fix: keep embargo tied to accepted folds and enforce PR-comment decision logs

docs/ouroboros/50_tpm_control_protocol.md

@@ -149,6 +149,7 @@ TPM 티켓 운영 규칙:
 - TPM은 합의된 변경을 이슈로 등록하고 우선순위(`P0/P1/P2`)를 지정한다.
 - PR 본문에는 TPM이 지정한 우선순위와 범위가 그대로 반영되어야 한다.
 - 우선순위 변경은 TPM 제안 + Main Agent 승인으로만 가능하다.
+- PM/TPM/Dev/Reviewer/Verifier/Runtime Verifier는 주요 의사결정 시점마다 PR 코멘트를 남겨 결정 근거를 추적 가능 상태로 유지한다.
 
 브랜치 운영 규칙:
 - TPM은 각 티켓에 대해 `ticket temp branch -> program feature branch` PR 경로를 지정한다.

docs/ouroboros/60_repo_enforcement_checklist.md

@@ -50,6 +50,7 @@ Updated: 2026-02-26
 - PR 본문에 `REQ-*`, `TASK-*`, `TEST-*` 매핑 표 존재
 - `src/core/risk_manager.py` 변경 없음
 - 주요 의사결정 체크포인트(DCP-01~04) 중 해당 단계 Main Agent 확인 기록 존재
+- 주요 의사결정(리뷰 지적/수정 합의/검증 승인)에 대한 에이전트 PR 코멘트 존재
 - 티켓 PR의 base가 `main`이 아닌 program feature branch인지 확인
 
 자동 점검:

docs/workflow.md

@@ -22,6 +22,7 @@
 - Ticket-level development happens only on **ticket temp branches** cut from the program feature branch.
 - Ticket PR merges into program feature branch are allowed after verifier approval.
 - Until final user sign-off, `main` merge is prohibited.
+- 각 에이전트는 주요 의사결정(리뷰 지적, 수정 방향, 검증 승인)마다 PR 코멘트를 적극 작성해 의사결정 과정을 남긴다.
 
 ## Gitea CLI Formatting Troubleshooting
 

src/analysis/triple_barrier.py

@@ -0,0 +1,111 @@
+"""Triple barrier labeler utilities.
+
+Implements first-touch labeling with upper/lower/time barriers.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import Literal, Sequence
+
+
+TieBreakMode = Literal["stop_first", "take_first"]
+
+
+@dataclass(frozen=True)
+class TripleBarrierSpec:
+    take_profit_pct: float
+    stop_loss_pct: float
+    max_holding_bars: int
+    tie_break: TieBreakMode = "stop_first"
+
+
+@dataclass(frozen=True)
+class TripleBarrierLabel:
+    label: int  # +1 take-profit first, -1 stop-loss first, 0 timeout
+    touched: Literal["take_profit", "stop_loss", "time"]
+    touch_bar: int
+    entry_price: float
+    upper_barrier: float
+    lower_barrier: float
+
+
+def label_with_triple_barrier(
+    *,
+    highs: Sequence[float],
+    lows: Sequence[float],
+    closes: Sequence[float],
+    entry_index: int,
+    side: int,
+    spec: TripleBarrierSpec,
+) -> TripleBarrierLabel:
+    """Label one entry using triple-barrier first-touch rule.
+
+    Args:
+        highs/lows/closes: OHLC components with identical length.
+        entry_index: Entry bar index in the sequences.
+        side: +1 for long, -1 for short.
+        spec: Barrier specification.
+    """
+    if side not in {1, -1}:
+        raise ValueError("side must be +1 or -1")
+    if len(highs) != len(lows) or len(highs) != len(closes):
+        raise ValueError("highs, lows, closes lengths must match")
+    if entry_index < 0 or entry_index >= len(closes):
+        raise IndexError("entry_index out of range")
+    if spec.max_holding_bars <= 0:
+        raise ValueError("max_holding_bars must be positive")
+
+    entry_price = float(closes[entry_index])
+    if entry_price <= 0:
+        raise ValueError("entry price must be positive")
+
+    if side == 1:
+        upper = entry_price * (1.0 + spec.take_profit_pct)
+        lower = entry_price * (1.0 - spec.stop_loss_pct)
+    else:
+        # For short side, favorable move is down.
+        upper = entry_price * (1.0 + spec.stop_loss_pct)
+        lower = entry_price * (1.0 - spec.take_profit_pct)
+
+    last_index = min(len(closes) - 1, entry_index + spec.max_holding_bars)
+    for idx in range(entry_index + 1, last_index + 1):
+        h = float(highs[idx])
+        l = float(lows[idx])
+
+        up_touch = h >= upper
+        down_touch = l <= lower
+        if not up_touch and not down_touch:
+            continue
+
+        if up_touch and down_touch:
+            if spec.tie_break == "stop_first":
+                touched = "stop_loss"
+                label = -1
+            else:
+                touched = "take_profit"
+                label = 1
+        elif up_touch:
+            touched = "take_profit" if side == 1 else "stop_loss"
+            label = 1 if side == 1 else -1
+        else:
+            touched = "stop_loss" if side == 1 else "take_profit"
+            label = -1 if side == 1 else 1
+
+        return TripleBarrierLabel(
+            label=label,
+            touched=touched,
+            touch_bar=idx,
+            entry_price=entry_price,
+            upper_barrier=upper,
+            lower_barrier=lower,
+        )
+
+    return TripleBarrierLabel(
+        label=0,
+        touched="time",
+        touch_bar=last_index,
+        entry_price=entry_price,
+        upper_barrier=upper,
+        lower_barrier=lower,
+    )

src/analysis/walk_forward_split.py

@@ -0,0 +1,74 @@
+"""Walk-forward splitter with purge/embargo controls."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+
+@dataclass(frozen=True)
+class WalkForwardFold:
+    train_indices: list[int]
+    test_indices: list[int]
+
+    @property
+    def train_size(self) -> int:
+        return len(self.train_indices)
+
+    @property
+    def test_size(self) -> int:
+        return len(self.test_indices)
+
+
+def generate_walk_forward_splits(
+    *,
+    n_samples: int,
+    train_size: int,
+    test_size: int,
+    step_size: int | None = None,
+    purge_size: int = 0,
+    embargo_size: int = 0,
+    min_train_size: int = 1,
+) -> list[WalkForwardFold]:
+    """Generate chronological folds with purge/embargo leakage controls."""
+    if n_samples <= 0:
+        raise ValueError("n_samples must be positive")
+    if train_size <= 0 or test_size <= 0:
+        raise ValueError("train_size and test_size must be positive")
+    if purge_size < 0 or embargo_size < 0:
+        raise ValueError("purge_size and embargo_size must be >= 0")
+    if min_train_size <= 0:
+        raise ValueError("min_train_size must be positive")
+
+    step = step_size if step_size is not None else test_size
+    if step <= 0:
+        raise ValueError("step_size must be positive")
+
+    folds: list[WalkForwardFold] = []
+    prev_test_end: int | None = None
+    test_start = train_size + purge_size
+
+    while test_start + test_size <= n_samples:
+        test_end = test_start + test_size - 1
+        train_end = test_start - purge_size - 1
+        if train_end < 0:
+            break
+
+        train_start = max(0, train_end - train_size + 1)
+        train_indices = list(range(train_start, train_end + 1))
+
+        if prev_test_end is not None and embargo_size > 0:
+            emb_from = prev_test_end + 1
+            emb_to = prev_test_end + embargo_size
+            train_indices = [i for i in train_indices if i < emb_from or i > emb_to]
+
+        if len(train_indices) >= min_train_size:
+            folds.append(
+                WalkForwardFold(
+                    train_indices=train_indices,
+                    test_indices=list(range(test_start, test_end + 1)),
+                )
+            )
+            prev_test_end = test_end
+        test_start += step
+
+    return folds

src/config.py

@@ -59,6 +59,7 @@ class Settings(BaseSettings):
     # KIS VTS overseas balance API returns errors for most accounts.
     # This value is used as a fallback when the balance API returns 0 in paper mode.
     PAPER_OVERSEAS_CASH: float = Field(default=50000.0, ge=0.0)
+    USD_BUFFER_MIN: float = Field(default=1000.0, ge=0.0)
 
     # Trading frequency mode (daily = batch API calls, realtime = per-stock calls)
     TRADE_MODE: str = Field(default="daily", pattern="^(daily|realtime)$")

src/main.py

@@ -429,6 +429,26 @@ def _determine_order_quantity(
     return quantity
 
 
+def _should_block_overseas_buy_for_fx_buffer(
+    *,
+    market: MarketInfo,
+    action: str,
+    total_cash: float,
+    order_amount: float,
+    settings: Settings | None,
+) -> tuple[bool, float, float]:
+    if (
+        market.is_domestic
+        or not market.code.startswith("US")
+        or action != "BUY"
+        or settings is None
+    ):
+        return False, total_cash - order_amount, 0.0
+    remaining = total_cash - order_amount
+    required = settings.USD_BUFFER_MIN
+    return remaining < required, remaining, required
+
+
 async def build_overseas_symbol_universe(
     db_conn: Any,
     overseas_broker: OverseasBroker,
@@ -1292,6 +1312,24 @@ async def trading_cycle(
             )
             return
         order_amount = current_price * quantity
+        fx_blocked, remaining_cash, required_buffer = _should_block_overseas_buy_for_fx_buffer(
+            market=market,
+            action=decision.action,
+            total_cash=total_cash,
+            order_amount=order_amount,
+            settings=settings,
+        )
+        if fx_blocked:
+            logger.warning(
+                "Skip BUY %s (%s): FX buffer guard (remaining=%.2f, required=%.2f, cash=%.2f, order=%.2f)",
+                stock_code,
+                market.name,
+                remaining_cash,
+                required_buffer,
+                total_cash,
+                order_amount,
+            )
+            return
 
         # 4. Check BUY cooldown (set when a prior BUY failed due to insufficient balance)
         if decision.action == "BUY" and buy_cooldown is not None:
@@ -2360,6 +2398,24 @@ async def run_daily_session(
                     )
                     continue
                 order_amount = stock_data["current_price"] * quantity
+                fx_blocked, remaining_cash, required_buffer = _should_block_overseas_buy_for_fx_buffer(
+                    market=market,
+                    action=decision.action,
+                    total_cash=total_cash,
+                    order_amount=order_amount,
+                    settings=settings,
+                )
+                if fx_blocked:
+                    logger.warning(
+                        "Skip BUY %s (%s): FX buffer guard (remaining=%.2f, required=%.2f, cash=%.2f, order=%.2f)",
+                        stock_code,
+                        market.name,
+                        remaining_cash,
+                        required_buffer,
+                        total_cash,
+                        order_amount,
+                    )
+                    continue
 
                 # Check BUY cooldown (insufficient balance)
                 if decision.action == "BUY":

tests/test_main.py

@@ -15,6 +15,7 @@ from src.evolution.scorecard import DailyScorecard
 from src.logging.decision_logger import DecisionLogger
 from src.main import (
     KILL_SWITCH,
+    _should_block_overseas_buy_for_fx_buffer,
     _trigger_emergency_kill_switch,
     _apply_dashboard_flag,
     _determine_order_quantity,
@@ -3690,6 +3691,81 @@ class TestOverseasBrokerIntegration:
         # DB도 브로커도 보유 없음 → BUY 주문이 실행되어야 함 (회귀 테스트)
         overseas_broker.send_overseas_order.assert_called_once()
 
+    @pytest.mark.asyncio
+    async def test_overseas_buy_blocked_by_usd_buffer_guard(self) -> None:
+        """Overseas BUY must be blocked when USD buffer would be breached."""
+        db_conn = init_db(":memory:")
+
+        overseas_broker = MagicMock()
+        overseas_broker.get_overseas_price = AsyncMock(
+            return_value={"output": {"last": "182.50"}}
+        )
+        overseas_broker.get_overseas_balance = AsyncMock(
+            return_value={
+                "output1": [],
+                "output2": [
+                    {
+                        "frcr_evlu_tota": "50000.00",
+                        "frcr_buy_amt_smtl": "0.00",
+                    }
+                ],
+            }
+        )
+        overseas_broker.get_overseas_buying_power = AsyncMock(
+            return_value={"output": {"ovrs_ord_psbl_amt": "50000.00"}}
+        )
+        overseas_broker.send_overseas_order = AsyncMock(return_value={"msg1": "주문접수"})
+
+        engine = MagicMock(spec=ScenarioEngine)
+        engine.evaluate = MagicMock(return_value=_make_buy_match("AAPL"))
+
+        market = MagicMock()
+        market.name = "NASDAQ"
+        market.code = "US_NASDAQ"
+        market.exchange_code = "NASD"
+        market.is_domestic = False
+
+        telegram = MagicMock()
+        telegram.notify_trade_execution = AsyncMock()
+        telegram.notify_fat_finger = AsyncMock()
+        telegram.notify_circuit_breaker = AsyncMock()
+        telegram.notify_scenario_matched = AsyncMock()
+
+        decision_logger = MagicMock()
+        decision_logger.log_decision = MagicMock(return_value="decision-id")
+
+        settings = MagicMock()
+        settings.POSITION_SIZING_ENABLED = False
+        settings.CONFIDENCE_THRESHOLD = 80
+        settings.USD_BUFFER_MIN = 49900.0
+        settings.MODE = "paper"
+        settings.PAPER_OVERSEAS_CASH = 50000.0
+
+        await trading_cycle(
+            broker=MagicMock(),
+            overseas_broker=overseas_broker,
+            scenario_engine=engine,
+            playbook=_make_playbook(market="US"),
+            risk=MagicMock(),
+            db_conn=db_conn,
+            decision_logger=decision_logger,
+            context_store=MagicMock(
+                get_latest_timeframe=MagicMock(return_value=None),
+                set_context=MagicMock(),
+            ),
+            criticality_assessor=MagicMock(
+                assess_market_conditions=MagicMock(return_value=MagicMock(value="NORMAL")),
+                get_timeout=MagicMock(return_value=5.0),
+            ),
+            telegram=telegram,
+            market=market,
+            stock_code="AAPL",
+            scan_candidates={},
+            settings=settings,
+        )
+
+        overseas_broker.send_overseas_order.assert_not_called()
+
 
 # ---------------------------------------------------------------------------
 # _retry_connection — unit tests (issue #209)
@@ -3723,7 +3799,6 @@ class TestRetryConnection:
         with patch("src.main.asyncio.sleep") as mock_sleep:
             mock_sleep.return_value = None
             result = await _retry_connection(flaky, label="flaky")
-
         assert result == "ok"
         assert call_count == 2
         mock_sleep.assert_called_once()
@@ -3778,6 +3853,48 @@ class TestRetryConnection:
         assert call_count == 1  # No retry for non-ConnectionError
 
 
+def test_fx_buffer_guard_applies_only_to_us_and_respects_boundary() -> None:
+    settings = MagicMock()
+    settings.USD_BUFFER_MIN = 1000.0
+
+    us_market = MagicMock()
+    us_market.is_domestic = False
+    us_market.code = "US_NASDAQ"
+
+    blocked, remaining, required = _should_block_overseas_buy_for_fx_buffer(
+        market=us_market,
+        action="BUY",
+        total_cash=5000.0,
+        order_amount=4001.0,
+        settings=settings,
+    )
+    assert blocked
+    assert remaining == 999.0
+    assert required == 1000.0
+
+    blocked_eq, _, _ = _should_block_overseas_buy_for_fx_buffer(
+        market=us_market,
+        action="BUY",
+        total_cash=5000.0,
+        order_amount=4000.0,
+        settings=settings,
+    )
+    assert not blocked_eq
+
+    jp_market = MagicMock()
+    jp_market.is_domestic = False
+    jp_market.code = "JP"
+    blocked_jp, _, required_jp = _should_block_overseas_buy_for_fx_buffer(
+        market=jp_market,
+        action="BUY",
+        total_cash=5000.0,
+        order_amount=4500.0,
+        settings=settings,
+    )
+    assert not blocked_jp
+    assert required_jp == 0.0
+
+
 # run_daily_session — daily CB baseline (daily_start_eval) tests (issue #207)
 # ---------------------------------------------------------------------------
 

tests/test_triple_barrier.py

@@ -0,0 +1,131 @@
+from __future__ import annotations
+
+from src.analysis.triple_barrier import TripleBarrierSpec, label_with_triple_barrier
+
+
+def test_long_take_profit_first() -> None:
+    highs = [100, 101, 103]
+    lows = [100, 99.6, 100]
+    closes = [100, 100, 102]
+    spec = TripleBarrierSpec(take_profit_pct=0.02, stop_loss_pct=0.01, max_holding_bars=3)
+    out = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=1,
+        spec=spec,
+    )
+    assert out.label == 1
+    assert out.touched == "take_profit"
+    assert out.touch_bar == 2
+
+
+def test_long_stop_loss_first() -> None:
+    highs = [100, 100.5, 101]
+    lows = [100, 98.8, 99]
+    closes = [100, 99.5, 100]
+    spec = TripleBarrierSpec(take_profit_pct=0.02, stop_loss_pct=0.01, max_holding_bars=3)
+    out = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=1,
+        spec=spec,
+    )
+    assert out.label == -1
+    assert out.touched == "stop_loss"
+    assert out.touch_bar == 1
+
+
+def test_time_barrier_timeout() -> None:
+    highs = [100, 100.8, 100.7]
+    lows = [100, 99.3, 99.4]
+    closes = [100, 100, 100]
+    spec = TripleBarrierSpec(take_profit_pct=0.02, stop_loss_pct=0.02, max_holding_bars=2)
+    out = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=1,
+        spec=spec,
+    )
+    assert out.label == 0
+    assert out.touched == "time"
+    assert out.touch_bar == 2
+
+
+def test_tie_break_stop_first_default() -> None:
+    highs = [100, 102.1]
+    lows = [100, 98.9]
+    closes = [100, 100]
+    spec = TripleBarrierSpec(take_profit_pct=0.02, stop_loss_pct=0.01, max_holding_bars=1)
+    out = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=1,
+        spec=spec,
+    )
+    assert out.label == -1
+    assert out.touched == "stop_loss"
+
+
+def test_short_side_inverts_barrier_semantics() -> None:
+    highs = [100, 100.5, 101.2]
+    lows = [100, 97.8, 98.0]
+    closes = [100, 99, 99]
+    spec = TripleBarrierSpec(take_profit_pct=0.02, stop_loss_pct=0.01, max_holding_bars=3)
+    out = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=-1,
+        spec=spec,
+    )
+    assert out.label == 1
+    assert out.touched == "take_profit"
+
+
+def test_short_tie_break_modes() -> None:
+    highs = [100, 101.1]
+    lows = [100, 97.9]
+    closes = [100, 100]
+
+    stop_first = TripleBarrierSpec(
+        take_profit_pct=0.02,
+        stop_loss_pct=0.01,
+        max_holding_bars=1,
+        tie_break="stop_first",
+    )
+    out_stop = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=-1,
+        spec=stop_first,
+    )
+    assert out_stop.label == -1
+    assert out_stop.touched == "stop_loss"
+
+    take_first = TripleBarrierSpec(
+        take_profit_pct=0.02,
+        stop_loss_pct=0.01,
+        max_holding_bars=1,
+        tie_break="take_first",
+    )
+    out_take = label_with_triple_barrier(
+        highs=highs,
+        lows=lows,
+        closes=closes,
+        entry_index=0,
+        side=-1,
+        spec=take_first,
+    )
+    assert out_take.label == 1
+    assert out_take.touched == "take_profit"

tests/test_walk_forward_split.py

@@ -0,0 +1,92 @@
+from __future__ import annotations
+
+import pytest
+
+from src.analysis.walk_forward_split import generate_walk_forward_splits
+
+
+def test_generates_sequential_folds() -> None:
+    folds = generate_walk_forward_splits(
+        n_samples=30,
+        train_size=10,
+        test_size=5,
+    )
+    assert len(folds) == 4
+    assert folds[0].train_indices == list(range(0, 10))
+    assert folds[0].test_indices == list(range(10, 15))
+    assert folds[1].train_indices == list(range(5, 15))
+    assert folds[1].test_indices == list(range(15, 20))
+
+
+def test_purge_removes_boundary_samples_before_test() -> None:
+    folds = generate_walk_forward_splits(
+        n_samples=25,
+        train_size=8,
+        test_size=4,
+        purge_size=2,
+    )
+    first = folds[0]
+    # test starts at 10, purge=2 => train end must be 7
+    assert first.train_indices == list(range(0, 8))
+    assert first.test_indices == list(range(10, 14))
+
+
+def test_embargo_excludes_post_test_samples_from_next_train() -> None:
+    folds = generate_walk_forward_splits(
+        n_samples=45,
+        train_size=15,
+        test_size=5,
+        step_size=10,
+        embargo_size=3,
+    )
+    assert len(folds) >= 2
+    # Fold1 test: 15..19, next fold train window: 10..24.
+    # embargo_size=3 should remove 20,21,22 from fold2 train.
+    second_train = folds[1].train_indices
+    assert 20 not in second_train
+    assert 21 not in second_train
+    assert 22 not in second_train
+    assert 23 in second_train
+
+
+def test_respects_min_train_size_and_returns_empty_when_impossible() -> None:
+    folds = generate_walk_forward_splits(
+        n_samples=15,
+        train_size=5,
+        test_size=5,
+        min_train_size=6,
+    )
+    assert folds == []
+
+
+def test_embargo_uses_last_accepted_fold_when_intermediate_fold_skips() -> None:
+    folds = generate_walk_forward_splits(
+        n_samples=30,
+        train_size=5,
+        test_size=3,
+        step_size=5,
+        embargo_size=1,
+        min_train_size=5,
+    )
+    # 1st fold accepted, 2nd skipped by min_train_size, subsequent folds still generated.
+    assert len(folds) == 3
+    assert folds[0].test_indices == [5, 6, 7]
+    assert folds[1].test_indices == [15, 16, 17]
+    assert folds[2].test_indices == [25, 26, 27]
+
+
+@pytest.mark.parametrize(
+    ("n_samples", "train_size", "test_size"),
+    [
+        (0, 10, 2),
+        (10, 0, 2),
+        (10, 5, 0),
+    ],
+)
+def test_invalid_args_raise(n_samples: int, train_size: int, test_size: int) -> None:
+    with pytest.raises(ValueError):
+        generate_walk_forward_splits(
+            n_samples=n_samples,
+            train_size=train_size,
+            test_size=test_size,
+        )