GT 총자산 비율 매수·leg 티어 배분과 시뮬/실거래 포지션 사이징을 통합한다.

타점·비중을 gt_model로 일반화하고, amount_krw 시각순 배분·EV/WF·상위 leg 대형 매수를 position_sizing과 시뮬 HTML(고정 ₩/회 비교)에 반영한다. Co-authored-by: Cursor <cursoragent@cursor.com>
2026-05-31 16:11:49 +09:00
parent 2cb67c42b3
commit 5842cc9fa3
14 changed files with 2073 additions and 182 deletions
--- a/deepcoin/matching/position_sizing.py
+++ b/deepcoin/matching/position_sizing.py
@@ -0,0 +1,378 @@
+"""
+총자산 대비 최적 매수율(비중) · 현금 한도 · leg 상위·EV/WF 통과 대형 매수.
+"""
+
+from __future__ import annotations
+
+import json
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+
+from config import (
+    GT_BUY_PCT_LARGE_LEG,
+    GT_BUY_PCT_SMALL_LEG,
+    GT_INITIAL_CASH_KRW,
+    GT_LARGE_LEG_TOP_PCT,
+    GT_MIN_ORDER_KRW,
+    LIVE_BUY_PCT_LARGE,
+    LIVE_BUY_PCT_SMALL,
+    MATCH_GT_TOLERANCE_MIN,
+    SIM_FEE_STRESS_MULT,
+    SIM_GO_MIN_HOLDOUT_EV,
+    SIM_GO_MIN_HOLDOUT_PF,
+    SIM_GO_WF_POSITIVE_RATIO,
+    SIM_WALK_FORWARD_MIN_MONTHS,
+    TRADING_FEE_RATE,
+)
+from deepcoin.matching.load_rules import load_matched_rules
+from deepcoin.paths import MATCHING_FIRE_OUTCOMES, MATCHING_MATCHED_RULES
+
+
+def portfolio_totals(
+    cash: float,
+    qty: float,
+    price: float,
+) -> tuple[float, float, float]:
+    """
+    총보유자산·코인평가·가용현금(=총자산-평가액)을 계산합니다.
+
+    Args:
+        cash: 현금.
+        qty: 보유 수량.
+        price: 평가·체결가.
+
+    Returns:
+        (total_asset_krw, holding_value_krw, cash_krw).
+    """
+    holding = qty * price
+    total = cash + holding
+    return total, holding, cash
+
+
+def optimal_weight_share(weight: float, weight_sum_remaining: float) -> float:
+    """
+    leg 내 남은 매수 비중 대비 이번 체결 최적 매수율(0~1).
+
+    Args:
+        weight: 이번 타점 weight.
+        weight_sum_remaining: 동일 leg 남은 매수 weight 합.
+
+    Returns:
+        비중 비율.
+    """
+    if weight_sum_remaining > 0:
+        return weight / weight_sum_remaining
+    return 1.0
+
+
+def compute_buy_amount_krw(
+    cash: float,
+    qty: float,
+    price: float,
+    weight: float,
+    weight_sum_remaining: float,
+    *,
+    asset_pct_scale: float,
+    min_order_krw: float = GT_MIN_ORDER_KRW,
+    fee_rate: float = TRADING_FEE_RATE,
+) -> float:
+    """
+    총자산 × (최적 매수율 × scale)을 목표로, 가용 현금을 넘지 않게 매수 원화를 산출합니다.
+
+    Args:
+        cash: 현금.
+        qty: 보유 수량.
+        price: 체결가.
+        weight: 타점 비중.
+        weight_sum_remaining: leg 내 남은 매수 weight 합.
+        asset_pct_scale: leg·규칙 티어(대형/소형) 스케일.
+        min_order_krw: 최소 주문 원화.
+        fee_rate: 수수료율.
+
+    Returns:
+        매수 원화(0이면 미체결).
+    """
+    if price <= 0:
+        return 0.0
+    total_asset, _, _ = portfolio_totals(cash, qty, price)
+    budget = max(cash / (1.0 + fee_rate), 0.0)
+    opt_rate = optimal_weight_share(weight, weight_sum_remaining) * asset_pct_scale
+    target = total_asset * opt_rate
+    amount = min(target, budget)
+    if budget >= min_order_krw and 0 < amount < min_order_krw:
+        amount = min(min_order_krw, budget)
+    return round(max(amount, 0.0), 0)
+
+
+def top_leg_ids_by_forward_return(
+    trades: list[dict[str, Any]],
+    top_pct: float = GT_LARGE_LEG_TOP_PCT,
+) -> set[int]:
+    """
+    leg별 최대 forward_return 기준 상위 n% leg_id 집합.
+
+    Args:
+        trades: GT trade dict.
+        top_pct: 상위 비율(0~1).
+
+    Returns:
+        대형 매수 leg_id set.
+    """
+    leg_ret: dict[int, float] = {}
+    for t in trades:
+        if t.get("action") != "sell":
+            continue
+        lid = int(t.get("leg_id", 0))
+        ret = float(t.get("forward_return_pct") or 0.0)
+        leg_ret[lid] = max(leg_ret.get(lid, 0.0), ret)
+    if not leg_ret:
+        return set()
+    ranked = sorted(leg_ret.items(), key=lambda x: x[1], reverse=True)
+    n = max(1, int(len(ranked) * top_pct + 0.999999))
+    return {lid for lid, _ in ranked[:n]}
+
+
+def leg_asset_pct_scale(leg_id: int, large_legs: set[int]) -> float:
+    """
+    leg 티어에 따른 총자산 대비 매수 스케일.
+
+    Args:
+        leg_id: leg 번호.
+        large_legs: 상위 leg 집합.
+
+    Returns:
+        GT_BUY_PCT_LARGE_LEG 또는 GT_BUY_PCT_SMALL_LEG.
+    """
+    if leg_id in large_legs:
+        return float(GT_BUY_PCT_LARGE_LEG)
+    return float(GT_BUY_PCT_SMALL_LEG)
+
+
+def _parse_dt(dt: str) -> datetime:
+    return datetime.fromisoformat(str(dt).replace("Z", "+00:00")[:19])
+
+
+def nearest_gt_leg_id(
+    dt: str,
+    gt_trades: list[dict[str, Any]],
+    tolerance_min: int = MATCH_GT_TOLERANCE_MIN,
+) -> int | None:
+    """
+    시각에 가장 가까운 GT trade의 leg_id (매수 우선).
+
+    Args:
+        dt: 발화 시각.
+        gt_trades: GT trades.
+        tolerance_min: 허용 분.
+
+    Returns:
+        leg_id 또는 None.
+    """
+    if not gt_trades:
+        return None
+    t0 = _parse_dt(dt)
+    best_buy: int | None = None
+    best_buy_min = float(tolerance_min) + 1.0
+    best_any: int | None = None
+    best_any_min = float(tolerance_min) + 1.0
+    for t in gt_trades:
+        try:
+            t1 = _parse_dt(t["dt"])
+        except ValueError:
+            continue
+        delta = abs((t0 - t1).total_seconds()) / 60.0
+        if delta > tolerance_min:
+            continue
+        lid = int(t.get("leg_id", 0))
+        if t.get("action") == "buy" and delta < best_buy_min:
+            best_buy_min = delta
+            best_buy = lid
+        if delta < best_any_min:
+            best_any_min = delta
+            best_any = lid
+    return best_buy if best_buy is not None else best_any
+
+
+_APPROVED_RULES_CACHE: set[str] | None = None
+
+
+def load_ev_wf_approved_rule_ids(
+    matched_path: Path | None = None,
+    outcomes_path: Path | None = None,
+) -> set[str]:
+    """
+    holdout EV·PF, walk-forward, 수수료 스트레스를 모두 통과한 rule_id.
+
+    Args:
+        matched_path: matched_rules.json.
+        outcomes_path: fire_outcomes.csv.
+
+    Returns:
+        통과 rule_id set. 산출 불가 시 monitor_rules 전체 fallback.
+    """
+    global _APPROVED_RULES_CACHE
+    if _APPROVED_RULES_CACHE is not None:
+        return set(_APPROVED_RULES_CACHE)
+
+    from deepcoin.matching.select_rules import _rule_metrics, _split_train_valid_holdout
+    from deepcoin.matching.simulation import (
+        evaluate_go_no_go,
+        simulate_live_order_cap,
+        walk_forward_by_month,
+        walk_forward_summary,
+    )
+
+    mp = matched_path or MATCHING_MATCHED_RULES
+    op = outcomes_path or MATCHING_FIRE_OUTCOMES
+    matched = load_matched_rules(mp)
+    rules = matched.get("monitor_rules") or []
+    if not rules or not op.is_file():
+        return {r["rule_id"] for r in rules}
+
+    import pandas as pd
+
+    from config import MATCH_FEE_RATE
+
+    outcomes = pd.read_csv(op)
+    outcomes["split"] = _split_train_valid_holdout(outcomes)
+    wf_sum = walk_forward_summary(walk_forward_by_month(outcomes))
+    fee_stress: dict[str, Any] = {}
+    for rid in outcomes["rule_id"].unique():
+        sub = outcomes[outcomes["rule_id"] == rid]
+        from deepcoin.matching.simulation import _fee_adjust_ret
+
+        adj = _fee_adjust_ret(sub["forward_ret_pct"], SIM_FEE_STRESS_MULT)
+        fee_stress[rid] = _rule_metrics(sub.assign(forward_ret_pct=adj))
+    monitor_ids = {r["rule_id"] for r in rules}
+    live_cap = simulate_live_order_cap(
+        outcomes, rule_ids=monitor_ids, holdout_only=True
+    )
+    go = evaluate_go_no_go(matched, wf_sum, fee_stress, live_cap)
+    passed = {c["rule_id"] for c in go.get("checks", []) if c.get("pass")}
+    if passed:
+        _APPROVED_RULES_CACHE = passed
+        return passed
+    fallback = monitor_ids
+    _APPROVED_RULES_CACHE = fallback
+    return fallback
+
+
+def live_buy_asset_pct_scale(
+    rule_id: str,
+    dt: str,
+    gt_trades: list[dict[str, Any]],
+    *,
+    approved_rules: set[str],
+    large_legs: set[int],
+) -> float:
+    """
+    실거래·시뮬 매수: EV/WF 통과 규칙 + leg 상위만 대형 비율.
+
+    Args:
+        rule_id: 규칙 ID.
+        dt: 체결 시각.
+        gt_trades: GT trades.
+        approved_rules: 통과 rule_id.
+        large_legs: 상위 leg.
+
+    Returns:
+        LIVE_BUY_PCT_LARGE 또는 LIVE_BUY_PCT_SMALL(또는 0에 가까운 소형).
+    """
+    if rule_id not in approved_rules:
+        return float(LIVE_BUY_PCT_SMALL)
+    lid = nearest_gt_leg_id(dt, gt_trades)
+    if lid is not None and lid in large_legs:
+        return float(LIVE_BUY_PCT_LARGE)
+    return float(LIVE_BUY_PCT_SMALL)
+
+
+def attach_dynamic_buy_amounts(
+    trades: list[dict[str, Any]],
+    *,
+    gt_trades: list[dict[str, Any]],
+    approved_rules: set[str] | None = None,
+    large_legs: set[int] | None = None,
+    initial_cash: float = GT_INITIAL_CASH_KRW,
+    default_weight: float = 1.0,
+    fee_rate: float = TRADING_FEE_RATE,
+) -> list[dict[str, Any]]:
+    """
+    시뮬 발화 trade dict에 amount_krw(총자산 비율·현금 한도)를 채웁니다.
+
+    Args:
+        trades: 시간순 {dt, action, price, rule_id, …}.
+        gt_trades: GT leg 매칭용.
+        approved_rules: EV/WF 통과 rule. None이면 전 규칙 대형 허용 안 함.
+        large_legs: 상위 leg. None이면 GT에서 산출.
+        initial_cash: 초기 현금.
+        default_weight: 매수 weight 기본값.
+        fee_rate: 수수료율.
+
+    Returns:
+        amount_krw가 채워진 동일 리스트.
+    """
+    if large_legs is None:
+        large_legs = top_leg_ids_by_forward_return(gt_trades)
+    if approved_rules is None:
+        approved_rules = set()
+
+    cash = float(initial_cash)
+    qty = 0.0
+    for t in sorted(trades, key=lambda x: x["dt"]):
+        action = t.get("action", t.get("side", ""))
+        price = float(t["price"])
+        if price <= 0:
+            continue
+        if action == "buy":
+            rid = str(t.get("rule_id", ""))
+            scale = live_buy_asset_pct_scale(
+                rid, t["dt"], gt_trades,
+                approved_rules=approved_rules,
+                large_legs=large_legs,
+            )
+            amount = compute_buy_amount_krw(
+                cash,
+                qty,
+                price,
+                float(t.get("weight", default_weight)),
+                float(t.get("weight", default_weight)),
+                asset_pct_scale=scale,
+                fee_rate=fee_rate,
+            )
+            t["amount_krw"] = amount
+            if amount > 0:
+                fee = amount * fee_rate
+                cash -= amount + fee
+                qty += amount / price
+        elif action == "sell" and qty > 1e-12:
+            gross = qty * price
+            t["amount_krw"] = round(gross, 0)
+            fee = gross * fee_rate
+            cash += gross - fee
+            qty = 0.0
+    return trades
+
+
+def load_sizing_context_from_gt(
+    gt_path: Path | None = None,
+) -> tuple[list[dict[str, Any]], set[int], set[str]]:
+    """
+    GT JSON에서 trades, 상위 leg, EV/WF 통과 rule을 로드합니다.
+
+    Args:
+        gt_path: ground_truth_trades.json.
+
+    Returns:
+        (gt_trades, large_legs, approved_rules).
+    """
+    from deepcoin.paths import resolve_ground_truth_file
+
+    p = gt_path or resolve_ground_truth_file()
+    trades: list[dict[str, Any]] = []
+    if p.is_file():
+        data = json.loads(p.read_text(encoding="utf-8"))
+        trades = data.get("trades") or []
+    large = top_leg_ids_by_forward_return(trades)
+    approved = load_ev_wf_approved_rule_ids()
+    return trades, large, approved