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale="Missing required fields",
+                    token_count=0,
+                    cached=False,
+                )
+                continue
+
+            action = str(item["action"]).upper()
+            if action not in VALID_ACTIONS:
+                logger.warning("Invalid action '%s' for %s — forcing HOLD", action, code)
+                action = "HOLD"
+
+            confidence = int(item["confidence"])
+            rationale = str(item["rationale"])
+
+            # Enforce confidence threshold
+            if confidence < self._confidence_threshold:
+                logger.info(
+                    "Confidence %d < threshold %d for %s — forcing HOLD",
+                    confidence,
+                    self._confidence_threshold,
+                    code,
+                )
+                action = "HOLD"
+
+            decisions[code] = TradeDecision(
+                action=action,
+                confidence=confidence,
+                rationale=rationale,
+                token_count=token_count // len(stocks_data),  # Split token cost
+                cached=False,
+            )
+            self._total_decisions += 1
+
+        # Fill in missing stocks with HOLD
+        for stock in stocks_data:
+            code = stock["stock_code"]
+            if code not in decisions:
+                logger.warning("No decision for %s in batch response — using HOLD", code)
+                decisions[code] = TradeDecision(
+                    action="HOLD",
+                    confidence=0,
+                    rationale="Not found in batch response",
+                    token_count=0,
+                    cached=False,
+                )
+
+        logger.info(
+            "Batch decision completed for %d stocks",
+            len(decisions),
+            extra={"tokens": token_count},
+        )
+
+        return decisions

src/broker/kis_api.py

@@ -55,6 +55,9 @@ class KISBroker:
         self._session: aiohttp.ClientSession | None = None
         self._access_token: str | None = None
         self._token_expires_at: float = 0.0
+        self._token_lock = asyncio.Lock()
+        self._last_refresh_attempt: float = 0.0
+        self._refresh_cooldown: float = 60.0  # Seconds (matches KIS 1/minute limit)
         self._rate_limiter = LeakyBucket(settings.RATE_LIMIT_RPS)
 
     def _get_session(self) -> aiohttp.ClientSession:
@@ -80,30 +83,54 @@ class KISBroker:
     # ------------------------------------------------------------------
 
     async def _ensure_token(self) -> str:
-        """Return a valid access token, refreshing if expired."""
+        """Return a valid access token, refreshing if expired.
+
+        Uses a lock to prevent concurrent token refresh attempts that would
+        hit the API's 1-per-minute rate limit (EGW00133).
+        """
+        # Fast path: check without lock
         now = asyncio.get_event_loop().time()
         if self._access_token and now < self._token_expires_at:
             return self._access_token
 
-        logger.info("Refreshing KIS access token")
-        session = self._get_session()
-        url = f"{self._base_url}/oauth2/tokenP"
-        body = {
-            "grant_type": "client_credentials",
-            "appkey": self._app_key,
-            "appsecret": self._app_secret,
-        }
+        # Slow path: acquire lock and refresh
+        async with self._token_lock:
+            # Re-check after acquiring lock (another coroutine may have refreshed)
+            now = asyncio.get_event_loop().time()
+            if self._access_token and now < self._token_expires_at:
+                return self._access_token
 
-        async with session.post(url, json=body) as resp:
-            if resp.status != 200:
-                text = await resp.text()
-                raise ConnectionError(f"Token refresh failed ({resp.status}): {text}")
-            data = await resp.json()
+            # Check cooldown period (prevents hitting EGW00133: 1/minute limit)
+            time_since_last_attempt = now - self._last_refresh_attempt
+            if time_since_last_attempt < self._refresh_cooldown:
+                remaining = self._refresh_cooldown - time_since_last_attempt
+                error_msg = (
+                    f"Token refresh on cooldown. "
+                    f"Retry in {remaining:.1f}s (KIS allows 1/minute)"
+                )
+                logger.warning(error_msg)
+                raise ConnectionError(error_msg)
 
-        self._access_token = data["access_token"]
-        self._token_expires_at = now + data.get("expires_in", 86400) - 60  # 1-min buffer
-        logger.info("Token refreshed successfully")
-        return self._access_token
+            logger.info("Refreshing KIS access token")
+            self._last_refresh_attempt = now
+            session = self._get_session()
+            url = f"{self._base_url}/oauth2/tokenP"
+            body = {
+                "grant_type": "client_credentials",
+                "appkey": self._app_key,
+                "appsecret": self._app_secret,
+            }
+
+            async with session.post(url, json=body) as resp:
+                if resp.status != 200:
+                    text = await resp.text()
+                    raise ConnectionError(f"Token refresh failed ({resp.status}): {text}")
+                data = await resp.json()
+
+            self._access_token = data["access_token"]
+            self._token_expires_at = now + data.get("expires_in", 86400) - 60  # 1-min buffer
+            logger.info("Token refreshed successfully")
+            return self._access_token
 
     # ------------------------------------------------------------------
     # Hash Key (required for POST bodies)
@@ -111,6 +138,7 @@ class KISBroker:
 
     async def _get_hash_key(self, body: dict[str, Any]) -> str:
         """Request a hash key from KIS for POST request body signing."""
+        await self._rate_limiter.acquire()
         session = self._get_session()
         url = f"{self._base_url}/uapi/hashkey"
         headers = {
@@ -252,3 +280,153 @@ class KISBroker:
                 return data
         except (TimeoutError, aiohttp.ClientError) as exc:
             raise ConnectionError(f"Network error sending order: {exc}") from exc
+
+    async def fetch_market_rankings(
+        self,
+        ranking_type: str = "volume",
+        limit: int = 30,
+    ) -> list[dict[str, Any]]:
+        """Fetch market rankings from KIS API.
+
+        Args:
+            ranking_type: Type of ranking ("volume" or "fluctuation")
+            limit: Maximum number of results to return
+
+        Returns:
+            List of stock data dicts with keys: stock_code, name, price, volume,
+            change_rate, volume_increase_rate
+
+        Raises:
+            ConnectionError: If API request fails
+        """
+        await self._rate_limiter.acquire()
+        session = self._get_session()
+
+        # TR_ID for volume ranking
+        tr_id = "FHPST01710000" if ranking_type == "volume" else "FHPST01710100"
+        headers = await self._auth_headers(tr_id)
+
+        params = {
+            "FID_COND_MRKT_DIV_CODE": "J",  # Stock/ETF/ETN
+            "FID_COND_SCR_DIV_CODE": "20001",  # Volume surge
+            "FID_INPUT_ISCD": "0000",  # All stocks
+            "FID_DIV_CLS_CODE": "0",  # All types
+            "FID_BLNG_CLS_CODE": "0",
+            "FID_TRGT_CLS_CODE": "111111111",
+            "FID_TRGT_EXLS_CLS_CODE": "000000",
+            "FID_INPUT_PRICE_1": "0",
+            "FID_INPUT_PRICE_2": "0",
+            "FID_VOL_CNT": "0",
+            "FID_INPUT_DATE_1": "",
+        }
+
+        url = f"{self._base_url}/uapi/domestic-stock/v1/quotations/volume-rank"
+
+        try:
+            async with session.get(url, headers=headers, params=params) as resp:
+                if resp.status != 200:
+                    text = await resp.text()
+                    raise ConnectionError(
+                        f"fetch_market_rankings failed ({resp.status}): {text}"
+                    )
+                data = await resp.json()
+
+            # Parse response - output is a list of ranked stocks
+            def _safe_float(value: str | float | None, default: float = 0.0) -> float:
+                if value is None or value == "":
+                    return default
+                try:
+                    return float(value)
+                except (ValueError, TypeError):
+                    return default
+
+            rankings = []
+            for item in data.get("output", [])[:limit]:
+                rankings.append({
+                    "stock_code": item.get("mksc_shrn_iscd", ""),
+                    "name": item.get("hts_kor_isnm", ""),
+                    "price": _safe_float(item.get("stck_prpr", "0")),
+                    "volume": _safe_float(item.get("acml_vol", "0")),
+                    "change_rate": _safe_float(item.get("prdy_ctrt", "0")),
+                    "volume_increase_rate": _safe_float(item.get("vol_inrt", "0")),
+                })
+            return rankings
+
+        except (TimeoutError, aiohttp.ClientError) as exc:
+            raise ConnectionError(f"Network error fetching rankings: {exc}") from exc
+
+    async def get_daily_prices(
+        self,
+        stock_code: str,
+        days: int = 20,
+    ) -> list[dict[str, Any]]:
+        """Fetch daily OHLCV price history for a stock.
+
+        Args:
+            stock_code: 6-digit stock code
+            days: Number of trading days to fetch (default 20 for RSI calculation)
+
+        Returns:
+            List of daily price dicts with keys: date, open, high, low, close, volume
+            Sorted oldest to newest
+
+        Raises:
+            ConnectionError: If API request fails
+        """
+        await self._rate_limiter.acquire()
+        session = self._get_session()
+
+        headers = await self._auth_headers("FHKST03010100")
+
+        # Calculate date range (today and N days ago)
+        from datetime import datetime, timedelta
+        end_date = datetime.now().strftime("%Y%m%d")
+        start_date = (datetime.now() - timedelta(days=days + 10)).strftime("%Y%m%d")
+
+        params = {
+            "FID_COND_MRKT_DIV_CODE": "J",
+            "FID_INPUT_ISCD": stock_code,
+            "FID_INPUT_DATE_1": start_date,
+            "FID_INPUT_DATE_2": end_date,
+            "FID_PERIOD_DIV_CODE": "D",  # Daily
+            "FID_ORG_ADJ_PRC": "0",  # Adjusted price
+        }
+
+        url = f"{self._base_url}/uapi/domestic-stock/v1/quotations/inquire-daily-itemchartprice"
+
+        try:
+            async with session.get(url, headers=headers, params=params) as resp:
+                if resp.status != 200:
+                    text = await resp.text()
+                    raise ConnectionError(
+                        f"get_daily_prices failed ({resp.status}): {text}"
+                    )
+                data = await resp.json()
+
+            # Parse response
+            def _safe_float(value: str | float | None, default: float = 0.0) -> float:
+                if value is None or value == "":
+                    return default
+                try:
+                    return float(value)
+                except (ValueError, TypeError):
+                    return default
+
+            prices = []
+            for item in data.get("output2", []):
+                prices.append({
+                    "date": item.get("stck_bsop_date", ""),
+                    "open": _safe_float(item.get("stck_oprc", "0")),
+                    "high": _safe_float(item.get("stck_hgpr", "0")),
+                    "low": _safe_float(item.get("stck_lwpr", "0")),
+                    "close": _safe_float(item.get("stck_clpr", "0")),
+                    "volume": _safe_float(item.get("acml_vol", "0")),
+                })
+
+            # Sort oldest to newest (KIS returns newest first)
+            prices.reverse()
+
+            return prices[:days]  # Return only requested number of days
+
+        except (TimeoutError, aiohttp.ClientError) as exc:
+            raise ConnectionError(f"Network error fetching daily prices: {exc}") from exc

src/config.py

@@ -24,22 +24,64 @@ class Settings(BaseSettings):
     NEWS_API_PROVIDER: str = "alphavantage"  # "alphavantage" or "newsapi"
     MARKET_DATA_API_KEY: str | None = None
 
+    # Legacy field names (for backward compatibility)
+    ALPHA_VANTAGE_API_KEY: str | None = None
+    NEWSAPI_KEY: str | None = None
+
     # Risk Management
     CIRCUIT_BREAKER_PCT: float = Field(default=-3.0, le=0.0)
     FAT_FINGER_PCT: float = Field(default=30.0, gt=0.0, le=100.0)
     CONFIDENCE_THRESHOLD: int = Field(default=80, ge=0, le=100)
 
+    # Smart Scanner Configuration
+    RSI_OVERSOLD_THRESHOLD: int = Field(default=30, ge=0, le=50)
+    RSI_MOMENTUM_THRESHOLD: int = Field(default=70, ge=50, le=100)
+    VOL_MULTIPLIER: float = Field(default=2.0, gt=1.0, le=10.0)
+    SCANNER_TOP_N: int = Field(default=3, ge=1, le=10)
+
     # Database
     DB_PATH: str = "data/trade_logs.db"
 
     # Rate Limiting (requests per second for KIS API)
-    RATE_LIMIT_RPS: float = 10.0
+    # Conservative limit to avoid EGW00201 "초당 거래건수 초과" errors.
+    # KIS API real limit is ~2 RPS; 2.0 provides maximum safety.
+    RATE_LIMIT_RPS: float = 2.0
 
     # Trading mode
     MODE: str = Field(default="paper", pattern="^(paper|live)$")
 
+    # Trading frequency mode (daily = batch API calls, realtime = per-stock calls)
+    TRADE_MODE: str = Field(default="daily", pattern="^(daily|realtime)$")
+    DAILY_SESSIONS: int = Field(default=4, ge=1, le=10)
+    SESSION_INTERVAL_HOURS: int = Field(default=6, ge=1, le=24)
+
+    # Pre-Market Planner
+    PRE_MARKET_MINUTES: int = Field(default=30, ge=10, le=120)
+    MAX_SCENARIOS_PER_STOCK: int = Field(default=5, ge=1, le=10)
+    PLANNER_TIMEOUT_SECONDS: int = Field(default=60, ge=10, le=300)
+    DEFENSIVE_PLAYBOOK_ON_FAILURE: bool = True
+    RESCAN_INTERVAL_SECONDS: int = Field(default=300, ge=60, le=900)
+
     # Market selection (comma-separated market codes)
-    ENABLED_MARKETS: str = "KR"
+    ENABLED_MARKETS: str = "KR,US"
+
+    # Backup and Disaster Recovery (optional)
+    BACKUP_ENABLED: bool = True
+    BACKUP_DIR: str = "data/backups"
+    S3_ENDPOINT_URL: str | None = None  # For MinIO, Backblaze B2, etc.
+    S3_ACCESS_KEY: str | None = None
+    S3_SECRET_KEY: str | None = None
+    S3_BUCKET_NAME: str | None = None
+    S3_REGION: str = "us-east-1"
+
+    # Telegram Notifications (optional)
+    TELEGRAM_BOT_TOKEN: str | None = None
+    TELEGRAM_CHAT_ID: str | None = None
+    TELEGRAM_ENABLED: bool = True
+
+    # Telegram Commands (optional)
+    TELEGRAM_COMMANDS_ENABLED: bool = True
+    TELEGRAM_POLLING_INTERVAL: float = 1.0  # seconds
 
     model_config = {"env_file": ".env", "env_file_encoding": "utf-8"}
 

src/db.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import json
 import sqlite3
 from datetime import UTC, datetime
 from pathlib import Path
@@ -38,6 +39,8 @@ def init_db(db_path: str) -> sqlite3.Connection:
         conn.execute("ALTER TABLE trades ADD COLUMN market TEXT DEFAULT 'KR'")
     if "exchange_code" not in columns:
         conn.execute("ALTER TABLE trades ADD COLUMN exchange_code TEXT DEFAULT 'KRX'")
+    if "selection_context" not in columns:
+        conn.execute("ALTER TABLE trades ADD COLUMN selection_context TEXT")
 
     # Context tree tables for multi-layered memory management
     conn.execute(
@@ -88,6 +91,27 @@ def init_db(db_path: str) -> sqlite3.Connection:
         """
     )
 
+    # Playbook storage for pre-market strategy persistence
+    conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS playbooks (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            date TEXT NOT NULL,
+            market TEXT NOT NULL,
+            status TEXT NOT NULL DEFAULT 'pending',
+            playbook_json TEXT NOT NULL,
+            generated_at TEXT NOT NULL,
+            token_count INTEGER DEFAULT 0,
+            scenario_count INTEGER DEFAULT 0,
+            match_count INTEGER DEFAULT 0,
+            UNIQUE(date, market)
+        )
+        """
+    )
+
+    conn.execute("CREATE INDEX IF NOT EXISTS idx_playbooks_date ON playbooks(date)")
+    conn.execute("CREATE INDEX IF NOT EXISTS idx_playbooks_market ON playbooks(market)")
+
     # Create indices for efficient context queries
     conn.execute("CREATE INDEX IF NOT EXISTS idx_contexts_layer ON contexts(layer)")
     conn.execute("CREATE INDEX IF NOT EXISTS idx_contexts_timeframe ON contexts(timeframe)")
@@ -118,15 +142,33 @@ def log_trade(
     pnl: float = 0.0,
     market: str = "KR",
     exchange_code: str = "KRX",
+    selection_context: dict[str, any] | None = None,
 ) -> None:
-    """Insert a trade record into the database."""
+    """Insert a trade record into the database.
+
+    Args:
+        conn: Database connection
+        stock_code: Stock code
+        action: Trade action (BUY/SELL/HOLD)
+        confidence: Confidence level (0-100)
+        rationale: AI decision rationale
+        quantity: Number of shares
+        price: Trade price
+        pnl: Profit/loss
+        market: Market code
+        exchange_code: Exchange code
+        selection_context: Scanner selection data (RSI, volume_ratio, signal, score)
+    """
+    # Serialize selection context to JSON
+    context_json = json.dumps(selection_context) if selection_context else None
+
     conn.execute(
         """
         INSERT INTO trades (
             timestamp, stock_code, action, confidence, rationale,
-            quantity, price, pnl, market, exchange_code
+            quantity, price, pnl, market, exchange_code, selection_context
         )
-        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
         """,
         (
             datetime.now(UTC).isoformat(),
@@ -139,6 +181,7 @@ def log_trade(
             pnl,
             market,
             exchange_code,
+            context_json,
         ),
     )
     conn.commit()

src/notifications/README.md

@@ -0,0 +1,350 @@
+# Telegram Notifications
+
+Real-time trading event notifications via Telegram Bot API.
+
+## Setup
+
+### 1. Create a Telegram Bot
+
+1. Open Telegram and message [@BotFather](https://t.me/BotFather)
+2. Send `/newbot` command
+3. Follow prompts to name your bot
+4. Save the **bot token** (looks like `1234567890:ABCdefGHIjklMNOpqrsTUVwxyz`)
+
+### 2. Get Your Chat ID
+
+**Option A: Using @userinfobot**
+1. Message [@userinfobot](https://t.me/userinfobot) on Telegram
+2. Send `/start`
+3. Save your numeric **chat ID** (e.g., `123456789`)
+
+**Option B: Using @RawDataBot**
+1. Message [@RawDataBot](https://t.me/rawdatabot) on Telegram
+2. Look for `"id":` in the JSON response
+3. Save your numeric **chat ID**
+
+### 3. Configure Environment
+
+Add to your `.env` file:
+
+```bash
+TELEGRAM_BOT_TOKEN=1234567890:ABCdefGHIjklMNOpqrsTUVwxyz
+TELEGRAM_CHAT_ID=123456789
+TELEGRAM_ENABLED=true
+```
+
+### 4. Test the Bot
+
+Start a conversation with your bot on Telegram first (send `/start`), then run:
+
+```bash
+python -m src.main --mode=paper
+```
+
+You should receive a startup notification.
+
+## Message Examples
+
+### Trade Execution
+```
+🟢 BUY
+Symbol: AAPL (United States)
+Quantity: 10 shares
+Price: 150.25
+Confidence: 85%
+```
+
+### Circuit Breaker
+```
+🚨 CIRCUIT BREAKER TRIPPED
+P&L: -3.15% (threshold: -3.0%)
+Trading halted for safety
+```
+
+### Fat-Finger Protection
+```
+⚠️ Fat-Finger Protection
+Order rejected: TSLA
+Attempted: 45.0% of cash
+Max allowed: 30%
+Amount: 45,000 / 100,000
+```
+
+### Market Open/Close
+```
+ℹ️ Market Open
+Korea trading session started
+
+ℹ️ Market Close
+Korea trading session ended
+📈 P&L: +1.25%
+```
+
+### System Status
+```
+📝 System Started
+Mode: PAPER
+Markets: KRX, NASDAQ
+
+System Shutdown
+Normal shutdown
+```
+
+## Notification Priorities
+
+| Priority | Emoji | Use Case |
+|----------|-------|----------|
+| LOW | ℹ️ | Market open/close |
+| MEDIUM | 📊 | Trade execution, system start/stop |
+| HIGH | ⚠️ | Fat-finger protection, errors |
+| CRITICAL | 🚨 | Circuit breaker trips |
+
+## Rate Limiting
+
+- Default: 1 message per second
+- Prevents hitting Telegram's global rate limits
+- Configurable via `rate_limit` parameter
+
+## Troubleshooting
+
+### No notifications received
+
+1. **Check bot configuration**
+   ```bash
+   # Verify env variables are set
+   grep TELEGRAM .env
+   ```
+
+2. **Start conversation with bot**
+   - Open bot in Telegram
+   - Send `/start` command
+   - Bot cannot message users who haven't started a conversation
+
+3. **Check logs**
+   ```bash
+   # Look for Telegram-related errors
+   python -m src.main --mode=paper 2>&1 | grep -i telegram
+   ```
+
+4. **Verify bot token**
+   ```bash
+   curl https://api.telegram.org/bot<YOUR_TOKEN>/getMe
+   # Should return bot info (not 401 error)
+   ```
+
+5. **Verify chat ID**
+   ```bash
+   curl -X POST https://api.telegram.org/bot<YOUR_TOKEN>/sendMessage \
+     -H 'Content-Type: application/json' \
+     -d '{"chat_id": "<YOUR_CHAT_ID>", "text": "Test"}'
+   # Should send a test message
+   ```
+
+### Notifications delayed
+
+- Check rate limiter settings
+- Verify network connection
+- Look for timeout errors in logs
+
+### "Chat not found" error
+
+- Incorrect chat ID
+- Bot blocked by user
+- Need to send `/start` to bot first
+
+### "Unauthorized" error
+
+- Invalid bot token
+- Token revoked (regenerate with @BotFather)
+
+## Graceful Degradation
+
+The system works without Telegram notifications:
+
+- Missing credentials → notifications disabled automatically
+- API errors → logged but trading continues
+- Network timeouts → trading loop unaffected
+- Rate limiting → messages queued, trading proceeds
+
+**Notifications never crash the trading system.**
+
+## Security Notes
+
+- Never commit `.env` file with credentials
+- Bot token grants full bot control
+- Chat ID is not sensitive (just a number)
+- Messages are sent over HTTPS
+- No trading credentials in notifications
+
+## Advanced Usage
+
+### Group Notifications
+
+1. Add bot to Telegram group
+2. Get group chat ID (negative number like `-123456789`)
+3. Use group chat ID in `TELEGRAM_CHAT_ID`
+
+### Multiple Recipients
+
+Create multiple bots or use a broadcast group with multiple members.
+
+### Custom Rate Limits
+
+Not currently exposed in config, but can be modified in code:
+
+```python
+telegram = TelegramClient(
+    bot_token=settings.TELEGRAM_BOT_TOKEN,
+    chat_id=settings.TELEGRAM_CHAT_ID,
+    rate_limit=2.0,  # 2 messages per second
+)
+```
+
+## Bidirectional Commands
+
+Control your trading bot remotely via Telegram commands. The bot not only sends notifications but also accepts commands for real-time control.
+
+### Available Commands
+
+| Command | Description |
+|---------|-------------|
+| `/start` | Welcome message with quick start guide |
+| `/help` | List all available commands |
+| `/status` | Current trading status (mode, markets, P&L, circuit breaker) |
+| `/positions` | View current holdings grouped by market |
+| `/stop` | Pause all trading operations |
+| `/resume` | Resume trading operations |
+
+### Command Examples
+
+**Check Trading Status**
+```
+You: /status
+
+Bot:
+📊 Trading Status
+
+Mode: PAPER
+Markets: Korea, United States
+Trading: Active
+
+Current P&L: +2.50%
+Circuit Breaker: -3.0%
+```
+
+**View Holdings**
+```
+You: /positions
+
+Bot:
+💼 Current Holdings
+
+🇰🇷 Korea
+• 005930: 10 shares @ 70,000
+• 035420: 5 shares @ 200,000
+
+🇺🇸 Overseas
+• AAPL: 15 shares @ 175
+• TSLA: 8 shares @ 245
+
+Cash: ₩5,000,000
+```
+
+**Pause Trading**
+```
+You: /stop
+
+Bot:
+⏸️ Trading Paused
+
+All trading operations have been suspended.
+Use /resume to restart trading.
+```
+
+**Resume Trading**
+```
+You: /resume
+
+Bot:
+▶️ Trading Resumed
+
+Trading operations have been restarted.
+```
+
+### Security
+
+**Chat ID Verification**
+- Commands are only accepted from the configured `TELEGRAM_CHAT_ID`
+- Unauthorized users receive no response
+- Command attempts from wrong chat IDs are logged
+
+**Authorization Required**
+- Only the bot owner (chat ID in `.env`) can control trading
+- No way for unauthorized users to discover or use commands
+- All command executions are logged for audit
+
+### Configuration
+
+Add to your `.env` file:
+
+```bash
+# Commands are enabled by default
+TELEGRAM_COMMANDS_ENABLED=true
+
+# Polling interval (seconds) - how often to check for commands
+TELEGRAM_POLLING_INTERVAL=1.0
+```
+
+To disable commands but keep notifications:
+```bash
+TELEGRAM_COMMANDS_ENABLED=false
+```
+
+### How It Works
+
+1. **Long Polling**: Bot checks Telegram API every second for new messages
+2. **Command Parsing**: Messages starting with `/` are parsed as commands
+3. **Authentication**: Chat ID is verified before executing any command
+4. **Execution**: Command handler is called with current bot state
+5. **Response**: Result is sent back via Telegram
+
+### Error Handling
+
+- Command parsing errors → "Unknown command" response
+- API failures → Graceful degradation, error logged
+- Invalid state → Appropriate message (e.g., "Trading is already paused")
+- Trading loop isolation → Command errors never crash trading
+
+### Troubleshooting Commands
+
+**Commands not responding**
+1. Check `TELEGRAM_COMMANDS_ENABLED=true` in `.env`
+2. Verify you started conversation with `/start`
+3. Check logs for command handler errors
+4. Confirm chat ID matches `.env` configuration
+
+**Wrong chat ID**
+- Commands from unauthorized chats are silently ignored
+- Check logs for "unauthorized chat_id" warnings
+
+**Delayed responses**
+- Polling interval is 1 second by default
+- Network latency may add delay
+- Check `TELEGRAM_POLLING_INTERVAL` setting
+
+## API Reference
+
+See `telegram_client.py` for full API documentation.
+
+### Notification Methods
+- `notify_trade_execution()` - Trade alerts
+- `notify_circuit_breaker()` - Emergency stops
+- `notify_fat_finger()` - Order rejections
+- `notify_market_open/close()` - Session tracking
+- `notify_system_start/shutdown()` - Lifecycle events
+- `notify_error()` - Error alerts
+
+### Command Handler
+- `TelegramCommandHandler` - Bidirectional command processing
+- `register_command()` - Register custom command handlers
+- `start_polling()` / `stop_polling()` - Lifecycle management

src/notifications/__init__.py

@@ -0,0 +1,5 @@
+"""Real-time notifications for trading events."""
+
+from src.notifications.telegram_client import TelegramClient
+
+__all__ = ["TelegramClient"]

src/notifications/telegram_client.py

@@ -0,0 +1,582 @@
+"""Telegram notification client for real-time trading alerts."""
+
+import asyncio
+import logging
+import time
+from collections.abc import Awaitable, Callable
+from dataclasses import dataclass
+from enum import Enum
+
+import aiohttp
+
+logger = logging.getLogger(__name__)
+
+
+class NotificationPriority(Enum):
+    """Priority levels for notifications with emoji indicators."""
+
+    LOW = ("ℹ️", "info")
+    MEDIUM = ("📊", "medium")
+    HIGH = ("⚠️", "warning")
+    CRITICAL = ("🚨", "critical")
+
+    def __init__(self, emoji: str, label: str) -> None:
+        self.emoji = emoji
+        self.label = label
+
+
+class LeakyBucket:
+    """Rate limiter using leaky bucket algorithm."""
+
+    def __init__(self, rate: float, capacity: int = 1) -> None:
+        """
+        Initialize rate limiter.
+
+        Args:
+            rate: Maximum requests per second
+            capacity: Bucket capacity (burst size)
+        """
+        self._rate = rate
+        self._capacity = capacity
+        self._tokens = float(capacity)
+        self._last_update = time.monotonic()
+        self._lock = asyncio.Lock()
+
+    async def acquire(self) -> None:
+        """Wait until a token is available, then consume it."""
+        async with self._lock:
+            now = time.monotonic()
+            elapsed = now - self._last_update
+            self._tokens = min(self._capacity, self._tokens + elapsed * self._rate)
+            self._last_update = now
+
+            if self._tokens < 1.0:
+                wait_time = (1.0 - self._tokens) / self._rate
+                await asyncio.sleep(wait_time)
+                self._tokens = 0.0
+            else:
+                self._tokens -= 1.0
+
+
+@dataclass
+class NotificationMessage:
+    """Internal notification message structure."""
+
+    priority: NotificationPriority
+    message: str
+
+
+class TelegramClient:
+    """Telegram Bot API client for sending trading notifications."""
+
+    API_BASE = "https://api.telegram.org/bot{token}"
+    DEFAULT_TIMEOUT = 5.0  # seconds
+    DEFAULT_RATE = 1.0  # messages per second
+
+    def __init__(
+        self,
+        bot_token: str | None = None,
+        chat_id: str | None = None,
+        enabled: bool = True,
+        rate_limit: float = DEFAULT_RATE,
+    ) -> None:
+        """
+        Initialize Telegram client.
+
+        Args:
+            bot_token: Telegram bot token from @BotFather
+            chat_id: Target chat ID (user or group)
+            enabled: Enable/disable notifications globally
+            rate_limit: Maximum messages per second
+        """
+        self._bot_token = bot_token
+        self._chat_id = chat_id
+        self._enabled = enabled
+        self._rate_limiter = LeakyBucket(rate=rate_limit)
+        self._session: aiohttp.ClientSession | None = None
+
+        if not enabled:
+            logger.info("Telegram notifications disabled via configuration")
+        elif bot_token is None or chat_id is None:
+            logger.warning(
+                "Telegram notifications disabled (missing bot_token or chat_id)"
+            )
+            self._enabled = False
+        else:
+            logger.info("Telegram notifications enabled for chat_id=%s", chat_id)
+
+    def _get_session(self) -> aiohttp.ClientSession:
+        """Get or create aiohttp session."""
+        if self._session is None or self._session.closed:
+            self._session = aiohttp.ClientSession(
+                timeout=aiohttp.ClientTimeout(total=self.DEFAULT_TIMEOUT)
+            )
+        return self._session
+
+    async def close(self) -> None:
+        """Close HTTP session."""
+        if self._session is not None and not self._session.closed:
+            await self._session.close()
+
+    async def send_message(self, text: str, parse_mode: str = "HTML") -> bool:
+        """
+        Send a generic text message to Telegram.
+
+        Args:
+            text: Message text to send
+            parse_mode: Parse mode for formatting (HTML or Markdown)
+
+        Returns:
+            True if message was sent successfully, False otherwise
+        """
+        if not self._enabled:
+            return False
+
+        try:
+            await self._rate_limiter.acquire()
+
+            url = f"{self.API_BASE.format(token=self._bot_token)}/sendMessage"
+            payload = {
+                "chat_id": self._chat_id,
+                "text": text,
+                "parse_mode": parse_mode,
+            }
+
+            session = self._get_session()
+            async with session.post(url, json=payload) as resp:
+                if resp.status != 200:
+                    error_text = await resp.text()
+                    logger.error(
+                        "Telegram API error (status=%d): %s", resp.status, error_text
+                    )
+                    return False
+                logger.debug("Telegram message sent: %s", text[:50])
+                return True
+
+        except asyncio.TimeoutError:
+            logger.error("Telegram message timeout")
+            return False
+        except aiohttp.ClientError as exc:
+            logger.error("Telegram message failed: %s", exc)
+            return False
+        except Exception as exc:
+            logger.error("Unexpected error sending message: %s", exc)
+            return False
+
+    async def _send_notification(self, msg: NotificationMessage) -> None:
+        """
+        Send notification to Telegram with graceful degradation.
+
+        Args:
+            msg: Notification message to send
+        """
+        formatted_message = f"{msg.priority.emoji} {msg.message}"
+        await self.send_message(formatted_message)
+
+    async def notify_trade_execution(
+        self,
+        stock_code: str,
+        market: str,
+        action: str,
+        quantity: int,
+        price: float,
+        confidence: float,
+    ) -> None:
+        """
+        Notify trade execution.
+
+        Args:
+            stock_code: Stock ticker symbol
+            market: Market name (e.g., "Korea", "United States")
+            action: "BUY" or "SELL"
+            quantity: Number of shares
+            price: Execution price
+            confidence: AI confidence level (0-100)
+        """
+        emoji = "🟢" if action == "BUY" else "🔴"
+        message = (
+            f"<b>{emoji} {action}</b>\n"
+            f"Symbol: <code>{stock_code}</code> ({market})\n"
+            f"Quantity: {quantity:,} shares\n"
+            f"Price: {price:,.2f}\n"
+            f"Confidence: {confidence:.0f}%"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.MEDIUM, message=message)
+        )
+
+    async def notify_market_open(self, market_name: str) -> None:
+        """
+        Notify market opening.
+
+        Args:
+            market_name: Name of the market (e.g., "Korea", "United States")
+        """
+        message = f"<b>Market Open</b>\n{market_name} trading session started"
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.LOW, message=message)
+        )
+
+    async def notify_market_close(self, market_name: str, pnl_pct: float) -> None:
+        """
+        Notify market closing.
+
+        Args:
+            market_name: Name of the market
+            pnl_pct: Final P&L percentage for the session
+        """
+        pnl_sign = "+" if pnl_pct >= 0 else ""
+        pnl_emoji = "📈" if pnl_pct >= 0 else "📉"
+        message = (
+            f"<b>Market Close</b>\n"
+            f"{market_name} trading session ended\n"
+            f"{pnl_emoji} P&L: {pnl_sign}{pnl_pct:.2f}%"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.LOW, message=message)
+        )
+
+    async def notify_circuit_breaker(
+        self, pnl_pct: float, threshold: float
+    ) -> None:
+        """
+        Notify circuit breaker activation.
+
+        Args:
+            pnl_pct: Current P&L percentage
+            threshold: Circuit breaker threshold
+        """
+        message = (
+            f"<b>CIRCUIT BREAKER TRIPPED</b>\n"
+            f"P&L: {pnl_pct:.2f}% (threshold: {threshold:.1f}%)\n"
+            f"Trading halted for safety"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.CRITICAL, message=message)
+        )
+
+    async def notify_fat_finger(
+        self,
+        stock_code: str,
+        order_amount: float,
+        total_cash: float,
+        max_pct: float,
+    ) -> None:
+        """
+        Notify fat-finger protection rejection.
+
+        Args:
+            stock_code: Stock ticker symbol
+            order_amount: Attempted order amount
+            total_cash: Total available cash
+            max_pct: Maximum allowed percentage
+        """
+        attempted_pct = (order_amount / total_cash) * 100 if total_cash > 0 else 0
+        message = (
+            f"<b>Fat-Finger Protection</b>\n"
+            f"Order rejected: <code>{stock_code}</code>\n"
+            f"Attempted: {attempted_pct:.1f}% of cash\n"
+            f"Max allowed: {max_pct:.0f}%\n"
+            f"Amount: {order_amount:,.0f} / {total_cash:,.0f}"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.HIGH, message=message)
+        )
+
+    async def notify_system_start(
+        self, mode: str, enabled_markets: list[str]
+    ) -> None:
+        """
+        Notify system startup.
+
+        Args:
+            mode: Trading mode ("paper" or "live")
+            enabled_markets: List of enabled market codes
+        """
+        mode_emoji = "📝" if mode == "paper" else "💰"
+        markets_str = ", ".join(enabled_markets)
+        message = (
+            f"<b>{mode_emoji} System Started</b>\n"
+            f"Mode: {mode.upper()}\n"
+            f"Markets: {markets_str}"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.MEDIUM, message=message)
+        )
+
+    async def notify_playbook_generated(
+        self,
+        market: str,
+        stock_count: int,
+        scenario_count: int,
+        token_count: int,
+    ) -> None:
+        """
+        Notify that a daily playbook was generated.
+
+        Args:
+            market: Market code (e.g., "KR", "US")
+            stock_count: Number of stocks in the playbook
+            scenario_count: Total number of scenarios
+            token_count: Gemini token usage for the playbook
+        """
+        message = (
+            f"<b>Playbook Generated</b>\n"
+            f"Market: {market}\n"
+            f"Stocks: {stock_count}\n"
+            f"Scenarios: {scenario_count}\n"
+            f"Tokens: {token_count}"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.MEDIUM, message=message)
+        )
+
+    async def notify_scenario_matched(
+        self,
+        stock_code: str,
+        action: str,
+        condition_summary: str,
+        confidence: float,
+    ) -> None:
+        """
+        Notify that a scenario matched for a stock.
+
+        Args:
+            stock_code: Stock ticker symbol
+            action: Scenario action (BUY/SELL/HOLD/REDUCE_ALL)
+            condition_summary: Short summary of the matched condition
+            confidence: Scenario confidence (0-100)
+        """
+        message = (
+            f"<b>Scenario Matched</b>\n"
+            f"Symbol: <code>{stock_code}</code>\n"
+            f"Action: {action}\n"
+            f"Condition: {condition_summary}\n"
+            f"Confidence: {confidence:.0f}%"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.HIGH, message=message)
+        )
+
+    async def notify_playbook_failed(self, market: str, reason: str) -> None:
+        """
+        Notify that playbook generation failed.
+
+        Args:
+            market: Market code (e.g., "KR", "US")
+            reason: Failure reason summary
+        """
+        message = (
+            f"<b>Playbook Failed</b>\n"
+            f"Market: {market}\n"
+            f"Reason: {reason[:200]}"
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.HIGH, message=message)
+        )
+
+    async def notify_system_shutdown(self, reason: str) -> None:
+        """
+        Notify system shutdown.
+
+        Args:
+            reason: Reason for shutdown (e.g., "Normal shutdown", "Circuit breaker")
+        """
+        message = f"<b>System Shutdown</b>\n{reason}"
+        priority = (
+            NotificationPriority.CRITICAL
+            if "circuit breaker" in reason.lower()
+            else NotificationPriority.MEDIUM
+        )
+        await self._send_notification(
+            NotificationMessage(priority=priority, message=message)
+        )
+
+    async def notify_error(
+        self, error_type: str, error_msg: str, context: str
+    ) -> None:
+        """
+        Notify system error.
+
+        Args:
+            error_type: Type of error (e.g., "Connection Error")
+            error_msg: Error message
+            context: Error context (e.g., stock code, market)
+        """
+        message = (
+            f"<b>Error: {error_type}</b>\n"
+            f"Context: {context}\n"
+            f"Message: {error_msg[:200]}"  # Truncate long errors
+        )
+        await self._send_notification(
+            NotificationMessage(priority=NotificationPriority.HIGH, message=message)
+        )
+
+
+class TelegramCommandHandler:
+    """Handles incoming Telegram commands via long polling."""
+
+    def __init__(
+        self, client: TelegramClient, polling_interval: float = 1.0
+    ) -> None:
+        """
+        Initialize command handler.
+
+        Args:
+            client: TelegramClient instance for sending responses
+            polling_interval: Polling interval in seconds
+        """
+        self._client = client
+        self._polling_interval = polling_interval
+        self._commands: dict[str, Callable[[], Awaitable[None]]] = {}
+        self._last_update_id = 0
+        self._polling_task: asyncio.Task[None] | None = None
+        self._running = False
+
+    def register_command(
+        self, command: str, handler: Callable[[], Awaitable[None]]
+    ) -> None:
+        """
+        Register a command handler.
+
+        Args:
+            command: Command name (without leading slash, e.g., "start")
+            handler: Async function to handle the command
+        """
+        self._commands[command] = handler
+        logger.debug("Registered command handler: /%s", command)
+
+    async def start_polling(self) -> None:
+        """Start long polling for commands."""
+        if self._running:
+            logger.warning("Command handler already running")
+            return
+
+        if not self._client._enabled:
+            logger.info("Command handler disabled (TelegramClient disabled)")
+            return
+
+        self._running = True
+        self._polling_task = asyncio.create_task(self._poll_loop())
+        logger.info("Started Telegram command polling")
+
+    async def stop_polling(self) -> None:
+        """Stop polling and cancel pending tasks."""
+        if not self._running:
+            return
+
+        self._running = False
+        if self._polling_task:
+            self._polling_task.cancel()
+            try:
+                await self._polling_task
+            except asyncio.CancelledError:
+                pass
+        logger.info("Stopped Telegram command polling")
+
+    async def _poll_loop(self) -> None:
+        """Main polling loop that fetches updates."""
+        while self._running:
+            try:
+                updates = await self._get_updates()
+                for update in updates:
+                    await self._handle_update(update)
+            except asyncio.CancelledError:
+                break
+            except Exception as exc:
+                logger.error("Error in polling loop: %s", exc)
+
+            await asyncio.sleep(self._polling_interval)
+
+    async def _get_updates(self) -> list[dict]:
+        """
+        Fetch updates from Telegram API.
+
+        Returns:
+            List of update objects
+        """
+        try:
+            url = f"{self._client.API_BASE.format(token=self._client._bot_token)}/getUpdates"
+            payload = {
+                "offset": self._last_update_id + 1,
+                "timeout": int(self._polling_interval),
+                "allowed_updates": ["message"],
+            }
+
+            session = self._client._get_session()
+            async with session.post(url, json=payload) as resp:
+                if resp.status != 200:
+                    error_text = await resp.text()
+                    logger.error(
+                        "getUpdates API error (status=%d): %s", resp.status, error_text
+                    )
+                    return []
+
+                data = await resp.json()
+                if not data.get("ok"):
+                    logger.error("getUpdates returned ok=false: %s", data)
+                    return []
+
+                updates = data.get("result", [])
+                if updates:
+                    self._last_update_id = updates[-1]["update_id"]
+
+                return updates
+
+        except asyncio.TimeoutError:
+            logger.debug("getUpdates timeout (normal)")
+            return []
+        except aiohttp.ClientError as exc:
+            logger.error("getUpdates failed: %s", exc)
+            return []
+        except Exception as exc:
+            logger.error("Unexpected error in _get_updates: %s", exc)
+            return []
+
+    async def _handle_update(self, update: dict) -> None:
+        """
+        Parse and handle a single update.
+
+        Args:
+            update: Update object from Telegram API
+        """
+        try:
+            message = update.get("message")
+            if not message:
+                return
+
+            # Verify chat_id matches configured chat
+            chat_id = str(message.get("chat", {}).get("id", ""))
+            if chat_id != self._client._chat_id:
+                logger.warning(
+                    "Ignoring command from unauthorized chat_id: %s", chat_id
+                )
+                return
+
+            # Extract command text
+            text = message.get("text", "").strip()
+            if not text.startswith("/"):
+                return
+
+            # Parse command (remove leading slash and extract command name)
+            command_parts = text[1:].split()
+            if not command_parts:
+                return
+
+            # Remove @botname suffix if present (for group chats)
+            command_name = command_parts[0].split("@")[0]
+
+            # Execute handler
+            handler = self._commands.get(command_name)
+            if handler:
+                logger.info("Executing command: /%s", command_name)
+                await handler()
+            else:
+                logger.debug("Unknown command: /%s", command_name)
+                await self._client.send_message(
+                    f"Unknown command: /{command_name}\nUse /help to see available commands."
+                )
+
+        except Exception as exc:
+            logger.error("Error handling update: %s", exc)
+            # Don't crash the polling loop on handler errors

src/strategy/models.py

@@ -0,0 +1,164 @@
+"""Pydantic models for pre-market scenario planning.
+
+Defines the data contracts for the proactive strategy system:
+- AI generates DayPlaybook before market open (structured JSON scenarios)
+- Local ScenarioEngine matches conditions during market hours (no API calls)
+"""
+
+from __future__ import annotations
+
+from datetime import UTC, date, datetime
+from enum import Enum
+
+from pydantic import BaseModel, Field, field_validator
+
+
+class ScenarioAction(str, Enum):
+    """Actions that can be taken by scenarios."""
+
+    BUY = "BUY"
+    SELL = "SELL"
+    HOLD = "HOLD"
+    REDUCE_ALL = "REDUCE_ALL"
+
+
+class MarketOutlook(str, Enum):
+    """AI's assessment of market direction."""
+
+    BULLISH = "bullish"
+    NEUTRAL_TO_BULLISH = "neutral_to_bullish"
+    NEUTRAL = "neutral"
+    NEUTRAL_TO_BEARISH = "neutral_to_bearish"
+    BEARISH = "bearish"
+
+
+class PlaybookStatus(str, Enum):
+    """Lifecycle status of a playbook."""
+
+    PENDING = "pending"
+    READY = "ready"
+    FAILED = "failed"
+    EXPIRED = "expired"
+
+
+class StockCondition(BaseModel):
+    """Condition fields for scenario matching (all optional, AND-combined).
+
+    The ScenarioEngine evaluates all non-None fields as AND conditions.
+    A condition matches only if ALL specified fields are satisfied.
+    """
+
+    rsi_below: float | None = None
+    rsi_above: float | None = None
+    volume_ratio_above: float | None = None
+    volume_ratio_below: float | None = None
+    price_above: float | None = None
+    price_below: float | None = None
+    price_change_pct_above: float | None = None
+    price_change_pct_below: float | None = None
+
+    def has_any_condition(self) -> bool:
+        """Check if at least one condition field is set."""
+        return any(
+            v is not None
+            for v in (
+                self.rsi_below,
+                self.rsi_above,
+                self.volume_ratio_above,
+                self.volume_ratio_below,
+                self.price_above,
+                self.price_below,
+                self.price_change_pct_above,
+                self.price_change_pct_below,
+            )
+        )
+
+
+class StockScenario(BaseModel):
+    """A single condition-action rule for one stock."""
+
+    condition: StockCondition
+    action: ScenarioAction
+    confidence: int = Field(ge=0, le=100)
+    allocation_pct: float = Field(ge=0, le=100, default=10.0)
+    stop_loss_pct: float = Field(le=0, default=-2.0)
+    take_profit_pct: float = Field(ge=0, default=3.0)
+    rationale: str = ""
+
+
+class StockPlaybook(BaseModel):
+    """All scenarios for a single stock (ordered by priority)."""
+
+    stock_code: str
+    stock_name: str = ""
+    scenarios: list[StockScenario] = Field(min_length=1)
+
+
+class GlobalRule(BaseModel):
+    """Portfolio-level rule (checked before stock-level scenarios)."""
+
+    condition: str  # e.g. "portfolio_pnl_pct < -2.0"
+    action: ScenarioAction
+    rationale: str = ""
+
+
+class CrossMarketContext(BaseModel):
+    """Summary of another market's state for cross-market awareness."""
+
+    market: str  # e.g. "US" or "KR"
+    date: str
+    total_pnl: float = 0.0
+    win_rate: float = 0.0
+    index_change_pct: float = 0.0  # e.g. KOSPI or S&P500 change
+    key_events: list[str] = Field(default_factory=list)
+    lessons: list[str] = Field(default_factory=list)
+
+
+class DayPlaybook(BaseModel):
+    """Complete playbook for a single trading day in a single market.
+
+    Generated by PreMarketPlanner (1 Gemini call per market per day).
+    Consumed by ScenarioEngine during market hours (0 API calls).
+    """
+
+    date: date
+    market: str  # "KR" or "US"
+    market_outlook: MarketOutlook = MarketOutlook.NEUTRAL
+    generated_at: str = ""  # ISO timestamp
+    gemini_model: str = ""
+    token_count: int = 0
+    global_rules: list[GlobalRule] = Field(default_factory=list)
+    stock_playbooks: list[StockPlaybook] = Field(default_factory=list)
+    default_action: ScenarioAction = ScenarioAction.HOLD
+    context_summary: dict = Field(default_factory=dict)
+    cross_market: CrossMarketContext | None = None
+
+    @field_validator("stock_playbooks")
+    @classmethod
+    def validate_unique_stocks(cls, v: list[StockPlaybook]) -> list[StockPlaybook]:
+        codes = [pb.stock_code for pb in v]
+        if len(codes) != len(set(codes)):
+            raise ValueError("Duplicate stock codes in playbook")
+        return v
+
+    def get_stock_playbook(self, stock_code: str) -> StockPlaybook | None:
+        """Find the playbook for a specific stock."""
+        for pb in self.stock_playbooks:
+            if pb.stock_code == stock_code:
+                return pb
+        return None
+
+    @property
+    def scenario_count(self) -> int:
+        """Total number of scenarios across all stocks."""
+        return sum(len(pb.scenarios) for pb in self.stock_playbooks)
+
+    @property
+    def stock_count(self) -> int:
+        """Number of stocks with scenarios."""
+        return len(self.stock_playbooks)
+
+    def model_post_init(self, __context: object) -> None:
+        """Set generated_at if not provided."""
+        if not self.generated_at:
+            self.generated_at = datetime.now(UTC).isoformat()

src/strategy/playbook_store.py

@@ -0,0 +1,184 @@
+"""Playbook persistence layer — CRUD for DayPlaybook in SQLite.
+
+Stores and retrieves market-specific daily playbooks with JSON serialization.
+Designed for the pre-market strategy system (one playbook per market per day).
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import sqlite3
+from datetime import date
+
+from src.strategy.models import DayPlaybook, PlaybookStatus
+
+logger = logging.getLogger(__name__)
+
+
+class PlaybookStore:
+    """CRUD operations for DayPlaybook persistence."""
+
+    def __init__(self, conn: sqlite3.Connection) -> None:
+        self._conn = conn
+
+    def save(self, playbook: DayPlaybook) -> int:
+        """Save or replace a playbook for a given date+market.
+
+        Uses INSERT OR REPLACE to enforce UNIQUE(date, market).
+
+        Returns:
+            The row id of the inserted/replaced record.
+        """
+        playbook_json = playbook.model_dump_json()
+        cursor = self._conn.execute(
+            """
+            INSERT OR REPLACE INTO playbooks
+                (date, market, status, playbook_json, generated_at,
+                 token_count, scenario_count, match_count)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """,
+            (
+                playbook.date.isoformat(),
+                playbook.market,
+                PlaybookStatus.READY.value,
+                playbook_json,
+                playbook.generated_at,
+                playbook.token_count,
+                playbook.scenario_count,
+                0,
+            ),
+        )
+        self._conn.commit()
+        row_id = cursor.lastrowid or 0
+        logger.info(
+            "Saved playbook for %s/%s (%d stocks, %d scenarios)",
+            playbook.date, playbook.market,
+            playbook.stock_count, playbook.scenario_count,
+        )
+        return row_id
+
+    def load(self, target_date: date, market: str) -> DayPlaybook | None:
+        """Load a playbook for a specific date and market.
+
+        Returns:
+            DayPlaybook if found, None otherwise.
+        """
+        row = self._conn.execute(
+            "SELECT playbook_json FROM playbooks WHERE date = ? AND market = ?",
+            (target_date.isoformat(), market),
+        ).fetchone()
+        if row is None:
+            return None
+        return DayPlaybook.model_validate_json(row[0])
+
+    def get_status(self, target_date: date, market: str) -> PlaybookStatus | None:
+        """Get the status of a playbook without deserializing the full JSON."""
+        row = self._conn.execute(
+            "SELECT status FROM playbooks WHERE date = ? AND market = ?",
+            (target_date.isoformat(), market),
+        ).fetchone()
+        if row is None:
+            return None
+        return PlaybookStatus(row[0])
+
+    def update_status(self, target_date: date, market: str, status: PlaybookStatus) -> bool:
+        """Update the status of a playbook.
+
+        Returns:
+            True if a row was updated, False if not found.
+        """
+        cursor = self._conn.execute(
+            "UPDATE playbooks SET status = ? WHERE date = ? AND market = ?",
+            (status.value, target_date.isoformat(), market),
+        )
+        self._conn.commit()
+        return cursor.rowcount > 0
+
+    def increment_match_count(self, target_date: date, market: str) -> bool:
+        """Increment the match_count for tracking scenario hits during the day.
+
+        Returns:
+            True if a row was updated, False if not found.
+        """
+        cursor = self._conn.execute(
+            "UPDATE playbooks SET match_count = match_count + 1 WHERE date = ? AND market = ?",
+            (target_date.isoformat(), market),
+        )
+        self._conn.commit()
+        return cursor.rowcount > 0
+
+    def get_stats(self, target_date: date, market: str) -> dict | None:
+        """Get playbook stats without full deserialization.
+
+        Returns:
+            Dict with status, token_count, scenario_count, match_count, or None.
+        """
+        row = self._conn.execute(
+            """
+            SELECT status, token_count, scenario_count, match_count, generated_at
+            FROM playbooks WHERE date = ? AND market = ?
+            """,
+            (target_date.isoformat(), market),
+        ).fetchone()
+        if row is None:
+            return None
+        return {
+            "status": row[0],
+            "token_count": row[1],
+            "scenario_count": row[2],
+            "match_count": row[3],
+            "generated_at": row[4],
+        }
+
+    def list_recent(self, market: str | None = None, limit: int = 7) -> list[dict]:
+        """List recent playbooks with summary info.
+
+        Args:
+            market: Filter by market code. None for all markets.
+            limit: Max number of results.
+
+        Returns:
+            List of dicts with date, market, status, scenario_count, match_count.
+        """
+        if market is not None:
+            rows = self._conn.execute(
+                """
+                SELECT date, market, status, scenario_count, match_count
+                FROM playbooks WHERE market = ?
+                ORDER BY date DESC LIMIT ?
+                """,
+                (market, limit),
+            ).fetchall()
+        else:
+            rows = self._conn.execute(
+                """
+                SELECT date, market, status, scenario_count, match_count
+                FROM playbooks
+                ORDER BY date DESC LIMIT ?
+                """,
+                (limit,),
+            ).fetchall()
+        return [
+            {
+                "date": row[0],
+                "market": row[1],
+                "status": row[2],
+                "scenario_count": row[3],
+                "match_count": row[4],
+            }
+            for row in rows
+        ]
+
+    def delete(self, target_date: date, market: str) -> bool:
+        """Delete a playbook.
+
+        Returns:
+            True if a row was deleted, False if not found.
+        """
+        cursor = self._conn.execute(
+            "DELETE FROM playbooks WHERE date = ? AND market = ?",
+            (target_date.isoformat(), market),
+        )
+        self._conn.commit()
+        return cursor.rowcount > 0

src/strategy/pre_market_planner.py

@@ -0,0 +1,419 @@
+"""Pre-market planner — generates DayPlaybook via Gemini before market open.
+
+One Gemini API call per market per day. Candidates come from SmartVolatilityScanner.
+On failure, returns a defensive playbook (all HOLD, no trades).
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+from datetime import date
+from typing import Any
+
+from src.analysis.smart_scanner import ScanCandidate
+from src.brain.context_selector import ContextSelector, DecisionType
+from src.brain.gemini_client import GeminiClient
+from src.config import Settings
+from src.context.store import ContextLayer, ContextStore
+from src.strategy.models import (
+    CrossMarketContext,
+    DayPlaybook,
+    GlobalRule,
+    MarketOutlook,
+    ScenarioAction,
+    StockCondition,
+    StockPlaybook,
+    StockScenario,
+)
+
+logger = logging.getLogger(__name__)
+
+# Mapping from string to MarketOutlook enum
+_OUTLOOK_MAP: dict[str, MarketOutlook] = {
+    "bullish": MarketOutlook.BULLISH,
+    "neutral_to_bullish": MarketOutlook.NEUTRAL_TO_BULLISH,
+    "neutral": MarketOutlook.NEUTRAL,
+    "neutral_to_bearish": MarketOutlook.NEUTRAL_TO_BEARISH,
+    "bearish": MarketOutlook.BEARISH,
+}
+
+_ACTION_MAP: dict[str, ScenarioAction] = {
+    "BUY": ScenarioAction.BUY,
+    "SELL": ScenarioAction.SELL,
+    "HOLD": ScenarioAction.HOLD,
+    "REDUCE_ALL": ScenarioAction.REDUCE_ALL,
+}
+
+
+class PreMarketPlanner:
+    """Generates a DayPlaybook by calling Gemini once before market open.
+
+    Flow:
+    1. Collect strategic context (L5-L7) + cross-market context
+    2. Build a structured prompt with scan candidates
+    3. Call Gemini for JSON scenario generation
+    4. Parse and validate response into DayPlaybook
+    5. On failure → defensive playbook (HOLD everything)
+    """
+
+    def __init__(
+        self,
+        gemini_client: GeminiClient,
+        context_store: ContextStore,
+        context_selector: ContextSelector,
+        settings: Settings,
+    ) -> None:
+        self._gemini = gemini_client
+        self._context_store = context_store
+        self._context_selector = context_selector
+        self._settings = settings
+
+    async def generate_playbook(
+        self,
+        market: str,
+        candidates: list[ScanCandidate],
+        today: date | None = None,
+    ) -> DayPlaybook:
+        """Generate a DayPlaybook for a market using Gemini.
+
+        Args:
+            market: Market code ("KR" or "US")
+            candidates: Stock candidates from SmartVolatilityScanner
+            today: Override date (defaults to date.today()). Use market-local date.
+
+        Returns:
+            DayPlaybook with scenarios. Empty/defensive if no candidates or failure.
+        """
+        if today is None:
+            today = date.today()
+
+        if not candidates:
+            logger.info("No candidates for %s — returning empty playbook", market)
+            return self._empty_playbook(today, market)
+
+        try:
+            # 1. Gather context
+            context_data = self._gather_context()
+            cross_market = self.build_cross_market_context(market, today)
+
+            # 2. Build prompt
+            prompt = self._build_prompt(market, candidates, context_data, cross_market)
+
+            # 3. Call Gemini
+            market_data = {
+                "stock_code": "PLANNER",
+                "current_price": 0,
+                "prompt_override": prompt,
+            }
+            decision = await self._gemini.decide(market_data)
+
+            # 4. Parse response
+            playbook = self._parse_response(
+                decision.rationale, today, market, candidates, cross_market
+            )
+            playbook_with_tokens = playbook.model_copy(
+                update={"token_count": decision.token_count}
+            )
+            logger.info(
+                "Generated playbook for %s: %d stocks, %d scenarios, %d tokens",
+                market,
+                playbook_with_tokens.stock_count,
+                playbook_with_tokens.scenario_count,
+                playbook_with_tokens.token_count,
+            )
+            return playbook_with_tokens
+
+        except Exception:
+            logger.exception("Playbook generation failed for %s", market)
+            if self._settings.DEFENSIVE_PLAYBOOK_ON_FAILURE:
+                return self._defensive_playbook(today, market, candidates)
+            return self._empty_playbook(today, market)
+
+    def build_cross_market_context(
+        self, target_market: str, today: date | None = None,
+    ) -> CrossMarketContext | None:
+        """Build cross-market context from the other market's L6 data.
+
+        KR planner → reads US scorecard from previous night.
+        US planner → reads KR scorecard from today.
+
+        Args:
+            target_market: The market being planned ("KR" or "US")
+            today: Override date (defaults to date.today()). Use market-local date.
+        """
+        other_market = "US" if target_market == "KR" else "KR"
+        if today is None:
+            today = date.today()
+        timeframe = today.isoformat()
+
+        scorecard_key = f"scorecard_{other_market}"
+        scorecard_data = self._context_store.get_context(
+            ContextLayer.L6_DAILY, timeframe, scorecard_key
+        )
+
+        if scorecard_data is None:
+            logger.debug("No cross-market scorecard found for %s", other_market)
+            return None
+
+        if isinstance(scorecard_data, str):
+            try:
+                scorecard_data = json.loads(scorecard_data)
+            except (json.JSONDecodeError, TypeError):
+                return None
+
+        if not isinstance(scorecard_data, dict):
+            return None
+
+        return CrossMarketContext(
+            market=other_market,
+            date=timeframe,
+            total_pnl=float(scorecard_data.get("total_pnl", 0.0)),
+            win_rate=float(scorecard_data.get("win_rate", 0.0)),
+            index_change_pct=float(scorecard_data.get("index_change_pct", 0.0)),
+            key_events=scorecard_data.get("key_events", []),
+            lessons=scorecard_data.get("lessons", []),
+        )
+
+    def _gather_context(self) -> dict[str, Any]:
+        """Gather strategic context using ContextSelector."""
+        layers = self._context_selector.select_layers(
+            decision_type=DecisionType.STRATEGIC,
+            include_realtime=True,
+        )
+        return self._context_selector.get_context_data(layers, max_items_per_layer=10)
+
+    def _build_prompt(
+        self,
+        market: str,
+        candidates: list[ScanCandidate],
+        context_data: dict[str, Any],
+        cross_market: CrossMarketContext | None,
+    ) -> str:
+        """Build a structured prompt for Gemini to generate scenario JSON."""
+        max_scenarios = self._settings.MAX_SCENARIOS_PER_STOCK
+
+        candidates_text = "\n".join(
+            f"  - {c.stock_code} ({c.name}): price={c.price}, "
+            f"RSI={c.rsi:.1f}, volume_ratio={c.volume_ratio:.1f}, "
+            f"signal={c.signal}, score={c.score:.1f}"
+            for c in candidates
+        )
+
+        cross_market_text = ""
+        if cross_market:
+            cross_market_text = (
+                f"\n## Other Market ({cross_market.market}) Summary\n"
+                f"- P&L: {cross_market.total_pnl:+.2f}%\n"
+                f"- Win Rate: {cross_market.win_rate:.0f}%\n"
+                f"- Index Change: {cross_market.index_change_pct:+.2f}%\n"
+            )
+            if cross_market.lessons:
+                cross_market_text += f"- Lessons: {'; '.join(cross_market.lessons[:3])}\n"
+
+        context_text = ""
+        if context_data:
+            context_text = "\n## Strategic Context\n"
+            for layer_name, layer_data in context_data.items():
+                if layer_data:
+                    context_text += f"### {layer_name}\n"
+                    for key, value in list(layer_data.items())[:5]:
+                        context_text += f"  - {key}: {value}\n"
+
+        return (
+            f"You are a pre-market trading strategist for the {market} market.\n"
+            f"Generate structured trading scenarios for today.\n\n"
+            f"## Candidates (from volatility scanner)\n{candidates_text}\n"
+            f"{cross_market_text}"
+            f"{context_text}\n"
+            f"## Instructions\n"
+            f"Return a JSON object with this exact structure:\n"
+            f'{{\n'
+            f'  "market_outlook": "bullish|neutral_to_bullish|neutral'
+            f'|neutral_to_bearish|bearish",\n'
+            f'  "global_rules": [\n'
+            f'    {{"condition": "portfolio_pnl_pct < -2.0",'
+            f' "action": "REDUCE_ALL", "rationale": "..."}}\n'
+            f'  ],\n'
+            f'  "stocks": [\n'
+            f'    {{\n'
+            f'      "stock_code": "...",\n'
+            f'      "scenarios": [\n'
+            f'        {{\n'
+            f'          "condition": {{"rsi_below": 30, "volume_ratio_above": 2.0}},\n'
+            f'          "action": "BUY|SELL|HOLD",\n'
+            f'          "confidence": 85,\n'
+            f'          "allocation_pct": 10.0,\n'
+            f'          "stop_loss_pct": -2.0,\n'
+            f'          "take_profit_pct": 3.0,\n'
+            f'          "rationale": "..."\n'
+            f'        }}\n'
+            f'      ]\n'
+            f'    }}\n'
+            f'  ]\n'
+            f'}}\n\n'
+            f"Rules:\n"
+            f"- Max {max_scenarios} scenarios per stock\n"
+            f"- Only use stocks from the candidates list\n"
+            f"- Confidence 0-100 (80+ for actionable trades)\n"
+            f"- stop_loss_pct must be <= 0, take_profit_pct must be >= 0\n"
+            f"- Return ONLY the JSON, no markdown fences or explanation\n"
+        )
+
+    def _parse_response(
+        self,
+        response_text: str,
+        today: date,
+        market: str,
+        candidates: list[ScanCandidate],
+        cross_market: CrossMarketContext | None,
+    ) -> DayPlaybook:
+        """Parse Gemini's JSON response into a validated DayPlaybook."""
+        cleaned = self._extract_json(response_text)
+        data = json.loads(cleaned)
+
+        valid_codes = {c.stock_code for c in candidates}
+
+        # Parse market outlook
+        outlook_str = data.get("market_outlook", "neutral")
+        market_outlook = _OUTLOOK_MAP.get(outlook_str, MarketOutlook.NEUTRAL)
+
+        # Parse global rules
+        global_rules = []
+        for rule_data in data.get("global_rules", []):
+            action_str = rule_data.get("action", "HOLD")
+            action = _ACTION_MAP.get(action_str, ScenarioAction.HOLD)
+            global_rules.append(
+                GlobalRule(
+                    condition=rule_data.get("condition", ""),
+                    action=action,
+                    rationale=rule_data.get("rationale", ""),
+                )
+            )
+
+        # Parse stock playbooks
+        stock_playbooks = []
+        max_scenarios = self._settings.MAX_SCENARIOS_PER_STOCK
+        for stock_data in data.get("stocks", []):
+            code = stock_data.get("stock_code", "")
+            if code not in valid_codes:
+                logger.warning("Gemini returned unknown stock %s — skipping", code)
+                continue
+
+            scenarios = []
+            for sc_data in stock_data.get("scenarios", [])[:max_scenarios]:
+                scenario = self._parse_scenario(sc_data)
+                if scenario:
+                    scenarios.append(scenario)
+
+            if scenarios:
+                stock_playbooks.append(
+                    StockPlaybook(
+                        stock_code=code,
+                        scenarios=scenarios,
+                    )
+                )
+
+        return DayPlaybook(
+            date=today,
+            market=market,
+            market_outlook=market_outlook,
+            global_rules=global_rules,
+            stock_playbooks=stock_playbooks,
+            cross_market=cross_market,
+        )
+
+    def _parse_scenario(self, sc_data: dict) -> StockScenario | None:
+        """Parse a single scenario from JSON data. Returns None if invalid."""
+        try:
+            cond_data = sc_data.get("condition", {})
+            condition = StockCondition(
+                rsi_below=cond_data.get("rsi_below"),
+                rsi_above=cond_data.get("rsi_above"),
+                volume_ratio_above=cond_data.get("volume_ratio_above"),
+                volume_ratio_below=cond_data.get("volume_ratio_below"),
+                price_above=cond_data.get("price_above"),
+                price_below=cond_data.get("price_below"),
+                price_change_pct_above=cond_data.get("price_change_pct_above"),
+                price_change_pct_below=cond_data.get("price_change_pct_below"),
+            )
+
+            if not condition.has_any_condition():
+                logger.warning("Scenario has no conditions — skipping")
+                return None
+
+            action_str = sc_data.get("action", "HOLD")
+            action = _ACTION_MAP.get(action_str, ScenarioAction.HOLD)
+
+            return StockScenario(
+                condition=condition,
+                action=action,
+                confidence=int(sc_data.get("confidence", 50)),
+                allocation_pct=float(sc_data.get("allocation_pct", 10.0)),
+                stop_loss_pct=float(sc_data.get("stop_loss_pct", -2.0)),
+                take_profit_pct=float(sc_data.get("take_profit_pct", 3.0)),
+                rationale=sc_data.get("rationale", ""),
+            )
+        except (ValueError, TypeError) as e:
+            logger.warning("Failed to parse scenario: %s", e)
+            return None
+
+    @staticmethod
+    def _extract_json(text: str) -> str:
+        """Extract JSON from response, stripping markdown fences if present."""
+        stripped = text.strip()
+        if stripped.startswith("```"):
+            # Remove first line (```json or ```) and last line (```)
+            lines = stripped.split("\n")
+            lines = lines[1:]  # Remove opening fence
+            if lines and lines[-1].strip() == "```":
+                lines = lines[:-1]
+            stripped = "\n".join(lines)
+        return stripped.strip()
+
+    @staticmethod
+    def _empty_playbook(today: date, market: str) -> DayPlaybook:
+        """Return an empty playbook (no stocks, no scenarios)."""
+        return DayPlaybook(
+            date=today,
+            market=market,
+            market_outlook=MarketOutlook.NEUTRAL,
+            stock_playbooks=[],
+        )
+
+    @staticmethod
+    def _defensive_playbook(
+        today: date,
+        market: str,
+        candidates: list[ScanCandidate],
+    ) -> DayPlaybook:
+        """Return a defensive playbook — HOLD everything with stop-loss ready."""
+        stock_playbooks = [
+            StockPlaybook(
+                stock_code=c.stock_code,
+                scenarios=[
+                    StockScenario(
+                        condition=StockCondition(price_change_pct_below=-3.0),
+                        action=ScenarioAction.SELL,
+                        confidence=90,
+                        stop_loss_pct=-3.0,
+                        rationale="Defensive stop-loss (planner failure)",
+                    ),
+                ],
+            )
+            for c in candidates
+        ]
+        return DayPlaybook(
+            date=today,
+            market=market,
+            market_outlook=MarketOutlook.NEUTRAL_TO_BEARISH,
+            default_action=ScenarioAction.HOLD,
+            stock_playbooks=stock_playbooks,
+            global_rules=[
+                GlobalRule(
+                    condition="portfolio_pnl_pct < -2.0",
+                    action=ScenarioAction.REDUCE_ALL,
+                    rationale="Defensive: reduce on loss threshold",
+                ),
+            ],
+        )

src/strategy/scenario_engine.py

@@ -0,0 +1,270 @@
+"""Local scenario engine for playbook execution.
+
+Matches real-time market conditions against pre-defined scenarios
+without any API calls. Designed for sub-100ms execution.
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from typing import Any
+
+from src.strategy.models import (
+    DayPlaybook,
+    GlobalRule,
+    ScenarioAction,
+    StockCondition,
+    StockScenario,
+)
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class ScenarioMatch:
+    """Result of matching market conditions against scenarios."""
+
+    stock_code: str
+    matched_scenario: StockScenario | None
+    action: ScenarioAction
+    confidence: int
+    rationale: str
+    global_rule_triggered: GlobalRule | None = None
+    match_details: dict[str, Any] = field(default_factory=dict)
+
+
+class ScenarioEngine:
+    """Evaluates playbook scenarios against real-time market data.
+
+    No API calls — pure Python condition matching.
+
+    Expected market_data keys: "rsi", "volume_ratio", "current_price", "price_change_pct".
+    Callers must normalize data source keys to match this contract.
+    """
+
+    def __init__(self) -> None:
+        self._warned_keys: set[str] = set()
+
+    @staticmethod
+    def _safe_float(value: Any) -> float | None:
+        """Safely cast a value to float. Returns None on failure."""
+        if value is None:
+            return None
+        try:
+            return float(value)
+        except (ValueError, TypeError):
+            return None
+
+    def _warn_missing_key(self, key: str) -> None:
+        """Log a missing-key warning once per key per engine instance."""
+        if key not in self._warned_keys:
+            self._warned_keys.add(key)
+            logger.warning("Condition requires '%s' but key missing from market_data", key)
+
+    def evaluate(
+        self,
+        playbook: DayPlaybook,
+        stock_code: str,
+        market_data: dict[str, Any],
+        portfolio_data: dict[str, Any],
+    ) -> ScenarioMatch:
+        """Match market conditions to scenarios and return a decision.
+
+        Algorithm:
+        1. Check global rules first (portfolio-level circuit breakers)
+        2. Find the StockPlaybook for the given stock_code
+        3. Iterate scenarios in order (first match wins)
+        4. If no match, return playbook.default_action (HOLD)
+
+        Args:
+            playbook: Today's DayPlaybook for this market
+            stock_code: Stock ticker to evaluate
+            market_data: Real-time market data (price, rsi, volume_ratio, etc.)
+            portfolio_data: Portfolio state (pnl_pct, total_cash, etc.)
+
+        Returns:
+            ScenarioMatch with the decision
+        """
+        # 1. Check global rules
+        triggered_rule = self.check_global_rules(playbook, portfolio_data)
+        if triggered_rule is not None:
+            logger.info(
+                "Global rule triggered for %s: %s -> %s",
+                stock_code,
+                triggered_rule.condition,
+                triggered_rule.action.value,
+            )
+            return ScenarioMatch(
+                stock_code=stock_code,
+                matched_scenario=None,
+                action=triggered_rule.action,
+                confidence=100,
+                rationale=f"Global rule: {triggered_rule.rationale or triggered_rule.condition}",
+                global_rule_triggered=triggered_rule,
+            )
+
+        # 2. Find stock playbook
+        stock_pb = playbook.get_stock_playbook(stock_code)
+        if stock_pb is None:
+            logger.debug("No playbook for %s — defaulting to %s", stock_code, playbook.default_action)
+            return ScenarioMatch(
+                stock_code=stock_code,
+                matched_scenario=None,
+                action=playbook.default_action,
+                confidence=0,
+                rationale=f"No scenarios defined for {stock_code}",
+            )
+
+        # 3. Iterate scenarios (first match wins)
+        for scenario in stock_pb.scenarios:
+            if self.evaluate_condition(scenario.condition, market_data):
+                logger.info(
+                    "Scenario matched for %s: %s (confidence=%d)",
+                    stock_code,
+                    scenario.action.value,
+                    scenario.confidence,
+                )
+                return ScenarioMatch(
+                    stock_code=stock_code,
+                    matched_scenario=scenario,
+                    action=scenario.action,
+                    confidence=scenario.confidence,
+                    rationale=scenario.rationale,
+                    match_details=self._build_match_details(scenario.condition, market_data),
+                )
+
+        # 4. No match — default action
+        logger.debug("No scenario matched for %s — defaulting to %s", stock_code, playbook.default_action)
+        return ScenarioMatch(
+            stock_code=stock_code,
+            matched_scenario=None,
+            action=playbook.default_action,
+            confidence=0,
+            rationale="No scenario conditions met — holding position",
+        )
+
+    def check_global_rules(
+        self,
+        playbook: DayPlaybook,
+        portfolio_data: dict[str, Any],
+    ) -> GlobalRule | None:
+        """Check portfolio-level rules. Returns first triggered rule or None."""
+        for rule in playbook.global_rules:
+            if self._evaluate_global_condition(rule.condition, portfolio_data):
+                return rule
+        return None
+
+    def evaluate_condition(
+        self,
+        condition: StockCondition,
+        market_data: dict[str, Any],
+    ) -> bool:
+        """Evaluate all non-None fields in condition as AND.
+
+        Returns True only if ALL specified conditions are met.
+        Empty condition (no fields set) returns False for safety.
+        """
+        if not condition.has_any_condition():
+            return False
+
+        checks: list[bool] = []
+
+        rsi = self._safe_float(market_data.get("rsi"))
+        if condition.rsi_below is not None or condition.rsi_above is not None:
+            if "rsi" not in market_data:
+                self._warn_missing_key("rsi")
+        if condition.rsi_below is not None:
+            checks.append(rsi is not None and rsi < condition.rsi_below)
+        if condition.rsi_above is not None:
+            checks.append(rsi is not None and rsi > condition.rsi_above)
+
+        volume_ratio = self._safe_float(market_data.get("volume_ratio"))
+        if condition.volume_ratio_above is not None or condition.volume_ratio_below is not None:
+            if "volume_ratio" not in market_data:
+                self._warn_missing_key("volume_ratio")
+        if condition.volume_ratio_above is not None:
+            checks.append(volume_ratio is not None and volume_ratio > condition.volume_ratio_above)
+        if condition.volume_ratio_below is not None:
+            checks.append(volume_ratio is not None and volume_ratio < condition.volume_ratio_below)
+
+        price = self._safe_float(market_data.get("current_price"))
+        if condition.price_above is not None or condition.price_below is not None:
+            if "current_price" not in market_data:
+                self._warn_missing_key("current_price")
+        if condition.price_above is not None:
+            checks.append(price is not None and price > condition.price_above)
+        if condition.price_below is not None:
+            checks.append(price is not None and price < condition.price_below)
+
+        price_change_pct = self._safe_float(market_data.get("price_change_pct"))
+        if condition.price_change_pct_above is not None or condition.price_change_pct_below is not None:
+            if "price_change_pct" not in market_data:
+                self._warn_missing_key("price_change_pct")
+        if condition.price_change_pct_above is not None:
+            checks.append(price_change_pct is not None and price_change_pct > condition.price_change_pct_above)
+        if condition.price_change_pct_below is not None:
+            checks.append(price_change_pct is not None and price_change_pct < condition.price_change_pct_below)
+
+        return len(checks) > 0 and all(checks)
+
+    def _evaluate_global_condition(
+        self,
+        condition_str: str,
+        portfolio_data: dict[str, Any],
+    ) -> bool:
+        """Evaluate a simple global condition string against portfolio data.
+
+        Supports: "field < value", "field > value", "field <= value", "field >= value"
+        """
+        parts = condition_str.strip().split()
+        if len(parts) != 3:
+            logger.warning("Invalid global condition format: %s", condition_str)
+            return False
+
+        field_name, operator, value_str = parts
+        try:
+            threshold = float(value_str)
+        except ValueError:
+            logger.warning("Invalid threshold in condition: %s", condition_str)
+            return False
+
+        actual = portfolio_data.get(field_name)
+        if actual is None:
+            return False
+
+        try:
+            actual_val = float(actual)
+        except (ValueError, TypeError):
+            return False
+
+        if operator == "<":
+            return actual_val < threshold
+        elif operator == ">":
+            return actual_val > threshold
+        elif operator == "<=":
+            return actual_val <= threshold
+        elif operator == ">=":
+            return actual_val >= threshold
+        else:
+            logger.warning("Unknown operator in condition: %s", operator)
+            return False
+
+    def _build_match_details(
+        self,
+        condition: StockCondition,
+        market_data: dict[str, Any],
+    ) -> dict[str, Any]:
+        """Build a summary of which conditions matched and their normalized values."""
+        details: dict[str, Any] = {}
+
+        if condition.rsi_below is not None or condition.rsi_above is not None:
+            details["rsi"] = self._safe_float(market_data.get("rsi"))
+        if condition.volume_ratio_above is not None or condition.volume_ratio_below is not None:
+            details["volume_ratio"] = self._safe_float(market_data.get("volume_ratio"))
+        if condition.price_above is not None or condition.price_below is not None:
+            details["current_price"] = self._safe_float(market_data.get("current_price"))
+        if condition.price_change_pct_above is not None or condition.price_change_pct_below is not None:
+            details["price_change_pct"] = self._safe_float(market_data.get("price_change_pct"))
+
+        return details

tests/test_backup.py

@@ -0,0 +1,365 @@
+"""Tests for backup and disaster recovery system."""
+
+from __future__ import annotations
+
+import sqlite3
+import tempfile
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+
+from src.backup.exporter import BackupExporter, ExportFormat
+from src.backup.health_monitor import HealthMonitor, HealthStatus
+from src.backup.scheduler import BackupPolicy, BackupScheduler
+
+
+@pytest.fixture
+def temp_db(tmp_path: Path) -> Path:
+    """Create a temporary test database."""
+    db_path = tmp_path / "test_trades.db"
+
+    conn = sqlite3.connect(str(db_path))
+    cursor = conn.cursor()
+
+    # Create trades table
+    cursor.execute("""
+        CREATE TABLE trades (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            timestamp TEXT NOT NULL,
+            stock_code TEXT NOT NULL,
+            action TEXT NOT NULL,
+            quantity INTEGER NOT NULL,
+            price REAL NOT NULL,
+            confidence INTEGER NOT NULL,
+            rationale TEXT,
+            pnl REAL DEFAULT 0.0
+        )
+    """)
+
+    # Insert test data
+    test_trades = [
+        ("2024-01-01T10:00:00Z", "005930", "BUY", 10, 70000.0, 85, "Test buy", 0.0),
+        ("2024-01-01T11:00:00Z", "005930", "SELL", 10, 71000.0, 90, "Test sell", 10000.0),
+        ("2024-01-02T10:00:00Z", "AAPL", "BUY", 5, 180.0, 88, "Tech buy", 0.0),
+    ]
+
+    cursor.executemany(
+        """
+        INSERT INTO trades (timestamp, stock_code, action, quantity, price, confidence, rationale, pnl)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+        """,
+        test_trades,
+    )
+
+    conn.commit()
+    conn.close()
+
+    return db_path
+
+
+class TestBackupExporter:
+    """Test BackupExporter functionality."""
+
+    def test_exporter_init(self, temp_db: Path) -> None:
+        """Test exporter initialization."""
+        exporter = BackupExporter(str(temp_db))
+        assert exporter.db_path == str(temp_db)
+
+    def test_export_json(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test JSON export."""
+        exporter = BackupExporter(str(temp_db))
+        output_dir = tmp_path / "exports"
+
+        results = exporter.export_all(
+            output_dir, formats=[ExportFormat.JSON], compress=False
+        )
+
+        assert ExportFormat.JSON in results
+        assert results[ExportFormat.JSON].exists()
+        assert results[ExportFormat.JSON].suffix == ".json"
+
+    def test_export_json_compressed(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test compressed JSON export."""
+        exporter = BackupExporter(str(temp_db))
+        output_dir = tmp_path / "exports"
+
+        results = exporter.export_all(
+            output_dir, formats=[ExportFormat.JSON], compress=True
+        )
+
+        assert ExportFormat.JSON in results
+        assert results[ExportFormat.JSON].suffix == ".gz"
+
+    def test_export_csv(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test CSV export."""
+        exporter = BackupExporter(str(temp_db))
+        output_dir = tmp_path / "exports"
+
+        results = exporter.export_all(
+            output_dir, formats=[ExportFormat.CSV], compress=False
+        )
+
+        assert ExportFormat.CSV in results
+        assert results[ExportFormat.CSV].exists()
+
+        # Verify CSV content
+        with open(results[ExportFormat.CSV], "r") as f:
+            lines = f.readlines()
+            assert len(lines) == 4  # Header + 3 rows
+
+    def test_export_all_formats(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test exporting all formats."""
+        exporter = BackupExporter(str(temp_db))
+        output_dir = tmp_path / "exports"
+
+        # Skip Parquet if pyarrow not available
+        try:
+            import pyarrow  # noqa: F401
+
+            formats = [ExportFormat.JSON, ExportFormat.CSV, ExportFormat.PARQUET]
+        except ImportError:
+            formats = [ExportFormat.JSON, ExportFormat.CSV]
+
+        results = exporter.export_all(output_dir, formats=formats, compress=False)
+
+        for fmt in formats:
+            assert fmt in results
+            assert results[fmt].exists()
+
+    def test_incremental_export(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test incremental export."""
+        exporter = BackupExporter(str(temp_db))
+        output_dir = tmp_path / "exports"
+
+        # Export only trades after Jan 2
+        cutoff = datetime(2024, 1, 2, tzinfo=UTC)
+        results = exporter.export_all(
+            output_dir,
+            formats=[ExportFormat.JSON],
+            compress=False,
+            incremental_since=cutoff,
+        )
+
+        # Should only have 1 trade (AAPL on Jan 2)
+        import json
+
+        with open(results[ExportFormat.JSON], "r") as f:
+            data = json.load(f)
+            assert data["record_count"] == 1
+            assert data["trades"][0]["stock_code"] == "AAPL"
+
+    def test_get_export_stats(self, temp_db: Path) -> None:
+        """Test export statistics."""
+        exporter = BackupExporter(str(temp_db))
+        stats = exporter.get_export_stats()
+
+        assert stats["total_trades"] == 3
+        assert "date_range" in stats
+        assert "db_size_bytes" in stats
+
+
+class TestBackupScheduler:
+    """Test BackupScheduler functionality."""
+
+    def test_scheduler_init(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test scheduler initialization."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        assert scheduler.db_path == temp_db
+        assert (backup_dir / "daily").exists()
+        assert (backup_dir / "weekly").exists()
+        assert (backup_dir / "monthly").exists()
+
+    def test_create_daily_backup(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test daily backup creation."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        metadata = scheduler.create_backup(BackupPolicy.DAILY, verify=True)
+
+        assert metadata.policy == BackupPolicy.DAILY
+        assert metadata.file_path.exists()
+        assert metadata.size_bytes > 0
+        assert metadata.checksum is not None
+
+    def test_create_weekly_backup(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test weekly backup creation."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        metadata = scheduler.create_backup(BackupPolicy.WEEKLY, verify=False)
+
+        assert metadata.policy == BackupPolicy.WEEKLY
+        assert metadata.file_path.exists()
+        assert metadata.checksum is None  # verify=False
+
+    def test_list_backups(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test listing backups."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        scheduler.create_backup(BackupPolicy.DAILY)
+        scheduler.create_backup(BackupPolicy.WEEKLY)
+
+        backups = scheduler.list_backups()
+        assert len(backups) == 2
+
+        daily_backups = scheduler.list_backups(BackupPolicy.DAILY)
+        assert len(daily_backups) == 1
+        assert daily_backups[0].policy == BackupPolicy.DAILY
+
+    def test_cleanup_old_backups(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test cleanup of old backups."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir, daily_retention_days=0)
+
+        # Create a backup
+        scheduler.create_backup(BackupPolicy.DAILY)
+
+        # Cleanup should remove it (0 day retention)
+        removed = scheduler.cleanup_old_backups()
+        assert removed[BackupPolicy.DAILY] >= 1
+
+    def test_backup_stats(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test backup statistics."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        scheduler.create_backup(BackupPolicy.DAILY)
+        scheduler.create_backup(BackupPolicy.MONTHLY)
+
+        stats = scheduler.get_backup_stats()
+
+        assert stats["daily"]["count"] == 1
+        assert stats["monthly"]["count"] == 1
+        assert stats["daily"]["total_size_bytes"] > 0
+
+    def test_restore_backup(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test backup restoration."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+
+        # Create backup
+        metadata = scheduler.create_backup(BackupPolicy.DAILY)
+
+        # Modify database
+        conn = sqlite3.connect(str(temp_db))
+        conn.execute("DELETE FROM trades")
+        conn.commit()
+        conn.close()
+
+        # Restore
+        scheduler.restore_backup(metadata, verify=True)
+
+        # Verify restoration
+        conn = sqlite3.connect(str(temp_db))
+        cursor = conn.execute("SELECT COUNT(*) FROM trades")
+        count = cursor.fetchone()[0]
+        conn.close()
+
+        assert count == 3  # Original 3 trades restored
+
+
+class TestHealthMonitor:
+    """Test HealthMonitor functionality."""
+
+    def test_monitor_init(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test monitor initialization."""
+        backup_dir = tmp_path / "backups"
+        monitor = HealthMonitor(str(temp_db), backup_dir)
+
+        assert monitor.db_path == temp_db
+
+    def test_check_database_health_ok(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test database health check (healthy)."""
+        monitor = HealthMonitor(str(temp_db), tmp_path / "backups")
+        result = monitor.check_database_health()
+
+        assert result.status == HealthStatus.HEALTHY
+        assert "healthy" in result.message.lower()
+        assert result.details is not None
+        assert result.details["trade_count"] == 3
+
+    def test_check_database_health_missing(self, tmp_path: Path) -> None:
+        """Test database health check (missing file)."""
+        non_existent = tmp_path / "missing.db"
+        monitor = HealthMonitor(str(non_existent), tmp_path / "backups")
+        result = monitor.check_database_health()
+
+        assert result.status == HealthStatus.UNHEALTHY
+        assert "not found" in result.message.lower()
+
+    def test_check_disk_space(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test disk space check."""
+        monitor = HealthMonitor(str(temp_db), tmp_path, min_disk_space_gb=0.001)
+        result = monitor.check_disk_space()
+
+        # Should be healthy with minimal requirement
+        assert result.status in [HealthStatus.HEALTHY, HealthStatus.DEGRADED]
+        assert result.details is not None
+        assert "free_gb" in result.details
+
+    def test_check_backup_recency_no_backups(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test backup recency check (no backups)."""
+        backup_dir = tmp_path / "backups"
+        backup_dir.mkdir()
+        (backup_dir / "daily").mkdir()
+
+        monitor = HealthMonitor(str(temp_db), backup_dir)
+        result = monitor.check_backup_recency()
+
+        assert result.status == HealthStatus.UNHEALTHY
+        assert "no" in result.message.lower()
+
+    def test_check_backup_recency_recent(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test backup recency check (recent backup)."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+        scheduler.create_backup(BackupPolicy.DAILY)
+
+        monitor = HealthMonitor(str(temp_db), backup_dir)
+        result = monitor.check_backup_recency()
+
+        assert result.status == HealthStatus.HEALTHY
+        assert "recent" in result.message.lower()
+
+    def test_run_all_checks(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test running all health checks."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+        scheduler.create_backup(BackupPolicy.DAILY)
+
+        monitor = HealthMonitor(str(temp_db), backup_dir, min_disk_space_gb=0.001)
+        checks = monitor.run_all_checks()
+
+        assert "database" in checks
+        assert "disk_space" in checks
+        assert "backup_recency" in checks
+        assert checks["database"].status == HealthStatus.HEALTHY
+
+    def test_get_overall_status(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test overall health status."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+        scheduler.create_backup(BackupPolicy.DAILY)
+
+        monitor = HealthMonitor(str(temp_db), backup_dir, min_disk_space_gb=0.001)
+        status = monitor.get_overall_status()
+
+        assert status in [HealthStatus.HEALTHY, HealthStatus.DEGRADED]
+
+    def test_get_health_report(self, temp_db: Path, tmp_path: Path) -> None:
+        """Test health report generation."""
+        backup_dir = tmp_path / "backups"
+        scheduler = BackupScheduler(str(temp_db), backup_dir)
+        scheduler.create_backup(BackupPolicy.DAILY)
+
+        monitor = HealthMonitor(str(temp_db), backup_dir, min_disk_space_gb=0.001)
+        report = monitor.get_health_report()
+
+        assert "overall_status" in report
+        assert "timestamp" in report
+        assert "checks" in report
+        assert len(report["checks"]) == 3

tests/test_brain.py

@@ -152,3 +152,121 @@ class TestPromptConstruction:
         assert "JSON" in prompt
         assert "action" in prompt
         assert "confidence" in prompt
+
+
+# ---------------------------------------------------------------------------
+# Batch Decision Making
+# ---------------------------------------------------------------------------
+
+
+class TestBatchDecisionParsing:
+    """Batch response parser must handle JSON arrays correctly."""
+
+    def test_parse_valid_batch_response(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [
+            {"stock_code": "AAPL", "current_price": 185.5},
+            {"stock_code": "MSFT", "current_price": 420.0},
+        ]
+        raw = """[
+            {"code": "AAPL", "action": "BUY", "confidence": 85, "rationale": "Strong momentum"},
+            {"code": "MSFT", "action": "HOLD", "confidence": 50, "rationale": "Wait for earnings"}
+        ]"""
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert len(decisions) == 2
+        assert decisions["AAPL"].action == "BUY"
+        assert decisions["AAPL"].confidence == 85
+        assert decisions["MSFT"].action == "HOLD"
+        assert decisions["MSFT"].confidence == 50
+
+    def test_parse_batch_with_markdown_wrapper(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = """```json
+[{"code": "AAPL", "action": "BUY", "confidence": 90, "rationale": "Good"}]
+```"""
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "BUY"
+        assert decisions["AAPL"].confidence == 90
+
+    def test_parse_batch_empty_response_returns_hold_for_all(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [
+            {"stock_code": "AAPL", "current_price": 185.5},
+            {"stock_code": "MSFT", "current_price": 420.0},
+        ]
+
+        decisions = client._parse_batch_response("", stocks_data, token_count=100)
+
+        assert len(decisions) == 2
+        assert decisions["AAPL"].action == "HOLD"
+        assert decisions["AAPL"].confidence == 0
+        assert decisions["MSFT"].action == "HOLD"
+
+    def test_parse_batch_malformed_json_returns_hold_for_all(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = "This is not JSON"
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "HOLD"
+        assert decisions["AAPL"].confidence == 0
+
+    def test_parse_batch_not_array_returns_hold_for_all(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = '{"code": "AAPL", "action": "BUY", "confidence": 90, "rationale": "Good"}'
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "HOLD"
+        assert decisions["AAPL"].confidence == 0
+
+    def test_parse_batch_missing_stock_gets_hold(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [
+            {"stock_code": "AAPL", "current_price": 185.5},
+            {"stock_code": "MSFT", "current_price": 420.0},
+        ]
+        # Response only has AAPL, MSFT is missing
+        raw = '[{"code": "AAPL", "action": "BUY", "confidence": 85, "rationale": "Good"}]'
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "BUY"
+        assert decisions["MSFT"].action == "HOLD"
+        assert decisions["MSFT"].confidence == 0
+
+    def test_parse_batch_invalid_action_becomes_hold(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = '[{"code": "AAPL", "action": "YOLO", "confidence": 90, "rationale": "Moon"}]'
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "HOLD"
+
+    def test_parse_batch_low_confidence_becomes_hold(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = '[{"code": "AAPL", "action": "BUY", "confidence": 65, "rationale": "Weak"}]'
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "HOLD"
+        assert decisions["AAPL"].confidence == 65
+
+    def test_parse_batch_missing_fields_gets_hold(self, settings):
+        client = GeminiClient(settings)
+        stocks_data = [{"stock_code": "AAPL", "current_price": 185.5}]
+        raw = '[{"code": "AAPL", "action": "BUY"}]'  # Missing confidence and rationale
+
+        decisions = client._parse_batch_response(raw, stocks_data, token_count=100)
+
+        assert decisions["AAPL"].action == "HOLD"
+        assert decisions["AAPL"].confidence == 0

tests/test_broker.py

@@ -49,6 +49,110 @@ class TestTokenManagement:
 
         await broker.close()
 
+    @pytest.mark.asyncio
+    async def test_concurrent_token_refresh_calls_api_once(self, settings):
+        """Multiple concurrent token requests should only call API once."""
+        broker = KISBroker(settings)
+
+        # Track how many times the mock API is called
+        call_count = [0]
+
+        def create_mock_resp():
+            call_count[0] += 1
+            mock_resp = AsyncMock()
+            mock_resp.status = 200
+            mock_resp.json = AsyncMock(
+                return_value={
+                    "access_token": "tok_concurrent",
+                    "token_type": "Bearer",
+                    "expires_in": 86400,
+                }
+            )
+            mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+            mock_resp.__aexit__ = AsyncMock(return_value=False)
+            return mock_resp
+
+        with patch("aiohttp.ClientSession.post", return_value=create_mock_resp()):
+            # Launch 5 concurrent token requests
+            tokens = await asyncio.gather(
+                broker._ensure_token(),
+                broker._ensure_token(),
+                broker._ensure_token(),
+                broker._ensure_token(),
+                broker._ensure_token(),
+            )
+
+            # All should get the same token
+            assert all(t == "tok_concurrent" for t in tokens)
+            # API should be called only once (due to lock)
+            assert call_count[0] == 1
+
+        await broker.close()
+
+    @pytest.mark.asyncio
+    async def test_token_refresh_cooldown_prevents_rapid_retries(self, settings):
+        """Token refresh should enforce cooldown after failure (issue #54)."""
+        broker = KISBroker(settings)
+        broker._refresh_cooldown = 2.0  # Short cooldown for testing
+
+        # First refresh attempt fails with 403 (EGW00133)
+        mock_resp_403 = AsyncMock()
+        mock_resp_403.status = 403
+        mock_resp_403.text = AsyncMock(
+            return_value='{"error_code":"EGW00133","error_description":"접근토큰 발급 잠시 후 다시 시도하세요(1분당 1회)"}'
+        )
+        mock_resp_403.__aenter__ = AsyncMock(return_value=mock_resp_403)
+        mock_resp_403.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp_403):
+            # First attempt should fail with 403
+            with pytest.raises(ConnectionError, match="Token refresh failed"):
+                await broker._ensure_token()
+
+            # Second attempt within cooldown should fail with cooldown error
+            with pytest.raises(ConnectionError, match="Token refresh on cooldown"):
+                await broker._ensure_token()
+
+        await broker.close()
+
+    @pytest.mark.asyncio
+    async def test_token_refresh_allowed_after_cooldown(self, settings):
+        """Token refresh should be allowed after cooldown period expires."""
+        broker = KISBroker(settings)
+        broker._refresh_cooldown = 0.1  # Very short cooldown for testing
+
+        # First attempt fails
+        mock_resp_403 = AsyncMock()
+        mock_resp_403.status = 403
+        mock_resp_403.text = AsyncMock(return_value='{"error_code":"EGW00133"}')
+        mock_resp_403.__aenter__ = AsyncMock(return_value=mock_resp_403)
+        mock_resp_403.__aexit__ = AsyncMock(return_value=False)
+
+        # Second attempt succeeds
+        mock_resp_200 = AsyncMock()
+        mock_resp_200.status = 200
+        mock_resp_200.json = AsyncMock(
+            return_value={
+                "access_token": "tok_after_cooldown",
+                "expires_in": 86400,
+            }
+        )
+        mock_resp_200.__aenter__ = AsyncMock(return_value=mock_resp_200)
+        mock_resp_200.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp_403):
+            with pytest.raises(ConnectionError, match="Token refresh failed"):
+                await broker._ensure_token()
+
+        # Wait for cooldown to expire
+        await asyncio.sleep(0.15)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp_200):
+            token = await broker._ensure_token()
+            assert token == "tok_after_cooldown"
+
+        await broker.close()
+
 
 # ---------------------------------------------------------------------------
 # Network Error Handling
@@ -107,6 +211,38 @@ class TestRateLimiter:
         await broker._rate_limiter.acquire()
         await broker.close()
 
+    @pytest.mark.asyncio
+    async def test_send_order_acquires_rate_limiter_twice(self, settings):
+        """send_order must acquire rate limiter for both hash key and order call."""
+        broker = KISBroker(settings)
+        broker._access_token = "tok"
+        broker._token_expires_at = asyncio.get_event_loop().time() + 3600
+
+        # Mock hash key response
+        mock_hash_resp = AsyncMock()
+        mock_hash_resp.status = 200
+        mock_hash_resp.json = AsyncMock(return_value={"HASH": "abc123"})
+        mock_hash_resp.__aenter__ = AsyncMock(return_value=mock_hash_resp)
+        mock_hash_resp.__aexit__ = AsyncMock(return_value=False)
+
+        # Mock order response
+        mock_order_resp = AsyncMock()
+        mock_order_resp.status = 200
+        mock_order_resp.json = AsyncMock(return_value={"rt_cd": "0"})
+        mock_order_resp.__aenter__ = AsyncMock(return_value=mock_order_resp)
+        mock_order_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch(
+            "aiohttp.ClientSession.post", side_effect=[mock_hash_resp, mock_order_resp]
+        ):
+            with patch.object(
+                broker._rate_limiter, "acquire", new_callable=AsyncMock
+            ) as mock_acquire:
+                await broker.send_order("005930", "BUY", 1, 50000)
+                assert mock_acquire.call_count == 2
+
+        await broker.close()
+
 
 # ---------------------------------------------------------------------------
 # Hash Key Generation
@@ -136,3 +272,27 @@ class TestHashKey:
             assert len(hash_key) > 0
 
         await broker.close()
+
+    @pytest.mark.asyncio
+    async def test_hash_key_acquires_rate_limiter(self, settings):
+        """_get_hash_key must go through the rate limiter to prevent burst."""
+        broker = KISBroker(settings)
+        broker._access_token = "tok"
+        broker._token_expires_at = asyncio.get_event_loop().time() + 3600
+
+        body = {"CANO": "12345678", "ACNT_PRDT_CD": "01"}
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.json = AsyncMock(return_value={"HASH": "abc123hash"})
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            with patch.object(
+                broker._rate_limiter, "acquire", new_callable=AsyncMock
+            ) as mock_acquire:
+                await broker._get_hash_key(body)
+                mock_acquire.assert_called_once()
+
+        await broker.close()

tests/test_playbook_store.py

@@ -0,0 +1,289 @@
+"""Tests for playbook persistence (PlaybookStore + DB schema)."""
+
+from __future__ import annotations
+
+from datetime import date
+
+import pytest
+
+from src.db import init_db
+from src.strategy.models import (
+    DayPlaybook,
+    GlobalRule,
+    MarketOutlook,
+    PlaybookStatus,
+    ScenarioAction,
+    StockCondition,
+    StockPlaybook,
+    StockScenario,
+)
+from src.strategy.playbook_store import PlaybookStore
+
+
+@pytest.fixture
+def conn():
+    """Create an in-memory DB with schema."""
+    connection = init_db(":memory:")
+    yield connection
+    connection.close()
+
+
+@pytest.fixture
+def store(conn) -> PlaybookStore:
+    return PlaybookStore(conn)
+
+
+def _make_playbook(
+    target_date: date = date(2026, 2, 8),
+    market: str = "KR",
+    outlook: MarketOutlook = MarketOutlook.NEUTRAL,
+    stock_codes: list[str] | None = None,
+) -> DayPlaybook:
+    """Create a test playbook with sensible defaults."""
+    if stock_codes is None:
+        stock_codes = ["005930"]
+    return DayPlaybook(
+        date=target_date,
+        market=market,
+        market_outlook=outlook,
+        token_count=150,
+        stock_playbooks=[
+            StockPlaybook(
+                stock_code=code,
+                scenarios=[
+                    StockScenario(
+                        condition=StockCondition(rsi_below=30.0),
+                        action=ScenarioAction.BUY,
+                        confidence=85,
+                        rationale=f"Oversold bounce for {code}",
+                    ),
+                ],
+            )
+            for code in stock_codes
+        ],
+        global_rules=[
+            GlobalRule(
+                condition="portfolio_pnl_pct < -2.0",
+                action=ScenarioAction.REDUCE_ALL,
+                rationale="Near circuit breaker",
+            ),
+        ],
+    )
+
+
+# ---------------------------------------------------------------------------
+# Schema
+# ---------------------------------------------------------------------------
+
+
+class TestSchema:
+    def test_playbooks_table_exists(self, conn) -> None:
+        row = conn.execute(
+            "SELECT name FROM sqlite_master WHERE type='table' AND name='playbooks'"
+        ).fetchone()
+        assert row is not None
+
+    def test_unique_constraint(self, store: PlaybookStore) -> None:
+        pb = _make_playbook()
+        store.save(pb)
+        # Saving again for same date+market should replace, not error
+        pb2 = _make_playbook(stock_codes=["005930", "000660"])
+        store.save(pb2)
+        loaded = store.load(date(2026, 2, 8), "KR")
+        assert loaded is not None
+        assert loaded.stock_count == 2
+
+
+# ---------------------------------------------------------------------------
+# Save / Load
+# ---------------------------------------------------------------------------
+
+
+class TestSaveLoad:
+    def test_save_and_load(self, store: PlaybookStore) -> None:
+        pb = _make_playbook()
+        row_id = store.save(pb)
+        assert row_id > 0
+
+        loaded = store.load(date(2026, 2, 8), "KR")
+        assert loaded is not None
+        assert loaded.date == date(2026, 2, 8)
+        assert loaded.market == "KR"
+        assert loaded.stock_count == 1
+        assert loaded.scenario_count == 1
+
+    def test_load_not_found(self, store: PlaybookStore) -> None:
+        result = store.load(date(2026, 1, 1), "KR")
+        assert result is None
+
+    def test_save_preserves_all_fields(self, store: PlaybookStore) -> None:
+        pb = _make_playbook(
+            outlook=MarketOutlook.BULLISH,
+            stock_codes=["005930", "AAPL"],
+        )
+        store.save(pb)
+        loaded = store.load(date(2026, 2, 8), "KR")
+        assert loaded is not None
+        assert loaded.market_outlook == MarketOutlook.BULLISH
+        assert loaded.stock_count == 2
+        assert loaded.global_rules[0].action == ScenarioAction.REDUCE_ALL
+        assert loaded.token_count == 150
+
+    def test_save_different_markets(self, store: PlaybookStore) -> None:
+        kr = _make_playbook(market="KR")
+        us = _make_playbook(market="US", stock_codes=["AAPL"])
+        store.save(kr)
+        store.save(us)
+
+        kr_loaded = store.load(date(2026, 2, 8), "KR")
+        us_loaded = store.load(date(2026, 2, 8), "US")
+        assert kr_loaded is not None
+        assert us_loaded is not None
+        assert kr_loaded.market == "KR"
+        assert us_loaded.market == "US"
+        assert kr_loaded.stock_playbooks[0].stock_code == "005930"
+        assert us_loaded.stock_playbooks[0].stock_code == "AAPL"
+
+    def test_save_different_dates(self, store: PlaybookStore) -> None:
+        d1 = _make_playbook(target_date=date(2026, 2, 7))
+        d2 = _make_playbook(target_date=date(2026, 2, 8))
+        store.save(d1)
+        store.save(d2)
+
+        assert store.load(date(2026, 2, 7), "KR") is not None
+        assert store.load(date(2026, 2, 8), "KR") is not None
+
+    def test_replace_updates_data(self, store: PlaybookStore) -> None:
+        pb1 = _make_playbook(outlook=MarketOutlook.BEARISH)
+        store.save(pb1)
+
+        pb2 = _make_playbook(outlook=MarketOutlook.BULLISH)
+        store.save(pb2)
+
+        loaded = store.load(date(2026, 2, 8), "KR")
+        assert loaded is not None
+        assert loaded.market_outlook == MarketOutlook.BULLISH
+
+
+# ---------------------------------------------------------------------------
+# Status
+# ---------------------------------------------------------------------------
+
+
+class TestStatus:
+    def test_get_status(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook())
+        status = store.get_status(date(2026, 2, 8), "KR")
+        assert status == PlaybookStatus.READY
+
+    def test_get_status_not_found(self, store: PlaybookStore) -> None:
+        assert store.get_status(date(2026, 1, 1), "KR") is None
+
+    def test_update_status(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook())
+        updated = store.update_status(date(2026, 2, 8), "KR", PlaybookStatus.EXPIRED)
+        assert updated is True
+
+        status = store.get_status(date(2026, 2, 8), "KR")
+        assert status == PlaybookStatus.EXPIRED
+
+    def test_update_status_not_found(self, store: PlaybookStore) -> None:
+        updated = store.update_status(date(2026, 1, 1), "KR", PlaybookStatus.FAILED)
+        assert updated is False
+
+
+# ---------------------------------------------------------------------------
+# Match count
+# ---------------------------------------------------------------------------
+
+
+class TestMatchCount:
+    def test_increment_match_count(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook())
+        store.increment_match_count(date(2026, 2, 8), "KR")
+        store.increment_match_count(date(2026, 2, 8), "KR")
+
+        stats = store.get_stats(date(2026, 2, 8), "KR")
+        assert stats is not None
+        assert stats["match_count"] == 2
+
+    def test_increment_not_found(self, store: PlaybookStore) -> None:
+        result = store.increment_match_count(date(2026, 1, 1), "KR")
+        assert result is False
+
+
+# ---------------------------------------------------------------------------
+# Stats
+# ---------------------------------------------------------------------------
+
+
+class TestStats:
+    def test_get_stats(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook())
+        stats = store.get_stats(date(2026, 2, 8), "KR")
+        assert stats is not None
+        assert stats["status"] == "ready"
+        assert stats["token_count"] == 150
+        assert stats["scenario_count"] == 1
+        assert stats["match_count"] == 0
+        assert stats["generated_at"] != ""
+
+    def test_get_stats_not_found(self, store: PlaybookStore) -> None:
+        assert store.get_stats(date(2026, 1, 1), "KR") is None
+
+
+# ---------------------------------------------------------------------------
+# List recent
+# ---------------------------------------------------------------------------
+
+
+class TestListRecent:
+    def test_list_recent(self, store: PlaybookStore) -> None:
+        for day in range(5, 10):
+            store.save(_make_playbook(target_date=date(2026, 2, day)))
+        results = store.list_recent(market="KR", limit=3)
+        assert len(results) == 3
+        # Most recent first
+        assert results[0]["date"] == "2026-02-09"
+        assert results[2]["date"] == "2026-02-07"
+
+    def test_list_recent_all_markets(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook(market="KR"))
+        store.save(_make_playbook(market="US", stock_codes=["AAPL"]))
+        results = store.list_recent(market=None, limit=10)
+        assert len(results) == 2
+
+    def test_list_recent_empty(self, store: PlaybookStore) -> None:
+        results = store.list_recent(market="KR")
+        assert results == []
+
+    def test_list_recent_filter_by_market(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook(market="KR"))
+        store.save(_make_playbook(market="US", stock_codes=["AAPL"]))
+        kr_only = store.list_recent(market="KR")
+        assert len(kr_only) == 1
+        assert kr_only[0]["market"] == "KR"
+
+
+# ---------------------------------------------------------------------------
+# Delete
+# ---------------------------------------------------------------------------
+
+
+class TestDelete:
+    def test_delete(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook())
+        deleted = store.delete(date(2026, 2, 8), "KR")
+        assert deleted is True
+        assert store.load(date(2026, 2, 8), "KR") is None
+
+    def test_delete_not_found(self, store: PlaybookStore) -> None:
+        deleted = store.delete(date(2026, 1, 1), "KR")
+        assert deleted is False
+
+    def test_delete_one_market_keeps_other(self, store: PlaybookStore) -> None:
+        store.save(_make_playbook(market="KR"))
+        store.save(_make_playbook(market="US", stock_codes=["AAPL"]))
+        store.delete(date(2026, 2, 8), "KR")
+        assert store.load(date(2026, 2, 8), "KR") is None
+        assert store.load(date(2026, 2, 8), "US") is not None

tests/test_pre_market_planner.py

@@ -0,0 +1,552 @@
+"""Tests for PreMarketPlanner — Gemini prompt builder + response parser."""
+
+from __future__ import annotations
+
+import json
+from datetime import date
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+from src.analysis.smart_scanner import ScanCandidate
+from src.brain.gemini_client import TradeDecision
+from src.config import Settings
+from src.context.store import ContextLayer
+from src.strategy.models import (
+    CrossMarketContext,
+    DayPlaybook,
+    MarketOutlook,
+    PlaybookStatus,
+    ScenarioAction,
+)
+from src.strategy.pre_market_planner import PreMarketPlanner
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+def _candidate(
+    code: str = "005930",
+    name: str = "Samsung",
+    price: float = 71000,
+    rsi: float = 28.5,
+    volume_ratio: float = 3.2,
+    signal: str = "oversold",
+    score: float = 82.0,
+) -> ScanCandidate:
+    return ScanCandidate(
+        stock_code=code,
+        name=name,
+        price=price,
+        volume=1_500_000,
+        volume_ratio=volume_ratio,
+        rsi=rsi,
+        signal=signal,
+        score=score,
+    )
+
+
+def _gemini_response_json(
+    outlook: str = "neutral_to_bullish",
+    stocks: list[dict] | None = None,
+    global_rules: list[dict] | None = None,
+) -> str:
+    """Build a valid Gemini JSON response."""
+    if stocks is None:
+        stocks = [
+            {
+                "stock_code": "005930",
+                "scenarios": [
+                    {
+                        "condition": {"rsi_below": 30, "volume_ratio_above": 2.5},
+                        "action": "BUY",
+                        "confidence": 85,
+                        "allocation_pct": 15.0,
+                        "stop_loss_pct": -2.0,
+                        "take_profit_pct": 4.0,
+                        "rationale": "Oversold bounce with high volume",
+                    }
+                ],
+            }
+        ]
+    if global_rules is None:
+        global_rules = [
+            {
+                "condition": "portfolio_pnl_pct < -2.0",
+                "action": "REDUCE_ALL",
+                "rationale": "Near circuit breaker",
+            }
+        ]
+    return json.dumps(
+        {"market_outlook": outlook, "global_rules": global_rules, "stocks": stocks}
+    )
+
+
+def _make_planner(
+    gemini_response: str = "",
+    token_count: int = 200,
+    context_data: dict | None = None,
+    scorecard_data: dict | None = None,
+) -> PreMarketPlanner:
+    """Create a PreMarketPlanner with mocked dependencies."""
+    if not gemini_response:
+        gemini_response = _gemini_response_json()
+
+    # Mock GeminiClient
+    gemini = AsyncMock()
+    gemini.decide = AsyncMock(
+        return_value=TradeDecision(
+            action="HOLD",
+            confidence=0,
+            rationale=gemini_response,
+            token_count=token_count,
+        )
+    )
+
+    # Mock ContextStore
+    store = MagicMock()
+    store.get_context = MagicMock(return_value=scorecard_data)
+
+    # Mock ContextSelector
+    selector = MagicMock()
+    selector.select_layers = MagicMock(return_value=[ContextLayer.L7_REALTIME, ContextLayer.L6_DAILY])
+    selector.get_context_data = MagicMock(return_value=context_data or {})
+
+    settings = Settings(
+        KIS_APP_KEY="test",
+        KIS_APP_SECRET="test",
+        KIS_ACCOUNT_NO="12345678-01",
+        GEMINI_API_KEY="test",
+    )
+
+    return PreMarketPlanner(gemini, store, selector, settings)
+
+
+# ---------------------------------------------------------------------------
+# generate_playbook
+# ---------------------------------------------------------------------------
+
+
+class TestGeneratePlaybook:
+    @pytest.mark.asyncio
+    async def test_basic_generation(self) -> None:
+        planner = _make_planner()
+        candidates = [_candidate()]
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert isinstance(pb, DayPlaybook)
+        assert pb.market == "KR"
+        assert pb.stock_count == 1
+        assert pb.scenario_count == 1
+        assert pb.market_outlook == MarketOutlook.NEUTRAL_TO_BULLISH
+        assert pb.token_count == 200
+
+    @pytest.mark.asyncio
+    async def test_empty_candidates_returns_empty_playbook(self) -> None:
+        planner = _make_planner()
+
+        pb = await planner.generate_playbook("KR", [], today=date(2026, 2, 8))
+
+        assert pb.stock_count == 0
+        assert pb.scenario_count == 0
+        assert pb.market_outlook == MarketOutlook.NEUTRAL
+
+    @pytest.mark.asyncio
+    async def test_gemini_failure_returns_defensive(self) -> None:
+        planner = _make_planner()
+        planner._gemini.decide = AsyncMock(side_effect=RuntimeError("API timeout"))
+        candidates = [_candidate()]
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert pb.default_action == ScenarioAction.HOLD
+        assert pb.market_outlook == MarketOutlook.NEUTRAL_TO_BEARISH
+        assert pb.stock_count == 1
+        # Defensive playbook has stop-loss scenarios
+        assert pb.stock_playbooks[0].scenarios[0].action == ScenarioAction.SELL
+
+    @pytest.mark.asyncio
+    async def test_gemini_failure_empty_when_defensive_disabled(self) -> None:
+        planner = _make_planner()
+        planner._settings.DEFENSIVE_PLAYBOOK_ON_FAILURE = False
+        planner._gemini.decide = AsyncMock(side_effect=RuntimeError("fail"))
+        candidates = [_candidate()]
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert pb.stock_count == 0
+
+    @pytest.mark.asyncio
+    async def test_multiple_candidates(self) -> None:
+        stocks = [
+            {
+                "stock_code": "005930",
+                "scenarios": [
+                    {
+                        "condition": {"rsi_below": 30},
+                        "action": "BUY",
+                        "confidence": 85,
+                        "rationale": "Oversold",
+                    }
+                ],
+            },
+            {
+                "stock_code": "AAPL",
+                "scenarios": [
+                    {
+                        "condition": {"rsi_above": 75},
+                        "action": "SELL",
+                        "confidence": 80,
+                        "rationale": "Overbought",
+                    }
+                ],
+            },
+        ]
+        planner = _make_planner(gemini_response=_gemini_response_json(stocks=stocks))
+        candidates = [_candidate(), _candidate(code="AAPL", name="Apple")]
+
+        pb = await planner.generate_playbook("US", candidates, today=date(2026, 2, 8))
+
+        assert pb.stock_count == 2
+        codes = [sp.stock_code for sp in pb.stock_playbooks]
+        assert "005930" in codes
+        assert "AAPL" in codes
+
+    @pytest.mark.asyncio
+    async def test_unknown_stock_in_response_skipped(self) -> None:
+        stocks = [
+            {
+                "stock_code": "005930",
+                "scenarios": [{"condition": {"rsi_below": 30}, "action": "BUY", "confidence": 85, "rationale": "ok"}],
+            },
+            {
+                "stock_code": "UNKNOWN",
+                "scenarios": [{"condition": {"rsi_below": 20}, "action": "BUY", "confidence": 90, "rationale": "bad"}],
+            },
+        ]
+        planner = _make_planner(gemini_response=_gemini_response_json(stocks=stocks))
+        candidates = [_candidate()]  # Only 005930
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert pb.stock_count == 1
+        assert pb.stock_playbooks[0].stock_code == "005930"
+
+    @pytest.mark.asyncio
+    async def test_global_rules_parsed(self) -> None:
+        planner = _make_planner()
+        candidates = [_candidate()]
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert len(pb.global_rules) == 1
+        assert pb.global_rules[0].action == ScenarioAction.REDUCE_ALL
+
+    @pytest.mark.asyncio
+    async def test_token_count_from_decision(self) -> None:
+        planner = _make_planner(token_count=450)
+        candidates = [_candidate()]
+
+        pb = await planner.generate_playbook("KR", candidates, today=date(2026, 2, 8))
+
+        assert pb.token_count == 450
+
+
+# ---------------------------------------------------------------------------
+# _parse_response
+# ---------------------------------------------------------------------------
+
+
+class TestParseResponse:
+    def test_parse_full_response(self) -> None:
+        planner = _make_planner()
+        response = _gemini_response_json(outlook="bearish")
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        assert pb.market_outlook == MarketOutlook.BEARISH
+        assert pb.stock_count == 1
+        assert pb.stock_playbooks[0].scenarios[0].confidence == 85
+
+    def test_parse_with_markdown_fences(self) -> None:
+        planner = _make_planner()
+        response = f"```json\n{_gemini_response_json()}\n```"
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        assert pb.stock_count == 1
+
+    def test_parse_unknown_outlook_defaults_neutral(self) -> None:
+        planner = _make_planner()
+        response = _gemini_response_json(outlook="super_bullish")
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        assert pb.market_outlook == MarketOutlook.NEUTRAL
+
+    def test_parse_scenario_with_all_condition_fields(self) -> None:
+        planner = _make_planner()
+        stocks = [
+            {
+                "stock_code": "005930",
+                "scenarios": [
+                    {
+                        "condition": {
+                            "rsi_below": 25,
+                            "volume_ratio_above": 3.0,
+                            "price_change_pct_below": -2.0,
+                        },
+                        "action": "BUY",
+                        "confidence": 92,
+                        "allocation_pct": 20.0,
+                        "stop_loss_pct": -3.0,
+                        "take_profit_pct": 5.0,
+                        "rationale": "Multi-condition entry",
+                    }
+                ],
+            }
+        ]
+        response = _gemini_response_json(stocks=stocks)
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        sc = pb.stock_playbooks[0].scenarios[0]
+        assert sc.condition.rsi_below == 25
+        assert sc.condition.volume_ratio_above == 3.0
+        assert sc.condition.price_change_pct_below == -2.0
+        assert sc.allocation_pct == 20.0
+        assert sc.stop_loss_pct == -3.0
+        assert sc.take_profit_pct == 5.0
+
+    def test_parse_empty_condition_scenario_skipped(self) -> None:
+        planner = _make_planner()
+        stocks = [
+            {
+                "stock_code": "005930",
+                "scenarios": [
+                    {
+                        "condition": {},
+                        "action": "BUY",
+                        "confidence": 85,
+                        "rationale": "No conditions",
+                    },
+                    {
+                        "condition": {"rsi_below": 30},
+                        "action": "BUY",
+                        "confidence": 80,
+                        "rationale": "Valid",
+                    },
+                ],
+            }
+        ]
+        response = _gemini_response_json(stocks=stocks)
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        # Empty condition scenario skipped, valid one kept
+        assert pb.stock_count == 1
+        assert pb.stock_playbooks[0].scenarios[0].confidence == 80
+
+    def test_parse_max_scenarios_enforced(self) -> None:
+        planner = _make_planner()
+        # Settings default MAX_SCENARIOS_PER_STOCK = 5
+        scenarios = [
+            {
+                "condition": {"rsi_below": 20 + i},
+                "action": "BUY",
+                "confidence": 80 + i,
+                "rationale": f"Scenario {i}",
+            }
+            for i in range(8)  # 8 scenarios, should be capped to 5
+        ]
+        stocks = [{"stock_code": "005930", "scenarios": scenarios}]
+        response = _gemini_response_json(stocks=stocks)
+        candidates = [_candidate()]
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, None)
+
+        assert len(pb.stock_playbooks[0].scenarios) == 5
+
+    def test_parse_invalid_json_raises(self) -> None:
+        planner = _make_planner()
+        candidates = [_candidate()]
+
+        with pytest.raises(json.JSONDecodeError):
+            planner._parse_response("not json at all", date(2026, 2, 8), "KR", candidates, None)
+
+    def test_parse_cross_market_preserved(self) -> None:
+        planner = _make_planner()
+        response = _gemini_response_json()
+        candidates = [_candidate()]
+        cross = CrossMarketContext(market="US", date="2026-02-07", total_pnl=1.5, win_rate=60)
+
+        pb = planner._parse_response(response, date(2026, 2, 8), "KR", candidates, cross)
+
+        assert pb.cross_market is not None
+        assert pb.cross_market.market == "US"
+        assert pb.cross_market.total_pnl == 1.5
+
+
+# ---------------------------------------------------------------------------
+# build_cross_market_context
+# ---------------------------------------------------------------------------
+
+
+class TestBuildCrossMarketContext:
+    def test_kr_reads_us_scorecard(self) -> None:
+        scorecard = {"total_pnl": 2.5, "win_rate": 65, "index_change_pct": 0.8, "lessons": ["Stay patient"]}
+        planner = _make_planner(scorecard_data=scorecard)
+
+        ctx = planner.build_cross_market_context("KR", today=date(2026, 2, 8))
+
+        assert ctx is not None
+        assert ctx.market == "US"
+        assert ctx.total_pnl == 2.5
+        assert ctx.win_rate == 65
+        assert "Stay patient" in ctx.lessons
+
+        # Verify it queried scorecard_US
+        planner._context_store.get_context.assert_called_once_with(
+            ContextLayer.L6_DAILY, "2026-02-08", "scorecard_US"
+        )
+
+    def test_us_reads_kr_scorecard(self) -> None:
+        scorecard = {"total_pnl": -1.0, "win_rate": 40, "index_change_pct": -0.5}
+        planner = _make_planner(scorecard_data=scorecard)
+
+        ctx = planner.build_cross_market_context("US", today=date(2026, 2, 8))
+
+        assert ctx is not None
+        assert ctx.market == "KR"
+        assert ctx.total_pnl == -1.0
+
+        planner._context_store.get_context.assert_called_once_with(
+            ContextLayer.L6_DAILY, "2026-02-08", "scorecard_KR"
+        )
+
+    def test_no_scorecard_returns_none(self) -> None:
+        planner = _make_planner(scorecard_data=None)
+
+        ctx = planner.build_cross_market_context("KR", today=date(2026, 2, 8))
+
+        assert ctx is None
+
+    def test_invalid_scorecard_returns_none(self) -> None:
+        planner = _make_planner(scorecard_data="not a dict and not json")
+
+        ctx = planner.build_cross_market_context("KR", today=date(2026, 2, 8))
+
+        assert ctx is None
+
+
+# ---------------------------------------------------------------------------
+# _build_prompt
+# ---------------------------------------------------------------------------
+
+
+class TestBuildPrompt:
+    def test_prompt_contains_candidates(self) -> None:
+        planner = _make_planner()
+        candidates = [_candidate(code="005930", name="Samsung")]
+
+        prompt = planner._build_prompt("KR", candidates, {}, None)
+
+        assert "005930" in prompt
+        assert "Samsung" in prompt
+        assert "RSI=" in prompt
+        assert "volume_ratio=" in prompt
+
+    def test_prompt_contains_cross_market(self) -> None:
+        planner = _make_planner()
+        cross = CrossMarketContext(
+            market="US", date="2026-02-07", total_pnl=1.5,
+            win_rate=60, index_change_pct=0.8, lessons=["Cut losses early"],
+        )
+
+        prompt = planner._build_prompt("KR", [_candidate()], {}, cross)
+
+        assert "Other Market (US)" in prompt
+        assert "+1.50%" in prompt
+        assert "Cut losses early" in prompt
+
+    def test_prompt_contains_context_data(self) -> None:
+        planner = _make_planner()
+        context = {"L6_DAILY": {"win_rate": 0.65, "total_pnl": 2.5}}
+
+        prompt = planner._build_prompt("KR", [_candidate()], context, None)
+
+        assert "Strategic Context" in prompt
+        assert "L6_DAILY" in prompt
+        assert "win_rate" in prompt
+
+    def test_prompt_contains_max_scenarios(self) -> None:
+        planner = _make_planner()
+        prompt = planner._build_prompt("KR", [_candidate()], {}, None)
+
+        assert f"Max {planner._settings.MAX_SCENARIOS_PER_STOCK} scenarios" in prompt
+
+    def test_prompt_market_name(self) -> None:
+        planner = _make_planner()
+        prompt = planner._build_prompt("US", [_candidate()], {}, None)
+        assert "US market" in prompt
+
+
+# ---------------------------------------------------------------------------
+# _extract_json
+# ---------------------------------------------------------------------------
+
+
+class TestExtractJson:
+    def test_plain_json(self) -> None:
+        assert PreMarketPlanner._extract_json('{"a": 1}') == '{"a": 1}'
+
+    def test_with_json_fence(self) -> None:
+        text = '```json\n{"a": 1}\n```'
+        assert PreMarketPlanner._extract_json(text) == '{"a": 1}'
+
+    def test_with_plain_fence(self) -> None:
+        text = '```\n{"a": 1}\n```'
+        assert PreMarketPlanner._extract_json(text) == '{"a": 1}'
+
+    def test_with_whitespace(self) -> None:
+        text = '  \n  {"a": 1}  \n  '
+        assert PreMarketPlanner._extract_json(text) == '{"a": 1}'
+
+
+# ---------------------------------------------------------------------------
+# Defensive playbook
+# ---------------------------------------------------------------------------
+
+
+class TestDefensivePlaybook:
+    def test_defensive_has_stop_loss(self) -> None:
+        candidates = [_candidate(code="005930"), _candidate(code="AAPL")]
+        pb = PreMarketPlanner._defensive_playbook(date(2026, 2, 8), "KR", candidates)
+
+        assert pb.default_action == ScenarioAction.HOLD
+        assert pb.market_outlook == MarketOutlook.NEUTRAL_TO_BEARISH
+        assert pb.stock_count == 2
+        for sp in pb.stock_playbooks:
+            assert sp.scenarios[0].action == ScenarioAction.SELL
+            assert sp.scenarios[0].stop_loss_pct == -3.0
+
+    def test_defensive_has_global_rule(self) -> None:
+        pb = PreMarketPlanner._defensive_playbook(date(2026, 2, 8), "KR", [_candidate()])
+
+        assert len(pb.global_rules) == 1
+        assert pb.global_rules[0].action == ScenarioAction.REDUCE_ALL
+
+    def test_empty_playbook(self) -> None:
+        pb = PreMarketPlanner._empty_playbook(date(2026, 2, 8), "US")
+
+        assert pb.stock_count == 0
+        assert pb.market == "US"
+        assert pb.market_outlook == MarketOutlook.NEUTRAL

tests/test_scenario_engine.py

@@ -0,0 +1,442 @@
+"""Tests for the local scenario engine."""
+
+from __future__ import annotations
+
+from datetime import date
+
+import pytest
+
+from src.strategy.models import (
+    DayPlaybook,
+    GlobalRule,
+    ScenarioAction,
+    StockCondition,
+    StockPlaybook,
+    StockScenario,
+)
+from src.strategy.scenario_engine import ScenarioEngine, ScenarioMatch
+
+
+@pytest.fixture
+def engine() -> ScenarioEngine:
+    return ScenarioEngine()
+
+
+def _scenario(
+    rsi_below: float | None = None,
+    rsi_above: float | None = None,
+    volume_ratio_above: float | None = None,
+    action: ScenarioAction = ScenarioAction.BUY,
+    confidence: int = 85,
+    **kwargs,
+) -> StockScenario:
+    return StockScenario(
+        condition=StockCondition(
+            rsi_below=rsi_below,
+            rsi_above=rsi_above,
+            volume_ratio_above=volume_ratio_above,
+            **kwargs,
+        ),
+        action=action,
+        confidence=confidence,
+        rationale=f"Test scenario: {action.value}",
+    )
+
+
+def _playbook(
+    stock_code: str = "005930",
+    scenarios: list[StockScenario] | None = None,
+    global_rules: list[GlobalRule] | None = None,
+    default_action: ScenarioAction = ScenarioAction.HOLD,
+) -> DayPlaybook:
+    if scenarios is None:
+        scenarios = [_scenario(rsi_below=30.0)]
+    return DayPlaybook(
+        date=date(2026, 2, 7),
+        market="KR",
+        stock_playbooks=[StockPlaybook(stock_code=stock_code, scenarios=scenarios)],
+        global_rules=global_rules or [],
+        default_action=default_action,
+    )
+
+
+# ---------------------------------------------------------------------------
+# evaluate_condition
+# ---------------------------------------------------------------------------
+
+
+class TestEvaluateCondition:
+    def test_rsi_below_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_below=30.0)
+        assert engine.evaluate_condition(cond, {"rsi": 25.0})
+
+    def test_rsi_below_no_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_below=30.0)
+        assert not engine.evaluate_condition(cond, {"rsi": 35.0})
+
+    def test_rsi_above_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_above=70.0)
+        assert engine.evaluate_condition(cond, {"rsi": 75.0})
+
+    def test_rsi_above_no_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_above=70.0)
+        assert not engine.evaluate_condition(cond, {"rsi": 65.0})
+
+    def test_volume_ratio_above_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(volume_ratio_above=3.0)
+        assert engine.evaluate_condition(cond, {"volume_ratio": 4.5})
+
+    def test_volume_ratio_below_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(volume_ratio_below=1.0)
+        assert engine.evaluate_condition(cond, {"volume_ratio": 0.5})
+
+    def test_price_above_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(price_above=50000)
+        assert engine.evaluate_condition(cond, {"current_price": 55000})
+
+    def test_price_below_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(price_below=50000)
+        assert engine.evaluate_condition(cond, {"current_price": 45000})
+
+    def test_price_change_pct_above_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(price_change_pct_above=2.0)
+        assert engine.evaluate_condition(cond, {"price_change_pct": 3.5})
+
+    def test_price_change_pct_below_match(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(price_change_pct_below=-3.0)
+        assert engine.evaluate_condition(cond, {"price_change_pct": -4.0})
+
+    def test_multiple_conditions_and_logic(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_below=30.0, volume_ratio_above=3.0)
+        # Both met
+        assert engine.evaluate_condition(cond, {"rsi": 25.0, "volume_ratio": 4.0})
+        # Only RSI met
+        assert not engine.evaluate_condition(cond, {"rsi": 25.0, "volume_ratio": 2.0})
+        # Only volume met
+        assert not engine.evaluate_condition(cond, {"rsi": 35.0, "volume_ratio": 4.0})
+        # Neither met
+        assert not engine.evaluate_condition(cond, {"rsi": 35.0, "volume_ratio": 2.0})
+
+    def test_empty_condition_returns_false(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition()
+        assert not engine.evaluate_condition(cond, {"rsi": 25.0})
+
+    def test_missing_data_returns_false(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_below=30.0)
+        assert not engine.evaluate_condition(cond, {})
+
+    def test_none_data_returns_false(self, engine: ScenarioEngine) -> None:
+        cond = StockCondition(rsi_below=30.0)
+        assert not engine.evaluate_condition(cond, {"rsi": None})
+
+    def test_boundary_value_not_matched(self, engine: ScenarioEngine) -> None:
+        """rsi_below=30 should NOT match rsi=30 (strict less than)."""
+        cond = StockCondition(rsi_below=30.0)
+        assert not engine.evaluate_condition(cond, {"rsi": 30.0})
+
+    def test_boundary_value_above_not_matched(self, engine: ScenarioEngine) -> None:
+        """rsi_above=70 should NOT match rsi=70 (strict greater than)."""
+        cond = StockCondition(rsi_above=70.0)
+        assert not engine.evaluate_condition(cond, {"rsi": 70.0})
+
+    def test_string_value_no_exception(self, engine: ScenarioEngine) -> None:
+        """String numeric value should not raise TypeError."""
+        cond = StockCondition(rsi_below=30.0)
+        # "25" can be cast to float → should match
+        assert engine.evaluate_condition(cond, {"rsi": "25"})
+        # "35" → should not match
+        assert not engine.evaluate_condition(cond, {"rsi": "35"})
+
+    def test_percent_string_returns_false(self, engine: ScenarioEngine) -> None:
+        """Percent string like '30%' cannot be cast to float → False, no exception."""
+        cond = StockCondition(rsi_below=30.0)
+        assert not engine.evaluate_condition(cond, {"rsi": "30%"})
+
+    def test_decimal_value_no_exception(self, engine: ScenarioEngine) -> None:
+        """Decimal values should be safely handled."""
+        from decimal import Decimal
+
+        cond = StockCondition(rsi_below=30.0)
+        assert engine.evaluate_condition(cond, {"rsi": Decimal("25.0")})
+
+    def test_mixed_invalid_types_no_exception(self, engine: ScenarioEngine) -> None:
+        """Various invalid types should not raise exceptions."""
+        cond = StockCondition(
+            rsi_below=30.0, volume_ratio_above=2.0,
+            price_above=100, price_change_pct_below=-1.0,
+        )
+        data = {
+            "rsi": [25],           # list
+            "volume_ratio": "bad",  # non-numeric string
+            "current_price": {},    # dict
+            "price_change_pct": object(),  # arbitrary object
+        }
+        # Should return False (invalid types → None → False), never raise
+        assert not engine.evaluate_condition(cond, data)
+
+    def test_missing_key_logs_warning_once(self, caplog) -> None:
+        """Missing key warning should fire only once per key per engine instance."""
+        import logging
+
+        eng = ScenarioEngine()
+        cond = StockCondition(rsi_below=30.0)
+        with caplog.at_level(logging.WARNING):
+            eng.evaluate_condition(cond, {})
+            eng.evaluate_condition(cond, {})
+            eng.evaluate_condition(cond, {})
+        # Warning should appear exactly once despite 3 calls
+        assert caplog.text.count("'rsi' but key missing") == 1
+
+
+# ---------------------------------------------------------------------------
+# check_global_rules
+# ---------------------------------------------------------------------------
+
+
+class TestCheckGlobalRules:
+    def test_no_rules(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(global_rules=[])
+        result = engine.check_global_rules(pb, {"portfolio_pnl_pct": -1.0})
+        assert result is None
+
+    def test_rule_triggered(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(
+                    condition="portfolio_pnl_pct < -2.0",
+                    action=ScenarioAction.REDUCE_ALL,
+                    rationale="Near circuit breaker",
+                ),
+            ]
+        )
+        result = engine.check_global_rules(pb, {"portfolio_pnl_pct": -2.5})
+        assert result is not None
+        assert result.action == ScenarioAction.REDUCE_ALL
+
+    def test_rule_not_triggered(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(
+                    condition="portfolio_pnl_pct < -2.0",
+                    action=ScenarioAction.REDUCE_ALL,
+                ),
+            ]
+        )
+        result = engine.check_global_rules(pb, {"portfolio_pnl_pct": -1.0})
+        assert result is None
+
+    def test_first_rule_wins(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="portfolio_pnl_pct < -2.0", action=ScenarioAction.REDUCE_ALL),
+                GlobalRule(condition="portfolio_pnl_pct < -1.0", action=ScenarioAction.HOLD),
+            ]
+        )
+        result = engine.check_global_rules(pb, {"portfolio_pnl_pct": -2.5})
+        assert result is not None
+        assert result.action == ScenarioAction.REDUCE_ALL
+
+    def test_greater_than_operator(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="volatility_index > 30", action=ScenarioAction.HOLD),
+            ]
+        )
+        result = engine.check_global_rules(pb, {"volatility_index": 35})
+        assert result is not None
+
+    def test_missing_field_not_triggered(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="unknown_field < -2.0", action=ScenarioAction.REDUCE_ALL),
+            ]
+        )
+        result = engine.check_global_rules(pb, {"portfolio_pnl_pct": -5.0})
+        assert result is None
+
+    def test_invalid_condition_format(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="bad format", action=ScenarioAction.HOLD),
+            ]
+        )
+        result = engine.check_global_rules(pb, {})
+        assert result is None
+
+    def test_le_operator(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="portfolio_pnl_pct <= -2.0", action=ScenarioAction.REDUCE_ALL),
+            ]
+        )
+        assert engine.check_global_rules(pb, {"portfolio_pnl_pct": -2.0}) is not None
+        assert engine.check_global_rules(pb, {"portfolio_pnl_pct": -1.9}) is None
+
+    def test_ge_operator(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            global_rules=[
+                GlobalRule(condition="volatility >= 80.0", action=ScenarioAction.HOLD),
+            ]
+        )
+        assert engine.check_global_rules(pb, {"volatility": 80.0}) is not None
+        assert engine.check_global_rules(pb, {"volatility": 79.9}) is None
+
+
+# ---------------------------------------------------------------------------
+# evaluate (full pipeline)
+# ---------------------------------------------------------------------------
+
+
+class TestEvaluate:
+    def test_scenario_match(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(scenarios=[_scenario(rsi_below=30.0)])
+        result = engine.evaluate(pb, "005930", {"rsi": 25.0}, {})
+        assert result.action == ScenarioAction.BUY
+        assert result.confidence == 85
+        assert result.matched_scenario is not None
+
+    def test_no_scenario_match_returns_default(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(scenarios=[_scenario(rsi_below=30.0)])
+        result = engine.evaluate(pb, "005930", {"rsi": 50.0}, {})
+        assert result.action == ScenarioAction.HOLD
+        assert result.confidence == 0
+        assert result.matched_scenario is None
+
+    def test_stock_not_in_playbook(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(stock_code="005930")
+        result = engine.evaluate(pb, "AAPL", {"rsi": 25.0}, {})
+        assert result.action == ScenarioAction.HOLD
+        assert result.confidence == 0
+
+    def test_global_rule_takes_priority(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            scenarios=[_scenario(rsi_below=30.0)],
+            global_rules=[
+                GlobalRule(
+                    condition="portfolio_pnl_pct < -2.0",
+                    action=ScenarioAction.REDUCE_ALL,
+                    rationale="Loss limit",
+                ),
+            ],
+        )
+        result = engine.evaluate(
+            pb,
+            "005930",
+            {"rsi": 25.0},  # Would match scenario
+            {"portfolio_pnl_pct": -2.5},  # But global rule triggers first
+        )
+        assert result.action == ScenarioAction.REDUCE_ALL
+        assert result.global_rule_triggered is not None
+        assert result.matched_scenario is None
+
+    def test_first_scenario_wins(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            scenarios=[
+                _scenario(rsi_below=30.0, action=ScenarioAction.BUY, confidence=90),
+                _scenario(rsi_below=25.0, action=ScenarioAction.BUY, confidence=95),
+            ]
+        )
+        result = engine.evaluate(pb, "005930", {"rsi": 20.0}, {})
+        # Both match, but first wins
+        assert result.confidence == 90
+
+    def test_sell_scenario(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            scenarios=[
+                _scenario(rsi_above=75.0, action=ScenarioAction.SELL, confidence=80),
+            ]
+        )
+        result = engine.evaluate(pb, "005930", {"rsi": 80.0}, {})
+        assert result.action == ScenarioAction.SELL
+
+    def test_empty_playbook(self, engine: ScenarioEngine) -> None:
+        pb = DayPlaybook(date=date(2026, 2, 7), market="KR", stock_playbooks=[])
+        result = engine.evaluate(pb, "005930", {"rsi": 25.0}, {})
+        assert result.action == ScenarioAction.HOLD
+
+    def test_match_details_populated(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(scenarios=[_scenario(rsi_below=30.0, volume_ratio_above=2.0)])
+        result = engine.evaluate(
+            pb, "005930", {"rsi": 25.0, "volume_ratio": 3.0}, {}
+        )
+        assert result.match_details.get("rsi") == 25.0
+        assert result.match_details.get("volume_ratio") == 3.0
+
+    def test_custom_default_action(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            scenarios=[_scenario(rsi_below=10.0)],  # Very unlikely to match
+            default_action=ScenarioAction.SELL,
+        )
+        result = engine.evaluate(pb, "005930", {"rsi": 50.0}, {})
+        assert result.action == ScenarioAction.SELL
+
+    def test_multiple_stocks_in_playbook(self, engine: ScenarioEngine) -> None:
+        pb = DayPlaybook(
+            date=date(2026, 2, 7),
+            market="US",
+            stock_playbooks=[
+                StockPlaybook(
+                    stock_code="AAPL",
+                    scenarios=[_scenario(rsi_below=25.0, confidence=90)],
+                ),
+                StockPlaybook(
+                    stock_code="MSFT",
+                    scenarios=[_scenario(rsi_above=75.0, action=ScenarioAction.SELL, confidence=80)],
+                ),
+            ],
+        )
+        aapl = engine.evaluate(pb, "AAPL", {"rsi": 20.0}, {})
+        assert aapl.action == ScenarioAction.BUY
+        assert aapl.confidence == 90
+
+        msft = engine.evaluate(pb, "MSFT", {"rsi": 80.0}, {})
+        assert msft.action == ScenarioAction.SELL
+
+    def test_complex_multi_condition(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(
+            scenarios=[
+                _scenario(
+                    rsi_below=30.0,
+                    volume_ratio_above=3.0,
+                    price_change_pct_below=-2.0,
+                    confidence=95,
+                ),
+            ]
+        )
+        # All conditions met
+        result = engine.evaluate(
+            pb,
+            "005930",
+            {"rsi": 22.0, "volume_ratio": 4.0, "price_change_pct": -3.0},
+            {},
+        )
+        assert result.action == ScenarioAction.BUY
+        assert result.confidence == 95
+
+        # One condition not met
+        result2 = engine.evaluate(
+            pb,
+            "005930",
+            {"rsi": 22.0, "volume_ratio": 4.0, "price_change_pct": -1.0},
+            {},
+        )
+        assert result2.action == ScenarioAction.HOLD
+
+    def test_scenario_match_returns_rationale(self, engine: ScenarioEngine) -> None:
+        pb = _playbook(scenarios=[_scenario(rsi_below=30.0)])
+        result = engine.evaluate(pb, "005930", {"rsi": 25.0}, {})
+        assert result.rationale != ""
+
+    def test_result_stock_code(self, engine: ScenarioEngine) -> None:
+        pb = _playbook()
+        result = engine.evaluate(pb, "005930", {"rsi": 25.0}, {})
+        assert result.stock_code == "005930"
+
+    def test_match_details_normalized(self, engine: ScenarioEngine) -> None:
+        """match_details should contain _safe_float normalized values, not raw."""
+        pb = _playbook(scenarios=[_scenario(rsi_below=30.0)])
+        # Pass string value — should be normalized to float in match_details
+        result = engine.evaluate(pb, "005930", {"rsi": "25.0"}, {})
+        assert result.action == ScenarioAction.BUY
+        assert result.match_details["rsi"] == 25.0
+        assert isinstance(result.match_details["rsi"], float)

tests/test_smart_scanner.py

@@ -0,0 +1,377 @@
+"""Tests for SmartVolatilityScanner."""
+
+from __future__ import annotations
+
+import pytest
+from unittest.mock import AsyncMock, MagicMock
+
+from src.analysis.smart_scanner import ScanCandidate, SmartVolatilityScanner
+from src.analysis.volatility import VolatilityAnalyzer
+from src.broker.kis_api import KISBroker
+from src.config import Settings
+
+
+@pytest.fixture
+def mock_settings() -> Settings:
+    """Create test settings."""
+    return Settings(
+        KIS_APP_KEY="test",
+        KIS_APP_SECRET="test",
+        KIS_ACCOUNT_NO="12345678-01",
+        GEMINI_API_KEY="test",
+        RSI_OVERSOLD_THRESHOLD=30,
+        RSI_MOMENTUM_THRESHOLD=70,
+        VOL_MULTIPLIER=2.0,
+        SCANNER_TOP_N=3,
+        DB_PATH=":memory:",
+    )
+
+
+@pytest.fixture
+def mock_broker(mock_settings: Settings) -> MagicMock:
+    """Create mock broker."""
+    broker = MagicMock(spec=KISBroker)
+    broker._settings = mock_settings
+    broker.fetch_market_rankings = AsyncMock()
+    broker.get_daily_prices = AsyncMock()
+    return broker
+
+
+@pytest.fixture
+def scanner(mock_broker: MagicMock, mock_settings: Settings) -> SmartVolatilityScanner:
+    """Create smart scanner instance."""
+    analyzer = VolatilityAnalyzer()
+    return SmartVolatilityScanner(
+        broker=mock_broker,
+        volatility_analyzer=analyzer,
+        settings=mock_settings,
+    )
+
+
+class TestSmartVolatilityScanner:
+    """Test suite for SmartVolatilityScanner."""
+
+    @pytest.mark.asyncio
+    async def test_scan_finds_oversold_candidates(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that scanner identifies oversold stocks with high volume."""
+        # Mock rankings
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": "005930",
+                "name": "Samsung",
+                "price": 70000,
+                "volume": 5000000,
+                "change_rate": -3.5,
+                "volume_increase_rate": 250,
+            },
+        ]
+
+        # Mock daily prices - trending down (oversold)
+        prices = []
+        for i in range(20):
+            prices.append({
+                "date": f"2026020{i:02d}",
+                "open": 75000 - i * 200,
+                "high": 75500 - i * 200,
+                "low": 74500 - i * 200,
+                "close": 75000 - i * 250,  # Steady decline
+                "volume": 2000000,
+            })
+        mock_broker.get_daily_prices.return_value = prices
+
+        candidates = await scanner.scan()
+
+        # Should find at least one candidate (depending on exact RSI calculation)
+        mock_broker.fetch_market_rankings.assert_called_once()
+        mock_broker.get_daily_prices.assert_called_once_with("005930", days=20)
+
+        # If qualified, should have oversold signal
+        if candidates:
+            assert candidates[0].signal in ["oversold", "momentum"]
+            assert candidates[0].volume_ratio >= scanner.vol_multiplier
+
+    @pytest.mark.asyncio
+    async def test_scan_finds_momentum_candidates(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that scanner identifies momentum stocks with high volume."""
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": "035420",
+                "name": "NAVER",
+                "price": 250000,
+                "volume": 3000000,
+                "change_rate": 5.0,
+                "volume_increase_rate": 300,
+            },
+        ]
+
+        # Mock daily prices - trending up (momentum)
+        prices = []
+        for i in range(20):
+            prices.append({
+                "date": f"2026020{i:02d}",
+                "open": 230000 + i * 500,
+                "high": 231000 + i * 500,
+                "low": 229000 + i * 500,
+                "close": 230500 + i * 500,  # Steady rise
+                "volume": 1000000,
+            })
+        mock_broker.get_daily_prices.return_value = prices
+
+        candidates = await scanner.scan()
+
+        mock_broker.fetch_market_rankings.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_scan_filters_low_volume(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that stocks with low volume ratio are filtered out."""
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": "000660",
+                "name": "SK Hynix",
+                "price": 150000,
+                "volume": 500000,
+                "change_rate": -5.0,
+                "volume_increase_rate": 50,  # Only 50% increase (< 200%)
+            },
+        ]
+
+        # Low volume
+        prices = []
+        for i in range(20):
+            prices.append({
+                "date": f"2026020{i:02d}",
+                "open": 150000 - i * 100,
+                "high": 151000 - i * 100,
+                "low": 149000 - i * 100,
+                "close": 150000 - i * 150,  # Declining (would be oversold)
+                "volume": 1000000,  # Current 500k < 2x prev day 1M
+            })
+        mock_broker.get_daily_prices.return_value = prices
+
+        candidates = await scanner.scan()
+
+        # Should be filtered out due to low volume ratio
+        assert len(candidates) == 0
+
+    @pytest.mark.asyncio
+    async def test_scan_filters_neutral_rsi(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that stocks with neutral RSI are filtered out."""
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": "051910",
+                "name": "LG Chem",
+                "price": 500000,
+                "volume": 3000000,
+                "change_rate": 0.5,
+                "volume_increase_rate": 300,  # High volume
+            },
+        ]
+
+        # Flat prices (neutral RSI ~50)
+        prices = []
+        for i in range(20):
+            prices.append({
+                "date": f"2026020{i:02d}",
+                "open": 500000 + (i % 2) * 100,  # Small oscillation
+                "high": 500500,
+                "low": 499500,
+                "close": 500000 + (i % 2) * 50,
+                "volume": 1000000,
+            })
+        mock_broker.get_daily_prices.return_value = prices
+
+        candidates = await scanner.scan()
+
+        # Should be filtered out (RSI ~50, not < 30 or > 70)
+        assert len(candidates) == 0
+
+    @pytest.mark.asyncio
+    async def test_scan_uses_fallback_on_api_error(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test fallback to static list when ranking API fails."""
+        mock_broker.fetch_market_rankings.side_effect = ConnectionError("API unavailable")
+
+        # Fallback stocks should still be analyzed
+        prices = []
+        for i in range(20):
+            prices.append({
+                "date": f"2026020{i:02d}",
+                "open": 50000 - i * 50,
+                "high": 51000 - i * 50,
+                "low": 49000 - i * 50,
+                "close": 50000 - i * 75,  # Declining
+                "volume": 1000000,
+            })
+        mock_broker.get_daily_prices.return_value = prices
+
+        candidates = await scanner.scan(fallback_stocks=["005930", "000660"])
+
+        # Should not crash
+        assert isinstance(candidates, list)
+
+    @pytest.mark.asyncio
+    async def test_scan_returns_top_n_only(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that scan returns at most top_n candidates."""
+        # Return many stocks
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": f"00{i}000",
+                "name": f"Stock{i}",
+                "price": 10000 * i,
+                "volume": 5000000,
+                "change_rate": -10,
+                "volume_increase_rate": 500,
+            }
+            for i in range(1, 10)
+        ]
+
+        # All oversold with high volume
+        def make_prices(code: str) -> list[dict]:
+            prices = []
+            for i in range(20):
+                prices.append({
+                    "date": f"2026020{i:02d}",
+                    "open": 10000 - i * 100,
+                    "high": 10500 - i * 100,
+                    "low": 9500 - i * 100,
+                    "close": 10000 - i * 150,
+                    "volume": 1000000,
+                })
+            return prices
+
+        mock_broker.get_daily_prices.side_effect = make_prices
+
+        candidates = await scanner.scan()
+
+        # Should respect top_n limit (3)
+        assert len(candidates) <= scanner.top_n
+
+    @pytest.mark.asyncio
+    async def test_scan_skips_insufficient_price_history(
+        self, scanner: SmartVolatilityScanner, mock_broker: MagicMock
+    ) -> None:
+        """Test that stocks with insufficient history are skipped."""
+        mock_broker.fetch_market_rankings.return_value = [
+            {
+                "stock_code": "005930",
+                "name": "Samsung",
+                "price": 70000,
+                "volume": 5000000,
+                "change_rate": -5.0,
+                "volume_increase_rate": 300,
+            },
+        ]
+
+        # Only 5 days of data (need 15+ for RSI)
+        mock_broker.get_daily_prices.return_value = [
+            {
+                "date": f"2026020{i:02d}",
+                "open": 70000,
+                "high": 71000,
+                "low": 69000,
+                "close": 70000,
+                "volume": 2000000,
+            }
+            for i in range(5)
+        ]
+
+        candidates = await scanner.scan()
+
+        # Should skip due to insufficient data
+        assert len(candidates) == 0
+
+    @pytest.mark.asyncio
+    async def test_get_stock_codes(
+        self, scanner: SmartVolatilityScanner
+    ) -> None:
+        """Test extraction of stock codes from candidates."""
+        candidates = [
+            ScanCandidate(
+                stock_code="005930",
+                name="Samsung",
+                price=70000,
+                volume=5000000,
+                volume_ratio=2.5,
+                rsi=28,
+                signal="oversold",
+                score=85.0,
+            ),
+            ScanCandidate(
+                stock_code="035420",
+                name="NAVER",
+                price=250000,
+                volume=3000000,
+                volume_ratio=3.0,
+                rsi=75,
+                signal="momentum",
+                score=88.0,
+            ),
+        ]
+
+        codes = scanner.get_stock_codes(candidates)
+
+        assert codes == ["005930", "035420"]
+
+
+class TestRSICalculation:
+    """Test RSI calculation in VolatilityAnalyzer."""
+
+    def test_rsi_oversold(self) -> None:
+        """Test RSI calculation for downtrending prices."""
+        analyzer = VolatilityAnalyzer()
+
+        # Steadily declining prices
+        prices = [100 - i * 0.5 for i in range(20)]
+        rsi = analyzer.calculate_rsi(prices, period=14)
+
+        assert rsi < 50  # Should be oversold territory
+
+    def test_rsi_overbought(self) -> None:
+        """Test RSI calculation for uptrending prices."""
+        analyzer = VolatilityAnalyzer()
+
+        # Steadily rising prices
+        prices = [100 + i * 0.5 for i in range(20)]
+        rsi = analyzer.calculate_rsi(prices, period=14)
+
+        assert rsi > 50  # Should be overbought territory
+
+    def test_rsi_neutral(self) -> None:
+        """Test RSI calculation for flat prices."""
+        analyzer = VolatilityAnalyzer()
+
+        # Flat prices with small oscillation
+        prices = [100 + (i % 2) * 0.1 for i in range(20)]
+        rsi = analyzer.calculate_rsi(prices, period=14)
+
+        assert 40 < rsi < 60  # Should be near neutral
+
+    def test_rsi_insufficient_data(self) -> None:
+        """Test RSI returns neutral when insufficient data."""
+        analyzer = VolatilityAnalyzer()
+
+        prices = [100, 101, 102]  # Only 3 prices, need 15+
+        rsi = analyzer.calculate_rsi(prices, period=14)
+
+        assert rsi == 50.0  # Default neutral
+
+    def test_rsi_all_gains(self) -> None:
+        """Test RSI returns 100 when all gains (no losses)."""
+        analyzer = VolatilityAnalyzer()
+
+        # Monotonic increase
+        prices = [100 + i for i in range(20)]
+        rsi = analyzer.calculate_rsi(prices, period=14)
+
+        assert rsi == 100.0  # Maximum RSI

tests/test_strategy_models.py

@@ -0,0 +1,366 @@
+"""Tests for strategy/playbook Pydantic models."""
+
+from __future__ import annotations
+
+from datetime import date
+
+import pytest
+from pydantic import ValidationError
+
+from src.strategy.models import (
+    CrossMarketContext,
+    DayPlaybook,
+    GlobalRule,
+    MarketOutlook,
+    PlaybookStatus,
+    ScenarioAction,
+    StockCondition,
+    StockPlaybook,
+    StockScenario,
+)
+
+
+# ---------------------------------------------------------------------------
+# StockCondition
+# ---------------------------------------------------------------------------
+
+
+class TestStockCondition:
+    def test_empty_condition(self) -> None:
+        cond = StockCondition()
+        assert not cond.has_any_condition()
+
+    def test_single_field(self) -> None:
+        cond = StockCondition(rsi_below=30.0)
+        assert cond.has_any_condition()
+
+    def test_multiple_fields(self) -> None:
+        cond = StockCondition(rsi_below=25.0, volume_ratio_above=3.0)
+        assert cond.has_any_condition()
+
+    def test_all_fields(self) -> None:
+        cond = StockCondition(
+            rsi_below=30,
+            rsi_above=10,
+            volume_ratio_above=2.0,
+            volume_ratio_below=10.0,
+            price_above=1000,
+            price_below=50000,
+            price_change_pct_above=-5.0,
+            price_change_pct_below=5.0,
+        )
+        assert cond.has_any_condition()
+
+
+# ---------------------------------------------------------------------------
+# StockScenario
+# ---------------------------------------------------------------------------
+
+
+class TestStockScenario:
+    def test_valid_scenario(self) -> None:
+        s = StockScenario(
+            condition=StockCondition(rsi_below=25.0),
+            action=ScenarioAction.BUY,
+            confidence=85,
+            allocation_pct=15.0,
+            stop_loss_pct=-2.0,
+            take_profit_pct=3.0,
+            rationale="Oversold bounce expected",
+        )
+        assert s.action == ScenarioAction.BUY
+        assert s.confidence == 85
+
+    def test_confidence_too_high(self) -> None:
+        with pytest.raises(ValidationError):
+            StockScenario(
+                condition=StockCondition(),
+                action=ScenarioAction.BUY,
+                confidence=101,
+            )
+
+    def test_confidence_too_low(self) -> None:
+        with pytest.raises(ValidationError):
+            StockScenario(
+                condition=StockCondition(),
+                action=ScenarioAction.BUY,
+                confidence=-1,
+            )
+
+    def test_allocation_too_high(self) -> None:
+        with pytest.raises(ValidationError):
+            StockScenario(
+                condition=StockCondition(),
+                action=ScenarioAction.BUY,
+                confidence=80,
+                allocation_pct=101.0,
+            )
+
+    def test_stop_loss_must_be_negative(self) -> None:
+        with pytest.raises(ValidationError):
+            StockScenario(
+                condition=StockCondition(),
+                action=ScenarioAction.BUY,
+                confidence=80,
+                stop_loss_pct=1.0,
+            )
+
+    def test_take_profit_must_be_positive(self) -> None:
+        with pytest.raises(ValidationError):
+            StockScenario(
+                condition=StockCondition(),
+                action=ScenarioAction.BUY,
+                confidence=80,
+                take_profit_pct=-1.0,
+            )
+
+    def test_defaults(self) -> None:
+        s = StockScenario(
+            condition=StockCondition(),
+            action=ScenarioAction.HOLD,
+            confidence=50,
+        )
+        assert s.allocation_pct == 10.0
+        assert s.stop_loss_pct == -2.0
+        assert s.take_profit_pct == 3.0
+        assert s.rationale == ""
+
+
+# ---------------------------------------------------------------------------
+# StockPlaybook
+# ---------------------------------------------------------------------------
+
+
+class TestStockPlaybook:
+    def test_valid_playbook(self) -> None:
+        pb = StockPlaybook(
+            stock_code="005930",
+            stock_name="Samsung Electronics",
+            scenarios=[
+                StockScenario(
+                    condition=StockCondition(rsi_below=25.0),
+                    action=ScenarioAction.BUY,
+                    confidence=85,
+                ),
+            ],
+        )
+        assert pb.stock_code == "005930"
+        assert len(pb.scenarios) == 1
+
+    def test_empty_scenarios_rejected(self) -> None:
+        with pytest.raises(ValidationError):
+            StockPlaybook(
+                stock_code="005930",
+                scenarios=[],
+            )
+
+    def test_multiple_scenarios(self) -> None:
+        pb = StockPlaybook(
+            stock_code="AAPL",
+            scenarios=[
+                StockScenario(
+                    condition=StockCondition(rsi_below=25.0),
+                    action=ScenarioAction.BUY,
+                    confidence=85,
+                ),
+                StockScenario(
+                    condition=StockCondition(rsi_above=75.0),
+                    action=ScenarioAction.SELL,
+                    confidence=80,
+                ),
+            ],
+        )
+        assert len(pb.scenarios) == 2
+
+
+# ---------------------------------------------------------------------------
+# GlobalRule
+# ---------------------------------------------------------------------------
+
+
+class TestGlobalRule:
+    def test_valid_rule(self) -> None:
+        rule = GlobalRule(
+            condition="portfolio_pnl_pct < -2.0",
+            action=ScenarioAction.REDUCE_ALL,
+            rationale="Risk limit approaching",
+        )
+        assert rule.action == ScenarioAction.REDUCE_ALL
+
+    def test_hold_rule(self) -> None:
+        rule = GlobalRule(
+            condition="volatility_index > 30",
+            action=ScenarioAction.HOLD,
+        )
+        assert rule.rationale == ""
+
+
+# ---------------------------------------------------------------------------
+# CrossMarketContext
+# ---------------------------------------------------------------------------
+
+
+class TestCrossMarketContext:
+    def test_valid_context(self) -> None:
+        ctx = CrossMarketContext(
+            market="US",
+            date="2026-02-07",
+            total_pnl=-1.5,
+            win_rate=40.0,
+            index_change_pct=-2.3,
+            key_events=["Fed rate decision"],
+            lessons=["Avoid tech sector on rate hike days"],
+        )
+        assert ctx.market == "US"
+        assert len(ctx.key_events) == 1
+
+    def test_defaults(self) -> None:
+        ctx = CrossMarketContext(market="KR", date="2026-02-07")
+        assert ctx.total_pnl == 0.0
+        assert ctx.key_events == []
+        assert ctx.lessons == []
+
+
+# ---------------------------------------------------------------------------
+# DayPlaybook
+# ---------------------------------------------------------------------------
+
+
+def _make_scenario(rsi_below: float = 25.0) -> StockScenario:
+    return StockScenario(
+        condition=StockCondition(rsi_below=rsi_below),
+        action=ScenarioAction.BUY,
+        confidence=85,
+    )
+
+
+def _make_playbook(**kwargs) -> DayPlaybook:
+    defaults = {
+        "date": date(2026, 2, 7),
+        "market": "KR",
+        "stock_playbooks": [
+            StockPlaybook(stock_code="005930", scenarios=[_make_scenario()]),
+        ],
+    }
+    defaults.update(kwargs)
+    return DayPlaybook(**defaults)
+
+
+class TestDayPlaybook:
+    def test_valid_playbook(self) -> None:
+        pb = _make_playbook()
+        assert pb.market == "KR"
+        assert pb.date == date(2026, 2, 7)
+        assert pb.default_action == ScenarioAction.HOLD
+        assert pb.scenario_count == 1
+        assert pb.stock_count == 1
+
+    def test_generated_at_auto_set(self) -> None:
+        pb = _make_playbook()
+        assert pb.generated_at != ""
+
+    def test_explicit_generated_at(self) -> None:
+        pb = _make_playbook(generated_at="2026-02-07T08:30:00")
+        assert pb.generated_at == "2026-02-07T08:30:00"
+
+    def test_duplicate_stocks_rejected(self) -> None:
+        with pytest.raises(ValidationError):
+            DayPlaybook(
+                date=date(2026, 2, 7),
+                market="KR",
+                stock_playbooks=[
+                    StockPlaybook(stock_code="005930", scenarios=[_make_scenario()]),
+                    StockPlaybook(stock_code="005930", scenarios=[_make_scenario(30)]),
+                ],
+            )
+
+    def test_empty_stock_playbooks_allowed(self) -> None:
+        pb = DayPlaybook(
+            date=date(2026, 2, 7),
+            market="KR",
+            stock_playbooks=[],
+        )
+        assert pb.stock_count == 0
+        assert pb.scenario_count == 0
+
+    def test_get_stock_playbook_found(self) -> None:
+        pb = _make_playbook()
+        result = pb.get_stock_playbook("005930")
+        assert result is not None
+        assert result.stock_code == "005930"
+
+    def test_get_stock_playbook_not_found(self) -> None:
+        pb = _make_playbook()
+        result = pb.get_stock_playbook("AAPL")
+        assert result is None
+
+    def test_with_global_rules(self) -> None:
+        pb = _make_playbook(
+            global_rules=[
+                GlobalRule(
+                    condition="portfolio_pnl_pct < -2.0",
+                    action=ScenarioAction.REDUCE_ALL,
+                ),
+            ],
+        )
+        assert len(pb.global_rules) == 1
+
+    def test_with_cross_market_context(self) -> None:
+        ctx = CrossMarketContext(market="US", date="2026-02-07", total_pnl=-1.5)
+        pb = _make_playbook(cross_market=ctx)
+        assert pb.cross_market is not None
+        assert pb.cross_market.market == "US"
+
+    def test_market_outlook(self) -> None:
+        pb = _make_playbook(market_outlook=MarketOutlook.BEARISH)
+        assert pb.market_outlook == MarketOutlook.BEARISH
+
+    def test_multiple_stocks_multiple_scenarios(self) -> None:
+        pb = DayPlaybook(
+            date=date(2026, 2, 7),
+            market="US",
+            stock_playbooks=[
+                StockPlaybook(
+                    stock_code="AAPL",
+                    scenarios=[_make_scenario(), _make_scenario(30)],
+                ),
+                StockPlaybook(
+                    stock_code="MSFT",
+                    scenarios=[_make_scenario()],
+                ),
+            ],
+        )
+        assert pb.stock_count == 2
+        assert pb.scenario_count == 3
+
+    def test_serialization_roundtrip(self) -> None:
+        pb = _make_playbook(
+            market_outlook=MarketOutlook.BULLISH,
+            cross_market=CrossMarketContext(market="US", date="2026-02-07"),
+        )
+        json_str = pb.model_dump_json()
+        restored = DayPlaybook.model_validate_json(json_str)
+        assert restored.market == pb.market
+        assert restored.date == pb.date
+        assert restored.scenario_count == pb.scenario_count
+        assert restored.cross_market is not None
+
+
+# ---------------------------------------------------------------------------
+# Enums
+# ---------------------------------------------------------------------------
+
+
+class TestEnums:
+    def test_scenario_action_values(self) -> None:
+        assert ScenarioAction.BUY.value == "BUY"
+        assert ScenarioAction.SELL.value == "SELL"
+        assert ScenarioAction.HOLD.value == "HOLD"
+        assert ScenarioAction.REDUCE_ALL.value == "REDUCE_ALL"
+
+    def test_market_outlook_values(self) -> None:
+        assert len(MarketOutlook) == 5
+
+    def test_playbook_status_values(self) -> None:
+        assert PlaybookStatus.READY.value == "ready"
+        assert PlaybookStatus.EXPIRED.value == "expired"

tests/test_telegram.py

@@ -0,0 +1,483 @@
+"""Tests for Telegram notification client."""
+
+from unittest.mock import AsyncMock, patch
+
+import aiohttp
+import pytest
+
+from src.notifications.telegram_client import NotificationPriority, TelegramClient
+
+
+class TestTelegramClientInit:
+    """Test client initialization scenarios."""
+
+    def test_disabled_via_flag(self) -> None:
+        """Client disabled via enabled=False flag."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=False
+        )
+        assert client._enabled is False
+
+    def test_disabled_missing_token(self) -> None:
+        """Client disabled when bot_token is None."""
+        client = TelegramClient(bot_token=None, chat_id="456", enabled=True)
+        assert client._enabled is False
+
+    def test_disabled_missing_chat_id(self) -> None:
+        """Client disabled when chat_id is None."""
+        client = TelegramClient(bot_token="123:abc", chat_id=None, enabled=True)
+        assert client._enabled is False
+
+    def test_enabled_with_credentials(self) -> None:
+        """Client enabled when credentials provided."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+        assert client._enabled is True
+
+
+class TestNotificationSending:
+    """Test notification sending behavior."""
+
+    @pytest.mark.asyncio
+    async def test_send_message_success(self) -> None:
+        """send_message returns True on successful send."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            result = await client.send_message("Test message")
+
+            assert result is True
+            assert mock_post.call_count == 1
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert payload["chat_id"] == "456"
+            assert payload["text"] == "Test message"
+            assert payload["parse_mode"] == "HTML"
+
+    @pytest.mark.asyncio
+    async def test_send_message_disabled_client(self) -> None:
+        """send_message returns False when client disabled."""
+        client = TelegramClient(enabled=False)
+
+        with patch("aiohttp.ClientSession.post") as mock_post:
+            result = await client.send_message("Test message")
+
+            assert result is False
+            mock_post.assert_not_called()
+
+    @pytest.mark.asyncio
+    async def test_send_message_api_error(self) -> None:
+        """send_message returns False on API error."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 400
+        mock_resp.text = AsyncMock(return_value="Bad Request")
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            result = await client.send_message("Test message")
+            assert result is False
+
+    @pytest.mark.asyncio
+    async def test_send_message_with_markdown(self) -> None:
+        """send_message supports different parse modes."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            result = await client.send_message("*bold*", parse_mode="Markdown")
+
+            assert result is True
+            payload = mock_post.call_args.kwargs["json"]
+            assert payload["parse_mode"] == "Markdown"
+
+    @pytest.mark.asyncio
+    async def test_no_send_when_disabled(self) -> None:
+        """Notifications not sent when client disabled."""
+        client = TelegramClient(enabled=False)
+
+        with patch("aiohttp.ClientSession.post") as mock_post:
+            await client.notify_trade_execution(
+                stock_code="AAPL",
+                market="United States",
+                action="BUY",
+                quantity=10,
+                price=150.0,
+                confidence=85.0,
+            )
+            mock_post.assert_not_called()
+
+    @pytest.mark.asyncio
+    async def test_trade_execution_format(self) -> None:
+        """Trade notification has correct format."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_trade_execution(
+                stock_code="TSLA",
+                market="United States",
+                action="SELL",
+                quantity=5,
+                price=250.50,
+                confidence=92.0,
+            )
+
+            # Verify API call was made
+            assert mock_post.call_count == 1
+            call_args = mock_post.call_args
+
+            # Check payload structure
+            payload = call_args.kwargs["json"]
+            assert payload["chat_id"] == "456"
+            assert "TSLA" in payload["text"]
+            assert "SELL" in payload["text"]
+            assert "5" in payload["text"]
+            assert "250.50" in payload["text"]
+            assert "92%" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_playbook_generated_format(self) -> None:
+        """Playbook generated notification has expected fields."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_playbook_generated(
+                market="KR",
+                stock_count=4,
+                scenario_count=12,
+                token_count=980,
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Playbook Generated" in payload["text"]
+            assert "Market: KR" in payload["text"]
+            assert "Stocks: 4" in payload["text"]
+            assert "Scenarios: 12" in payload["text"]
+            assert "Tokens: 980" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_scenario_matched_format(self) -> None:
+        """Scenario matched notification has expected fields."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_scenario_matched(
+                stock_code="AAPL",
+                action="BUY",
+                condition_summary="RSI < 30, volume_ratio > 2.0",
+                confidence=88.2,
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Scenario Matched" in payload["text"]
+            assert "AAPL" in payload["text"]
+            assert "Action: BUY" in payload["text"]
+            assert "RSI < 30" in payload["text"]
+            assert "88%" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_playbook_failed_format(self) -> None:
+        """Playbook failed notification has expected fields."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_playbook_failed(
+                market="US",
+                reason="Gemini timeout",
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Playbook Failed" in payload["text"]
+            assert "Market: US" in payload["text"]
+            assert "Gemini timeout" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_circuit_breaker_priority(self) -> None:
+        """Circuit breaker uses CRITICAL priority."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_circuit_breaker(pnl_pct=-3.15, threshold=-3.0)
+
+            payload = mock_post.call_args.kwargs["json"]
+            # CRITICAL priority has 🚨 emoji
+            assert NotificationPriority.CRITICAL.emoji in payload["text"]
+            assert "-3.15%" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_api_error_handling(self) -> None:
+        """API errors logged but don't crash."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 400
+        mock_resp.text = AsyncMock(return_value="Bad Request")
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            # Should not raise exception
+            await client.notify_system_start(mode="paper", enabled_markets=["KR"])
+
+    @pytest.mark.asyncio
+    async def test_timeout_handling(self) -> None:
+        """Timeouts logged but don't crash."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        with patch(
+            "aiohttp.ClientSession.post",
+            side_effect=aiohttp.ClientError("Connection timeout"),
+        ):
+            # Should not raise exception
+            await client.notify_error(
+                error_type="Test Error", error_msg="Test", context="test"
+            )
+
+    @pytest.mark.asyncio
+    async def test_session_management(self) -> None:
+        """Session created and reused correctly."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        # Session should be None initially
+        assert client._session is None
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            await client.notify_market_open("Korea")
+            # Session should be created
+            assert client._session is not None
+
+            session1 = client._session
+            await client.notify_market_close("Korea", 1.5)
+            # Same session should be reused
+            assert client._session is session1
+
+
+class TestRateLimiting:
+    """Test rate limiter behavior."""
+
+    @pytest.mark.asyncio
+    async def test_rate_limiter_enforced(self) -> None:
+        """Rate limiter delays rapid requests."""
+        import time
+
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True, rate_limit=2.0
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            start = time.monotonic()
+
+            # Send 3 messages (rate: 2/sec = 0.5s per message)
+            await client.notify_market_open("Korea")
+            await client.notify_market_open("United States")
+            await client.notify_market_open("Japan")
+
+            elapsed = time.monotonic() - start
+
+            # Should take at least 0.4 seconds (3 msgs at 2/sec with some tolerance)
+            assert elapsed >= 0.4
+
+
+class TestMessagePriorities:
+    """Test priority-based messaging."""
+
+    @pytest.mark.asyncio
+    async def test_low_priority_uses_info_emoji(self) -> None:
+        """LOW priority uses ℹ️ emoji."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_market_open("Korea")
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert NotificationPriority.LOW.emoji in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_critical_priority_uses_alarm_emoji(self) -> None:
+        """CRITICAL priority uses 🚨 emoji."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_system_shutdown("Circuit breaker tripped")
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert NotificationPriority.CRITICAL.emoji in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_playbook_generated_priority(self) -> None:
+        """Playbook generated uses MEDIUM priority emoji."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_playbook_generated(
+                market="KR",
+                stock_count=2,
+                scenario_count=4,
+                token_count=123,
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert NotificationPriority.MEDIUM.emoji in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_playbook_failed_priority(self) -> None:
+        """Playbook failed uses HIGH priority emoji."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_playbook_failed(
+                market="KR",
+                reason="Invalid JSON",
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert NotificationPriority.HIGH.emoji in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_scenario_matched_priority(self) -> None:
+        """Scenario matched uses HIGH priority emoji."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            await client.notify_scenario_matched(
+                stock_code="AAPL",
+                action="BUY",
+                condition_summary="RSI < 30",
+                confidence=80.0,
+            )
+
+            payload = mock_post.call_args.kwargs["json"]
+            assert NotificationPriority.HIGH.emoji in payload["text"]
+
+
+class TestClientCleanup:
+    """Test client cleanup behavior."""
+
+    @pytest.mark.asyncio
+    async def test_close_closes_session(self) -> None:
+        """close() closes the HTTP session."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        mock_session = AsyncMock()
+        mock_session.closed = False
+        mock_session.close = AsyncMock()
+        client._session = mock_session
+
+        await client.close()
+        mock_session.close.assert_called_once()
+
+    @pytest.mark.asyncio
+    async def test_close_handles_no_session(self) -> None:
+        """close() handles None session gracefully."""
+        client = TelegramClient(
+            bot_token="123:abc", chat_id="456", enabled=True
+        )
+
+        # Should not raise exception
+        await client.close()

tests/test_telegram_commands.py

@@ -0,0 +1,777 @@
+"""Tests for Telegram command handler."""
+
+from unittest.mock import AsyncMock, patch
+
+import pytest
+
+from src.notifications.telegram_client import TelegramClient, TelegramCommandHandler
+
+
+class TestCommandHandlerInit:
+    """Test command handler initialization."""
+
+    def test_init_with_client(self) -> None:
+        """Handler initializes with TelegramClient."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        assert handler._client is client
+        assert handler._polling_interval == 1.0
+        assert handler._commands == {}
+        assert handler._running is False
+
+    def test_custom_polling_interval(self) -> None:
+        """Handler accepts custom polling interval."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client, polling_interval=2.5)
+
+        assert handler._polling_interval == 2.5
+
+
+class TestCommandRegistration:
+    """Test command registration."""
+
+    @pytest.mark.asyncio
+    async def test_register_command(self) -> None:
+        """Commands can be registered."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        async def test_handler() -> None:
+            pass
+
+        handler.register_command("test", test_handler)
+
+        assert "test" in handler._commands
+        assert handler._commands["test"] is test_handler
+
+    @pytest.mark.asyncio
+    async def test_register_multiple_commands(self) -> None:
+        """Multiple commands can be registered."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        async def handler1() -> None:
+            pass
+
+        async def handler2() -> None:
+            pass
+
+        handler.register_command("start", handler1)
+        handler.register_command("help", handler2)
+
+        assert len(handler._commands) == 2
+        assert handler._commands["start"] is handler1
+        assert handler._commands["help"] is handler2
+
+
+class TestPollingLifecycle:
+    """Test polling start/stop."""
+
+    @pytest.mark.asyncio
+    async def test_start_polling(self) -> None:
+        """Polling can be started."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        with patch.object(handler, "_poll_loop", new_callable=AsyncMock):
+            await handler.start_polling()
+
+            assert handler._running is True
+            assert handler._polling_task is not None
+
+        await handler.stop_polling()
+
+    @pytest.mark.asyncio
+    async def test_start_polling_disabled_client(self) -> None:
+        """Polling not started when client disabled."""
+        client = TelegramClient(enabled=False)
+        handler = TelegramCommandHandler(client)
+
+        await handler.start_polling()
+
+        assert handler._running is False
+        assert handler._polling_task is None
+
+    @pytest.mark.asyncio
+    async def test_stop_polling(self) -> None:
+        """Polling can be stopped."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        with patch.object(handler, "_poll_loop", new_callable=AsyncMock):
+            await handler.start_polling()
+            await handler.stop_polling()
+
+            assert handler._running is False
+
+    @pytest.mark.asyncio
+    async def test_double_start_ignored(self) -> None:
+        """Starting already running handler is ignored."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        with patch.object(handler, "_poll_loop", new_callable=AsyncMock):
+            await handler.start_polling()
+            task1 = handler._polling_task
+
+            await handler.start_polling()  # Second start
+            task2 = handler._polling_task
+
+            # Should be the same task
+            assert task1 is task2
+
+        await handler.stop_polling()
+
+
+class TestUpdateHandling:
+    """Test update parsing and handling."""
+
+    @pytest.mark.asyncio
+    async def test_handle_valid_command(self) -> None:
+        """Valid commands are executed."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        executed = False
+
+        async def test_command() -> None:
+            nonlocal executed
+            executed = True
+
+        handler.register_command("test", test_command)
+
+        update = {
+            "update_id": 1,
+            "message": {
+                "chat": {"id": 456},
+                "text": "/test",
+            },
+        }
+
+        await handler._handle_update(update)
+        assert executed is True
+
+    @pytest.mark.asyncio
+    async def test_handle_unknown_command(self) -> None:
+        """Unknown commands send help message."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/unknown",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Should send error message
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Unknown command" in payload["text"]
+            assert "/unknown" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_ignore_unauthorized_chat(self) -> None:
+        """Commands from unauthorized chats are ignored."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        executed = False
+
+        async def test_command() -> None:
+            nonlocal executed
+            executed = True
+
+        handler.register_command("test", test_command)
+
+        update = {
+            "update_id": 1,
+            "message": {
+                "chat": {"id": 999},  # Wrong chat_id
+                "text": "/test",
+            },
+        }
+
+        await handler._handle_update(update)
+        assert executed is False
+
+    @pytest.mark.asyncio
+    async def test_ignore_non_command_text(self) -> None:
+        """Non-command text is ignored."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        executed = False
+
+        async def test_command() -> None:
+            nonlocal executed
+            executed = True
+
+        handler.register_command("test", test_command)
+
+        update = {
+            "update_id": 1,
+            "message": {
+                "chat": {"id": 456},
+                "text": "Hello, not a command",
+            },
+        }
+
+        await handler._handle_update(update)
+        assert executed is False
+
+    @pytest.mark.asyncio
+    async def test_handle_command_with_botname(self) -> None:
+        """Commands with @botname suffix are handled correctly."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        executed = False
+
+        async def test_command() -> None:
+            nonlocal executed
+            executed = True
+
+        handler.register_command("start", test_command)
+
+        update = {
+            "update_id": 1,
+            "message": {
+                "chat": {"id": 456},
+                "text": "/start@mybot",
+            },
+        }
+
+        await handler._handle_update(update)
+        assert executed is True
+
+    @pytest.mark.asyncio
+    async def test_handle_update_error_isolation(self) -> None:
+        """Errors in handlers don't crash the system."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        async def failing_command() -> None:
+            raise ValueError("Test error")
+
+        handler.register_command("fail", failing_command)
+
+        update = {
+            "update_id": 1,
+            "message": {
+                "chat": {"id": 456},
+                "text": "/fail",
+            },
+        }
+
+        # Should not raise exception
+        await handler._handle_update(update)
+
+
+class TestTradingControlCommands:
+    """Test trading control commands."""
+
+    @pytest.mark.asyncio
+    async def test_stop_command_pauses_trading(self) -> None:
+        """Stop command clears pause event."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        # Create mock pause event
+        import asyncio
+
+        pause_event = asyncio.Event()
+        pause_event.set()  # Initially active
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_stop() -> None:
+            """Mock /stop handler."""
+            if not pause_event.is_set():
+                await client.send_message("⏸️ Trading is already paused")
+                return
+
+            pause_event.clear()
+            await client.send_message(
+                "<b>⏸️ Trading Paused</b>\n\n"
+                "All trading operations have been suspended.\n"
+                "Use /resume to restart trading."
+            )
+
+        handler.register_command("stop", mock_stop)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/stop",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify pause event was cleared
+            assert not pause_event.is_set()
+
+            # Verify message was sent
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Trading Paused" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_resume_command_resumes_trading(self) -> None:
+        """Resume command sets pause event."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        # Create mock pause event (initially paused)
+        import asyncio
+
+        pause_event = asyncio.Event()
+        pause_event.clear()  # Initially paused
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_resume() -> None:
+            """Mock /resume handler."""
+            if pause_event.is_set():
+                await client.send_message("▶️ Trading is already active")
+                return
+
+            pause_event.set()
+            await client.send_message(
+                "<b>▶️ Trading Resumed</b>\n\n"
+                "Trading operations have been restarted."
+            )
+
+        handler.register_command("resume", mock_resume)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/resume",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify pause event was set
+            assert pause_event.is_set()
+
+            # Verify message was sent
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Trading Resumed" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_stop_when_already_paused(self) -> None:
+        """Stop command when already paused sends appropriate message."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        # Create mock pause event (already paused)
+        import asyncio
+
+        pause_event = asyncio.Event()
+        pause_event.clear()
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_stop() -> None:
+            """Mock /stop handler."""
+            if not pause_event.is_set():
+                await client.send_message("⏸️ Trading is already paused")
+                return
+
+            pause_event.clear()
+
+        handler.register_command("stop", mock_stop)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/stop",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            payload = mock_post.call_args.kwargs["json"]
+            assert "already paused" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_resume_when_already_active(self) -> None:
+        """Resume command when already active sends appropriate message."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        # Create mock pause event (already active)
+        import asyncio
+
+        pause_event = asyncio.Event()
+        pause_event.set()
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_resume() -> None:
+            """Mock /resume handler."""
+            if pause_event.is_set():
+                await client.send_message("▶️ Trading is already active")
+                return
+
+            pause_event.set()
+
+        handler.register_command("resume", mock_resume)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/resume",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            payload = mock_post.call_args.kwargs["json"]
+            assert "already active" in payload["text"]
+
+
+class TestStatusCommands:
+    """Test status query commands."""
+
+    @pytest.mark.asyncio
+    async def test_status_command_shows_trading_info(self) -> None:
+        """Status command displays mode, markets, and P&L."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_status() -> None:
+            """Mock /status handler."""
+            message = (
+                "<b>📊 Trading Status</b>\n\n"
+                "<b>Mode:</b> PAPER\n"
+                "<b>Markets:</b> Korea, United States\n"
+                "<b>Trading:</b> Active\n\n"
+                "<b>Current P&L:</b> +2.50%\n"
+                "<b>Circuit Breaker:</b> -3.0%"
+            )
+            await client.send_message(message)
+
+        handler.register_command("status", mock_status)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/status",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Trading Status" in payload["text"]
+            assert "PAPER" in payload["text"]
+            assert "P&L" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_status_command_error_handling(self) -> None:
+        """Status command handles errors gracefully."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_status_error() -> None:
+            """Mock /status handler with error."""
+            await client.send_message(
+                "<b>⚠️ Error</b>\n\nFailed to retrieve trading status."
+            )
+
+        handler.register_command("status", mock_status_error)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/status",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Should send error message
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Error" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_positions_command_shows_holdings(self) -> None:
+        """Positions command displays account summary."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_positions() -> None:
+            """Mock /positions handler."""
+            message = (
+                "<b>💼 Account Summary</b>\n\n"
+                "<b>Total Evaluation:</b> ₩10,500,000\n"
+                "<b>Available Cash:</b> ₩5,000,000\n"
+                "<b>Purchase Total:</b> ₩10,000,000\n"
+                "<b>P&L:</b> +5.00%\n\n"
+                "<i>Note: Individual position details require API enhancement</i>"
+            )
+            await client.send_message(message)
+
+        handler.register_command("positions", mock_positions)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/positions",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Account Summary" in payload["text"]
+            assert "Total Evaluation" in payload["text"]
+            assert "P&L" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_positions_command_empty_holdings(self) -> None:
+        """Positions command handles empty portfolio."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_positions_empty() -> None:
+            """Mock /positions handler with no positions."""
+            message = (
+                "<b>💼 Account Summary</b>\n\n"
+                "No balance information available."
+            )
+            await client.send_message(message)
+
+        handler.register_command("positions", mock_positions_empty)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/positions",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            payload = mock_post.call_args.kwargs["json"]
+            assert "No balance information available" in payload["text"]
+
+    @pytest.mark.asyncio
+    async def test_positions_command_error_handling(self) -> None:
+        """Positions command handles errors gracefully."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_positions_error() -> None:
+            """Mock /positions handler with error."""
+            await client.send_message(
+                "<b>⚠️ Error</b>\n\nFailed to retrieve positions."
+            )
+
+        handler.register_command("positions", mock_positions_error)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/positions",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Should send error message
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Error" in payload["text"]
+
+
+class TestBasicCommands:
+    """Test basic command implementations."""
+
+    @pytest.mark.asyncio
+    async def test_help_command_content(self) -> None:
+        """Help command lists all available commands."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        async def mock_help() -> None:
+            """Mock /help handler."""
+            message = (
+                "<b>📖 Available Commands</b>\n\n"
+                "/help - Show available commands\n"
+                "/status - Trading status (mode, markets, P&L)\n"
+                "/positions - Current holdings\n"
+                "/stop - Pause trading\n"
+                "/resume - Resume trading"
+            )
+            await client.send_message(message)
+
+        handler.register_command("help", mock_help)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp) as mock_post:
+            update = {
+                "update_id": 1,
+                "message": {
+                    "chat": {"id": 456},
+                    "text": "/help",
+                },
+            }
+
+            await handler._handle_update(update)
+
+            # Verify message was sent
+            assert mock_post.call_count == 1
+            payload = mock_post.call_args.kwargs["json"]
+            assert "Available Commands" in payload["text"]
+            assert "/help" in payload["text"]
+            assert "/status" in payload["text"]
+            assert "/positions" in payload["text"]
+            assert "/stop" in payload["text"]
+            assert "/resume" in payload["text"]
+
+
+class TestGetUpdates:
+    """Test getUpdates API interaction."""
+
+    @pytest.mark.asyncio
+    async def test_get_updates_success(self) -> None:
+        """getUpdates fetches and parses updates."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.json = AsyncMock(
+            return_value={
+                "ok": True,
+                "result": [
+                    {"update_id": 1, "message": {"text": "/test"}},
+                    {"update_id": 2, "message": {"text": "/help"}},
+                ],
+            }
+        )
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            updates = await handler._get_updates()
+
+            assert len(updates) == 2
+            assert updates[0]["update_id"] == 1
+            assert updates[1]["update_id"] == 2
+            assert handler._last_update_id == 2
+
+    @pytest.mark.asyncio
+    async def test_get_updates_api_error(self) -> None:
+        """getUpdates handles API errors gracefully."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 400
+        mock_resp.text = AsyncMock(return_value="Bad Request")
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            updates = await handler._get_updates()
+
+            assert updates == []
+
+    @pytest.mark.asyncio
+    async def test_get_updates_empty_result(self) -> None:
+        """getUpdates handles empty results."""
+        client = TelegramClient(bot_token="123:abc", chat_id="456", enabled=True)
+        handler = TelegramCommandHandler(client)
+
+        mock_resp = AsyncMock()
+        mock_resp.status = 200
+        mock_resp.json = AsyncMock(return_value={"ok": True, "result": []})
+        mock_resp.__aenter__ = AsyncMock(return_value=mock_resp)
+        mock_resp.__aexit__ = AsyncMock(return_value=False)
+
+        with patch("aiohttp.ClientSession.post", return_value=mock_resp):
+            updates = await handler._get_updates()
+
+            assert updates == []

tests/test_volatility.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import asyncio
 import sqlite3
 from typing import Any
 from unittest.mock import AsyncMock
@@ -338,6 +339,28 @@ class TestMarketScanner:
         assert metrics.stock_code == "AAPL"
         assert metrics.current_price == 150.50
 
+    @pytest.mark.asyncio
+    async def test_scan_stock_overseas_empty_price(
+        self,
+        scanner: MarketScanner,
+        mock_overseas_broker: OverseasBroker,
+        context_store: ContextStore,
+    ) -> None:
+        """Test scanning overseas stock with empty price string (issue #49)."""
+        mock_overseas_broker.get_overseas_price.return_value = {
+            "output": {
+                "last": "",  # Empty string
+                "tvol": "",  # Empty string
+            }
+        }
+
+        market = MARKETS["US_NASDAQ"]
+        metrics = await scanner.scan_stock("AAPL", market)
+
+        assert metrics is not None
+        assert metrics.stock_code == "AAPL"
+        assert metrics.current_price == 0.0  # Should default to 0.0
+
     @pytest.mark.asyncio
     async def test_scan_stock_error_handling(
         self,
@@ -509,3 +532,45 @@ class TestMarketScanner:
         new_additions = [code for code in updated if code not in current_watchlist]
         assert len(new_additions) <= 1
         assert len(updated) == len(current_watchlist)
+
+    @pytest.mark.asyncio
+    async def test_scan_market_respects_concurrency_limit(
+        self,
+        mock_broker: KISBroker,
+        mock_overseas_broker: OverseasBroker,
+        volatility_analyzer: VolatilityAnalyzer,
+        context_store: ContextStore,
+    ) -> None:
+        """scan_market should limit concurrent scans to max_concurrent_scans."""
+        max_concurrent = 2
+        scanner = MarketScanner(
+            broker=mock_broker,
+            overseas_broker=mock_overseas_broker,
+            volatility_analyzer=volatility_analyzer,
+            context_store=context_store,
+            top_n=5,
+            max_concurrent_scans=max_concurrent,
+        )
+
+        # Track peak concurrency
+        active_count = 0
+        peak_count = 0
+
+        original_scan = scanner.scan_stock
+
+        async def tracking_scan(code: str, market: Any) -> VolatilityMetrics:
+            nonlocal active_count, peak_count
+            active_count += 1
+            peak_count = max(peak_count, active_count)
+            await asyncio.sleep(0.05)  # Simulate API call duration
+            active_count -= 1
+            return VolatilityMetrics(code, 50000, 500, 1.0, 1.0, 1.0, 1.0, 10.0, 50.0)
+
+        scanner.scan_stock = tracking_scan  # type: ignore[method-assign]
+
+        market = MARKETS["KR"]
+        stock_codes = ["001", "002", "003", "004", "005", "006"]
+
+        await scanner.scan_market(market, stock_codes)
+
+        assert peak_count <= max_concurrent