feat: 리밸런싱 엔진 V1 + GAS 버그 수정 (2026-06-13)

주요 변경: - tools/build_rebalance_engine_v1.py: REBALANCE_ENGINE_V1 신규 * account_snapshot 직접 합산(_build_snap_position_map) → 소수주 분리 행 병합 * 레짐 소스 macro.REGIME_PRELIM 최우선 (GAS 와 동일) - src/gas_adapter_parts/gdf_06_rebalance.gs: runRebalanceSheet_() 신규 * Logger.log / getSpreadsheet_() 로 run_all 연동 수정 - src/gas_adapter_parts/gdc_01_fetch_fundamentals.gs * _mergePositionRecord_(): 소수주 중복 행 합산 신규 * parseInt → parseFloat (qty, availQty) - src/gas_adapter_parts/gdf_01_price_metrics.gs * 미보유 종목 SELL_READY → WATCH_EXIT_SIGNAL - spec/41_release_dag.yaml: build_rebalance_sheet 노드 추가 (step_count 63) - spec/51_formula_lifecycle_registry.yaml: REBALANCE_ENGINE_V1 등록 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-13 13:20:14 +09:00
commit ee3e799de1
1474 changed files with 176087 additions and 0 deletions
@@ -0,0 +1,636 @@
+"""build_rebalance_engine_v1.py — REBALANCE_ENGINE_V1
+
+리밸런싱 엔진: bucket drift 측정 → 레짐 적응 밴드 → 비용효익 게이트 → 3단계 분할 실행 계획.
+ABS_FLOOR / TIME_STOP 강제 매도 신호 통합.
+
+P3: 레짐 적응 밴드 (Risk-On ±15%p, Neutral ±5%p, Risk-Off +2%p/−10%p)
+P4: 비용효익 게이트 (|drift| − TX_COST_ROUNDTRIP > 0.5%p)
+P5: 3단계 분할 실행 (Stage1 30%, Stage2 30%, Stage3 40%)
+P6: ABS_FLOOR / TIME_STOP 강제 매도 통합
+
+입력: GatherTradingData.json + Temp/computed_harness_v1.json (optional)
+출력: Temp/rebalance_engine_v1.json
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import math
+from pathlib import Path
+from typing import Any
+
+ROOT = Path(__file__).resolve().parents[1]
+TEMP = ROOT / "Temp"
+DEFAULT_JSON = ROOT / "GatherTradingData.json"
+DEFAULT_OUT  = TEMP / "rebalance_engine_v1.json"
+FORMULA_ID   = "REBALANCE_ENGINE_V1"
+
+# ── 버킷 목표 (gdf_01_price_metrics.gs THRESHOLDS 와 동기화) ──────────────────
+BUCKET_CONFIG: dict[str, dict] = {
+    "Core":      {"target": 66.0, "min": 60.0, "max": 72.0},
+    "Satellite": {"target": 17.5, "min": 10.0, "max": 25.0},
+    "Cash":      {"target": 16.5, "min": 10.0, "max": 22.0},
+}
+
+# 직접 코어 주도주 (gdc_02_account_satellite.gs isCoreLeader 기준)
+CORE_TICKERS_BASE: set[str] = {"005930", "000660", "000270"}
+
+# ── 레짐 적응 밴드 (P3) ──────────────────────────────────────────────────────
+# expand = 목표 대비 상단 확장%p, contract = 목표 대비 하단 축소%p (절대값)
+REGIME_BANDS: dict[str, dict] = {
+    "RISK_ON":            {"label": "RISK_ON ±15%p",    "expand": 15.0, "contract": 15.0},
+    "SECULAR_LEADER_RISK_ON": {"label": "RISK_ON ±15%p", "expand": 15.0, "contract": 15.0},
+    "NEUTRAL":            {"label": "NEUTRAL ±5%p",     "expand":  5.0, "contract":  5.0},
+    "RISK_OFF_CANDIDATE": {"label": "RISK_OFF_CANDIDATE +2/−10%p", "expand": 2.0, "contract": 10.0},
+    "RISK_OFF":           {"label": "RISK_OFF +2/−10%p", "expand":  2.0, "contract": 10.0},
+    "EVENT_SHOCK":        {"label": "RISK_OFF +2/−10%p", "expand":  2.0, "contract": 10.0},
+    "_DEFAULT":           {"label": "NEUTRAL ±5%p",     "expand":  5.0, "contract":  5.0},
+}
+
+# ── 비용효익 게이트 (P4) ─────────────────────────────────────────────────────
+TX_COST_ONE_WAY    = 0.0035   # 0.35% (수수료 + 세금)
+TX_COST_ROUNDTRIP  = TX_COST_ONE_WAY * 2   # 0.70%
+COST_BENEFIT_THRESHOLD = 0.005             # 0.50%p — 이보다 클 때만 주문 생성
+MIN_ACTIONABLE_DRIFT_PCT = (TX_COST_ROUNDTRIP + COST_BENEFIT_THRESHOLD) * 100  # 1.20%p
+
+# ── 3단계 분할 실행 비율 (P5) ────────────────────────────────────────────────
+STAGE_RATIOS = [0.30, 0.30, 0.40]  # Stage1, Stage2, Stage3
+
+# 매도 지정가 할인율 (직전 종가의 0.2% 아래로 지정)
+LIMIT_PRICE_DISCOUNT = 0.002
+
+
+# ── 유틸 ─────────────────────────────────────────────────────────────────────
+def _load(path: Path) -> Any:
+    if not path.exists():
+        return {}
+    try:
+        return json.loads(path.read_text(encoding="utf-8"))
+    except Exception:
+        return {}
+
+
+def _f(v: Any, default: float = 0.0) -> float:
+    try:
+        return float(v)
+    except (TypeError, ValueError):
+        return default
+
+
+def _i(v: Any, default: int = 0) -> int:
+    try:
+        return int(float(v))
+    except (TypeError, ValueError):
+        return default
+
+
+def _s(v: Any, default: str = "") -> str:
+    return str(v) if v is not None else default
+
+
+def _round2(v: float) -> float:
+    return round(v, 2)
+
+
+def _band_for_regime(regime: str) -> dict:
+    return REGIME_BANDS.get(regime.upper(), REGIME_BANDS["_DEFAULT"])
+
+
+def _detect_force_signal(row: dict) -> str:
+    """Sell_Reason 또는 Final_Action에서 ABS_FLOOR / TIME_STOP 강제 신호 추출."""
+    combined = " ".join([
+        _s(row.get("Sell_Reason")),
+        _s(row.get("Final_Action")),
+        _s(row.get("Sell_Action")),
+    ]).upper()
+    if "ABS_FLOOR" in combined:
+        return "ABS_FLOOR"
+    if "TIME_STOP" in combined or "TIME_EXIT" in combined or "TIME_TRIM" in combined:
+        return "TIME_STOP"
+    return ""
+
+
+def _stage_qty_split(total_qty: int) -> list[int]:
+    """3단계 수량 분할: [30%, 30%, 40%]. 각 stage는 최소 1주."""
+    if total_qty <= 0:
+        return [0, 0, 0]
+    if total_qty < 3:
+        return [total_qty, 0, 0]
+    s1 = max(1, math.floor(total_qty * STAGE_RATIOS[0]))
+    s2 = max(1, math.floor(total_qty * STAGE_RATIOS[1]))
+    s3 = total_qty - s1 - s2
+    if s3 < 0:
+        s2 = total_qty - s1
+        s3 = 0
+    return [s1, s2, max(0, s3)]
+
+
+def _extract_trim_guidance(hc: dict) -> dict:
+    tg = hc.get("regime_trim_guidance_json") or {}
+    if isinstance(tg, str):
+        try:
+            tg = json.loads(tg)
+        except Exception:
+            tg = {}
+    return tg if isinstance(tg, dict) else {}
+
+
+def _extract_df_rows(payload: Any) -> list[dict]:
+    """GatherTradingData.json 에서 data_feed 행 배열 추출."""
+    if not isinstance(payload, dict):
+        return []
+    data = payload.get("data") or {}
+    df = data.get("data_feed") or []
+    if not isinstance(df, list):
+        return []
+    return [r for r in df if isinstance(r, dict)]
+
+
+def _build_snap_position_map(payload: Any, total_krw: float) -> dict[str, dict]:
+    """
+    account_snapshot 행 배열을 티커별로 합산한 맵 반환.
+    동일 티커 복수 행(소수 분리 계좌 포함) → qty·MV·원가 합산.
+    GAS readAccountSnapshotMap_ 의 _mergePositionRecord_ 와 동일 로직.
+
+    반환: { ticker: { qty, market_value, weight_pct, position_type, name } }
+    """
+    data = payload.get("data") or {} if isinstance(payload, dict) else {}
+    snap = data.get("account_snapshot") or []
+    if not isinstance(snap, list):
+        return {}
+
+    merged: dict[str, dict] = {}
+    for row in snap:
+        if not isinstance(row, dict):
+            continue
+        if _s(row.get("parse_status")).strip().upper() != "CAPTURE_READ_OK":
+            continue
+        # 연금저축 계좌는 주문 불가 — 건너뜀
+        if "연금저축" in _s(row.get("account_type")):
+            continue
+        ticker = _s(row.get("ticker")).strip()
+        if not ticker:
+            continue
+        # US 주식(알파벳만): Naver 가격 없음 → weight 0 → 건너뜀
+        if ticker.replace(".", "").isalpha():
+            continue
+        qty = _f(row.get("holding_quantity"))
+        if qty <= 0:
+            continue
+        mv      = _f(row.get("market_value"))
+        cost    = _f(row.get("total_cost"))
+        pl      = _f(row.get("profit_loss"))
+        name    = _s(row.get("name"))
+        pt      = _s(row.get("position_type")).lower()
+
+        if ticker not in merged:
+            merged[ticker] = {
+                "qty": qty, "market_value": mv,
+                "total_cost": cost, "profit_loss": pl,
+                "name": name, "position_type": pt,
+            }
+        else:
+            ex = merged[ticker]
+            ex["qty"]          += qty
+            ex["market_value"] += mv
+            ex["total_cost"]   += cost
+            ex["profit_loss"]  += pl
+            # "(소수)" 없는 이름 우선
+            if "소수" in ex["name"] and "소수" not in name:
+                ex["name"] = name
+            if not ex["position_type"] and pt:
+                ex["position_type"] = pt
+
+    # weight_pct 계산
+    for v in merged.values():
+        v["weight_pct"] = _round2(v["market_value"] / total_krw * 100) if total_krw > 0 else 0.0
+
+    return merged
+
+
+def _extract_portfolio_totals(payload: Any) -> tuple[float, float]:
+    """
+    (total_portfolio_krw, cash_krw) 추출 우선순위:
+    1. settings.total_asset_krw / settings.settlement_cash_d2_krw  (가장 정확)
+    2. account_snapshot cash row 의 settlement_cash_d2
+    3. holdings 합산 (fallback)
+    """
+    data     = payload.get("data") or {} if isinstance(payload, dict) else {}
+    settings = data.get("settings") or {}
+
+    # settings 에서 읽기 (ChatGPT → settings 탭에 기록된 총자산)
+    total_krw = _f(settings.get("total_asset_krw"))
+    cash_krw  = _f(settings.get("settlement_cash_d2_krw"))
+
+    if total_krw <= 0:
+        # account_snapshot cash row fallback
+        snap = data.get("account_snapshot") or []
+        if isinstance(snap, list):
+            for row in snap:
+                if isinstance(row, dict) and _s(row.get("position_type")).lower() == "cash":
+                    cash_krw  = _f(row.get("settlement_cash_d2") or row.get("immediate_cash"))
+                    break
+        # holdings 합산 (최후 fallback)
+        df_rows = _extract_df_rows(payload)
+        holdings_total = sum(_f(r.get("Account_Market_Value")) for r in df_rows)
+        total_krw = holdings_total + cash_krw
+
+    return total_krw, cash_krw
+
+
+_REGIME_NORMALIZE: dict[str, str] = {
+    "BREAKDOWN":              "RISK_OFF",
+    "EVENT_SHOCK":            "RISK_OFF",
+    "RISK_OFF_CANDIDATE":     "RISK_OFF_CANDIDATE",
+    "RISK_OFF":               "RISK_OFF",
+    "NEUTRAL":                "NEUTRAL",
+    "RISK_ON":                "RISK_ON",
+    "SECULAR_LEADER_RISK_ON": "RISK_ON",
+}
+
+
+def _normalize_regime(raw: str) -> str:
+    return _REGIME_NORMALIZE.get(raw.upper(), raw.upper())
+
+
+def _extract_regime(harness: Any, payload: Any) -> str:
+    """
+    레짐 추출 우선순위:
+    1. payload.data.macro[Symbol=REGIME_PRELIM].Close  — GAS가 실제 사용하는 소스(최신)
+    2. payload.data.settings.prev_market_regime         — ChatGPT 설정값
+    3. payload.data._harness_context.market_regime_state — harness (최대 수일 stale 가능)
+    4. payload.data._harness_context.regime_transition_json.current_regime
+    5. computed_harness_v1.json 의 market_regime 키
+    6. 기본값 "NEUTRAL"
+
+    BREAKDOWN → RISK_OFF 정규화.
+    GAS gdc_01_fetch_fundamentals.gs:2237 과 동일한 소스 사용.
+    """
+    data = payload.get("data") or {} if isinstance(payload, dict) else {}
+
+    # 1. macro 시트 REGIME_PRELIM 행 (GAS 와 동일한 소스 — 항상 최신)
+    macro_rows = data.get("macro") or []
+    if isinstance(macro_rows, list):
+        for r in macro_rows:
+            if isinstance(r, dict) and _s(r.get("Symbol")).strip() == "REGIME_PRELIM":
+                v = _s(r.get("Close")).strip()
+                if v:
+                    return _normalize_regime(v)
+
+    # 2. settings.prev_market_regime
+    settings = data.get("settings") or {}
+    v = settings.get("prev_market_regime") or settings.get("REGIME_PRELIM")
+    if v and isinstance(v, str) and v.strip():
+        return _normalize_regime(v.strip())
+
+    # 3-4. _harness_context (수일 stale 가능 — fallback 전용)
+    hc = data.get("_harness_context") or {}
+    if isinstance(hc, dict):
+        v = hc.get("market_regime_state")
+        if v and isinstance(v, str):
+            return _normalize_regime(v)
+        rt = hc.get("regime_transition_json") or {}
+        if isinstance(rt, str):
+            try:
+                rt = json.loads(rt)
+            except Exception:
+                rt = {}
+        if isinstance(rt, dict):
+            v = rt.get("current_regime")
+            if v and isinstance(v, str):
+                return _normalize_regime(v)
+
+    # 5. computed_harness_v1.json
+    if isinstance(harness, dict):
+        for key in ("market_regime", "Market_Regime", "regime"):
+            v = harness.get(key)
+            if v and isinstance(v, str):
+                return _normalize_regime(v)
+
+    return "NEUTRAL"
+
+
+def _assign_bucket(ticker: str, row: dict, snap_position_type: str = "") -> str:
+    """
+    우선순위: snap_position_type (account_snapshot) > row.position_type > CORE_TICKERS_BASE.
+    ETF/펀드 코드(알파벳 포함)는 Satellite.
+    """
+    for pt_raw in (snap_position_type, _s(row.get("position_type") or row.get("Position_Type"))):
+        pt = pt_raw.lower().strip()
+        if pt == "core":
+            return "Core"
+        if pt == "satellite":
+            return "Satellite"
+    return "Core" if ticker in CORE_TICKERS_BASE else "Satellite"
+
+
+def _compute_limit_price(close: float, action: str) -> float:
+    """매도 지정가: 종가의 (1 - LIMIT_PRICE_DISCOUNT). 매수는 종가 그대로."""
+    if close <= 0:
+        return 0.0
+    if action == "SELL":
+        return round(close * (1 - LIMIT_PRICE_DISCOUNT))
+    return round(close)
+
+
+# ── 메인 엔진 ─────────────────────────────────────────────────────────────────
+def main() -> int:
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--json",    default=str(DEFAULT_JSON))
+    ap.add_argument("--harness", default=str(TEMP / "computed_harness_v1.json"))
+    ap.add_argument("--out",     default=str(DEFAULT_OUT))
+    args = ap.parse_args()
+
+    json_path    = Path(args.json)
+    harness_path = Path(args.harness)
+    out_path     = Path(args.out)
+    for p in (json_path, harness_path, out_path):
+        if not p.is_absolute():
+            p = ROOT / p
+
+    payload = _load(json_path)
+    harness = _load(harness_path)
+
+    df_rows = _extract_df_rows(payload)
+    regime  = _extract_regime(harness, payload)
+    band    = _band_for_regime(regime)
+
+    total_portfolio_krw, cash_krw = _extract_portfolio_totals(payload)
+
+    # harness_context 에서 현금 바닥 목표 읽기 (BUCKET_CONFIG.Cash.min 보다 우선)
+    hc_data = (payload.get("data") or {}).get("_harness_context") or {}
+    cash_floor_pct = float(hc_data.get("cash_floor_min_pct") or BUCKET_CONFIG["Cash"]["min"])
+    cash_shortfall_krw = float(hc_data.get("cash_shortfall_min_krw") or 0)
+    # Cash 버킷 최소값을 harness 현금바닥으로 override
+    BUCKET_CONFIG["Cash"]["min"] = max(BUCKET_CONFIG["Cash"]["min"], cash_floor_pct)
+
+    # ── 1. account_snapshot 합산 맵 (Weight_Pct 재계산 기준) ─────────────────
+    # GAS readAccountSnapshotMap_ 에 소수 분리 행 덮어쓰기 버그가 있는 경우를 대비,
+    # Python 엔진은 account_snapshot 을 직접 합산하여 올바른 weight_pct 를 사용한다.
+    snap_map = _build_snap_position_map(payload, total_portfolio_krw)
+
+    # ── 2. 보유 종목 필터링 (snap_map 우선, data_feed 보조) ───────────────────
+    holdings = []
+    seen_tickers: set[str] = set()
+
+    # snap_map 에 있는 종목이 authoritative (GAS 소수 분리 버그 무관)
+    df_row_map = {_s(r.get("Ticker")).strip(): r for r in df_rows if _s(r.get("Ticker")).strip()}
+
+    for ticker, sp in snap_map.items():
+        if sp["weight_pct"] <= 0:
+            continue
+        seen_tickers.add(ticker)
+        df_row = df_row_map.get(ticker, {})
+        holdings.append({
+            "ticker":       ticker,
+            "name":         sp["name"] or _s(df_row.get("Name")),
+            "bucket":       _assign_bucket(ticker, df_row, sp.get("position_type", "")),
+            "weight_pct":   sp["weight_pct"],
+            "acct_mv_krw":  sp["market_value"],
+            "holding_qty":  int(round(sp["qty"])),  # 주문 수량은 정수
+            "close":        _f(df_row.get("Close")),
+            "final_action": _s(df_row.get("Final_Action")),
+            "sell_reason":  _s(df_row.get("Sell_Reason")),
+            "force_signal": _detect_force_signal(df_row),
+        })
+
+    # data_feed 에만 있는 보유 종목 보완 (snap 에 없는 경우 — 비정상이지만 방어)
+    for row in df_rows:
+        ticker = _s(row.get("Ticker")).strip()
+        if not ticker or ticker in seen_tickers:
+            continue
+        weight_pct = _f(row.get("Weight_Pct"))
+        acct_mv    = _f(row.get("Account_Market_Value"))
+        if weight_pct <= 0 and acct_mv <= 0:
+            continue
+        holdings.append({
+            "ticker":       ticker,
+            "name":         _s(row.get("Name")),
+            "bucket":       _assign_bucket(ticker, row),
+            "weight_pct":   weight_pct,
+            "acct_mv_krw":  acct_mv,
+            "holding_qty":  _i(row.get("Account_Holding_Qty")),
+            "close":        _f(row.get("Close")),
+            "final_action": _s(row.get("Final_Action")),
+            "sell_reason":  _s(row.get("Sell_Reason")),
+            "force_signal": _detect_force_signal(row),
+        })
+
+    # ── 3. 버킷별 현재 비중 집계 ─────────────────────────────────────────────
+    # weight_pct 는 account_snapshot MV / total_asset_krw 기반 (재계산 완료).
+    # Cash: settings.settlement_cash_d2_krw / total_asset_krw 로 직접 계산 (정확).
+    core_pct = sum(h["weight_pct"] for h in holdings if h["bucket"] == "Core")
+    sat_pct  = sum(h["weight_pct"] for h in holdings if h["bucket"] == "Satellite")
+    if total_portfolio_krw > 0 and cash_krw > 0:
+        cash_pct = _round2(cash_krw / total_portfolio_krw * 100)
+    else:
+        cash_pct = max(0.0, 100.0 - core_pct - sat_pct)
+
+    bucket_current = {"Core": core_pct, "Satellite": sat_pct, "Cash": cash_pct}
+
+    # ── 3. 버킷 drift + 밴드 계산 ────────────────────────────────────────────
+    bucket_rows: list[dict] = []
+    rebalance_needed = False
+
+    for bname, bcfg in BUCKET_CONFIG.items():
+        target     = bcfg["target"]
+        current    = _round2(bucket_current[bname])
+        drift      = _round2(current - target)
+        band_min   = _round2(target - band["contract"])
+        band_max   = _round2(target + band["expand"])
+
+        if current < band_min:
+            drift_status = "BREACH_LOW"
+            rebalance_needed = True
+        elif current > band_max:
+            drift_status = "BREACH_HIGH"
+            rebalance_needed = True
+        elif abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT / 2:
+            drift_status = "WARN"
+        else:
+            drift_status = "NORMAL"
+
+        bucket_rows.append({
+            "bucket":        bname,
+            "target_pct":    target,
+            "current_pct":   current,
+            "drift_pct":     drift,
+            "band_min":      band_min,
+            "band_max":      band_max,
+            "regime_band":   band["label"],
+            "drift_status":  drift_status,
+        })
+
+    # ── 4. 종목별 분석 ───────────────────────────────────────────────────────
+    # 버킷 내 equal-weight target
+    bucket_ticker_count: dict[str, int] = {}
+    for h in holdings:
+        bucket_ticker_count[h["bucket"]] = bucket_ticker_count.get(h["bucket"], 0) + 1
+
+    ticker_rows: list[dict] = []
+    for h in holdings:
+        bname       = h["bucket"]
+        bcfg        = BUCKET_CONFIG.get(bname, BUCKET_CONFIG["Satellite"])
+        n_tickers   = bucket_ticker_count.get(bname, 1)
+        target_pct  = _round2(bcfg["target"] / n_tickers)
+        current_pct = _round2(h["weight_pct"])
+        drift       = _round2(current_pct - target_pct)
+        band_min    = _round2(target_pct - band["contract"])
+        band_max    = _round2(target_pct + band["expand"])
+
+        # 드리프트 상태
+        force = h["force_signal"]
+        if force:
+            drift_status = "FORCE_" + force
+            action       = "SELL"
+            gate_status  = "FORCE_OVERRIDE"
+        elif current_pct > band_max:
+            drift_status = "BREACH_HIGH"
+            action       = "SELL" if abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT else "WATCH"
+            gate_status  = "PASS" if abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT else "BLOCKED_BY_COST"
+        elif current_pct < band_min:
+            drift_status = "BREACH_LOW"
+            action       = "BUY" if abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT else "WATCH"
+            gate_status  = "PASS" if abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT else "BLOCKED_BY_COST"
+        elif abs(drift) >= MIN_ACTIONABLE_DRIFT_PCT / 2:
+            drift_status = "WARN"
+            gate_status  = "BLOCKED_BY_COST"
+            action       = "WATCH"
+        else:
+            drift_status = "NORMAL"
+            gate_status  = "BLOCKED_BY_COST"
+            action       = "HOLD"
+
+        # 3단계 분할 수량 계산 (P5) — SELL/BUY 액션에만 적용
+        stage_qtys: list[int] = [0, 0, 0]
+        limit_prices: list[float] = [0.0, 0.0, 0.0]
+        trade_value_krw = 0.0
+        cost_est_krw    = 0.0
+        net_benefit_pct = 0.0
+
+        if action in ("SELL", "BUY") and h["holding_qty"] > 0 and h["close"] > 0:
+            # 매도/매수 수량: 드리프트 해소에 필요한 수량
+            # |drift|%p ÷ current_pct * 보유수량 근사
+            if action == "SELL" and current_pct > 0:
+                adjust_ratio  = min(abs(drift) / current_pct, 1.0)
+                adjust_qty    = max(1, round(h["holding_qty"] * adjust_ratio))
+            else:
+                adjust_qty = max(1, round(h["holding_qty"] * 0.10))  # BUY: 10% 추가 근사
+
+            stage_qtys   = _stage_qty_split(adjust_qty)
+            limit_price  = _compute_limit_price(h["close"], action)
+            limit_prices = [limit_price, limit_price, limit_price]
+
+            total_action_qty = sum(stage_qtys)
+            trade_value_krw  = _round2(total_action_qty * limit_price)
+            cost_est_krw     = _round2(trade_value_krw * TX_COST_ROUNDTRIP)
+            # 비용 차감 후 드리프트 개선 효과
+            net_benefit_pct  = _round2(abs(drift) - TX_COST_ROUNDTRIP * 100)
+
+        ticker_rows.append({
+            "ticker":        h["ticker"],
+            "name":          h["name"],
+            "bucket":        bname,
+            "target_pct":    target_pct,
+            "current_pct":   current_pct,
+            "drift_pct":     drift,
+            "band_min":      band_min,
+            "band_max":      band_max,
+            "regime_band":   band["label"],
+            "drift_status":  drift_status,
+            "force_signal":  force,
+            "gate_status":   gate_status,
+            "action":        action,
+            "stage1_qty":    stage_qtys[0],
+            "stage1_price":  limit_prices[0],
+            "stage2_qty":    stage_qtys[1],
+            "stage2_price":  limit_prices[1],
+            "stage3_qty":    stage_qtys[2],
+            "stage3_price":  limit_prices[2],
+            "trade_value_krw": trade_value_krw,
+            "cost_est_krw":    cost_est_krw,
+            "net_benefit_pct": net_benefit_pct,
+            "close":           h["close"],
+        })
+
+    # ── 5. ORDERS 생성 (P4: gate_status=PASS 또는 FORCE_OVERRIDE のみ) ────────
+    order_rows: list[dict] = []
+    order_no = 1
+    # Priority: FORCE_OVERRIDE > BREACH_HIGH(SELL) > BREACH_LOW(BUY)
+    priority_map = {"FORCE_OVERRIDE": 0, "PASS": 1}
+    sorted_tickers = sorted(
+        [t for t in ticker_rows if t["gate_status"] in ("PASS", "FORCE_OVERRIDE")],
+        key=lambda t: (priority_map.get(t["gate_status"], 9), -abs(t["drift_pct"]))
+    )
+
+    for t in sorted_tickers:
+        for stage_idx, (qty, price) in enumerate(zip(
+            [t["stage1_qty"], t["stage2_qty"], t["stage3_qty"]],
+            [t["stage1_price"], t["stage2_price"], t["stage3_price"]],
+        ), start=1):
+            if qty <= 0:
+                continue
+            reason = t["force_signal"] if t["force_signal"] else t["drift_status"]
+            order_rows.append({
+                "order_no":       order_no,
+                "ticker":         t["ticker"],
+                "name":           t["name"],
+                "bucket":         t["bucket"],
+                "action":         t["action"],
+                "stage":          stage_idx,
+                "qty":            qty,
+                "limit_price_krw": price,
+                "trade_value_krw": round(qty * price),
+                "reason":         reason,
+            })
+            order_no += 1
+
+    # ── 6. SUMMARY ───────────────────────────────────────────────────────────
+    summary = {
+        "run_date":           _s((payload.get("metadata") or {}).get("converted_at", "")),
+        "regime":             regime,
+        "regime_band":        band["label"],
+        "total_portfolio_krw": total_portfolio_krw,
+        "cash_krw":           cash_krw,
+        "core_pct":           _round2(core_pct),
+        "satellite_pct":      _round2(sat_pct),
+        "cash_pct":           _round2(cash_pct),
+        "target_core_pct":    BUCKET_CONFIG["Core"]["target"],
+        "target_sat_pct":     BUCKET_CONFIG["Satellite"]["target"],
+        "target_cash_pct":    BUCKET_CONFIG["Cash"]["target"],
+        "rebalance_needed":   rebalance_needed,
+        "holdings_count":     len(holdings),
+        "orders_count":       len(order_rows),
+        "min_actionable_drift_pct": MIN_ACTIONABLE_DRIFT_PCT,
+        "cash_floor_min_pct": cash_floor_pct,
+        "cash_shortfall_krw": cash_shortfall_krw,
+        "regime_trim_lock":   bool(hc_data.get("regime_trim_lock", False)),
+        "new_buy_gate":       _s((_extract_trim_guidance(hc_data)).get("new_buy_gate", "")),
+    }
+
+    # ── 7. 출력 ──────────────────────────────────────────────────────────────
+    out = {
+        "formula_id": FORMULA_ID,
+        "metadata": {
+            "per_ticker_target_method": "equal_weight_within_bucket",
+            "regime_source": "macro.REGIME_PRELIM > settings > harness_context > computed_harness",
+        },
+        "summary":      summary,
+        "buckets":      bucket_rows,
+        "tickers":      ticker_rows,
+        "orders":       order_rows,
+    }
+
+    out_path.parent.mkdir(parents=True, exist_ok=True)
+    out_path.write_text(json.dumps(out, ensure_ascii=False, indent=2), encoding="utf-8")
+
+    action_count = len([t for t in ticker_rows if t["action"] not in ("HOLD", "WATCH")])
+    print(f"[{FORMULA_ID}] regime={regime} rebalance_needed={rebalance_needed} "
+          f"holdings={len(holdings)} action_tickers={action_count} orders={len(order_rows)}")
+    print(f"  core={core_pct:.1f}% sat={sat_pct:.1f}% cash={cash_pct:.1f}%")
+    print(f"  -> {out_path}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())