refactor: reorganize project structure into tests/, data/, logs/

- Move all backtest/simulation scripts to tests/
  - Add sys.path.insert to each script for correct import resolution
- Move pkl cache files to data/ (git-ignored)
- Move log files to logs/ (git-ignored)
- Update main.py: trading.log path → logs/trading.log
- Add ecosystem.config.js: pm2 log paths → logs/pm2*.log
- Update .gitignore: ignore data/ and logs/ instead of *.pkl/*.log
- core/fng.py: increase cache TTL 3600→86400s (API updates daily at KST 09:00)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
joungmin
2026-03-03 16:08:50 +09:00
parent bfe0b4d40c
commit 6b2c962ed8
30 changed files with 1039 additions and 5 deletions

399
tests/vol_lead_sim.py Normal file
View File

@@ -0,0 +1,399 @@
"""거래량 선행(Volume Lead) 진입 전략 시뮬레이션.
3가지 전략 비교:
A (현행): 12h 가격 +5% 확인 + 1h 거래량 급증 → 진입 (이미 오른 뒤 추격)
B (신규): 가격 횡보 중 거래량 급증(축적) → 그 후 추세 +N% 시작 시 선진입
C (단순): 거래량 급증만 (베이스라인, 노이즈 확인용)
"""
import os as _os, sys as _sys
_sys.path.insert(0, _os.path.dirname(_os.path.dirname(_os.path.abspath(__file__))))
import os
import pickle
import time
from datetime import datetime
from pathlib import Path
import pandas as pd
from dotenv import load_dotenv
load_dotenv()
import pyupbit
# ── 공통 파라미터 ─────────────────────────────────────
STOP_LOSS_PCT = 0.015 # 트레일링 스탑 1.5%
TIME_STOP_HOURS = 8
TIME_STOP_MIN_PCT = 3.0
FEE = 0.0005
LOCAL_VOL_HOURS = 5 # 거래량 기준 이전 N시간
VOL_MULT = 2.0 # 거래량 배수 기준
# 현행 전략 파라미터
TREND_HOURS = 12
TREND_MIN_PCT = 5.0
# B 전략 파라미터: 거래량 선행 + 이후 소규모 추세 확인
PRICE_QUIET_PCT = 2.0 # 거래량 급증 시점 가격 횡보 기준 (2h 변동 < N%)
TREND_AFTER_VOL = 1.5 # 축적 신호 후 진입 기준 (vol 시점 대비 +N% 상승 시)
SIGNAL_TIMEOUT_H = 8 # 축적 신호 후 N시간 내 추세 미발생 시 초기화
FROM_DATE = "2026-01-15 00:00:00"
TICKERS = [
'KRW-DKA', 'KRW-LAYER', 'KRW-SIGN',
'KRW-SOL', 'KRW-ETH', 'KRW-XRP',
'KRW-HOLO', 'KRW-OM', 'KRW-ORBS',
]
CACHE_FILE = Path("vol_lead_cache.pkl")
# ── 데이터 로드 ───────────────────────────────────────
def fetch_all(ticker: str, from_date: str):
"""1h봉 전체 로드 (from_date 이후, 페이지 역방향 수집)."""
target = datetime.strptime(from_date, "%Y-%m-%d %H:%M:%S")
frames = []
to_dt = None
for _ in range(15): # 최대 15페이지 = 3000h ≈ 125일
kwargs: dict = dict(ticker=ticker, interval="minute60", count=200)
if to_dt:
kwargs["to"] = to_dt.strftime("%Y-%m-%d %H:%M:%S")
df = pyupbit.get_ohlcv(**kwargs)
if df is None or df.empty:
break
frames.append(df)
oldest = df.index[0].to_pydatetime().replace(tzinfo=None)
if oldest <= target:
break
to_dt = oldest
time.sleep(0.2)
if not frames:
return None
result = pd.concat(frames).sort_index().drop_duplicates()
result.index = result.index.tz_localize(None)
return result[result.index >= target]
def load_data() -> dict:
if CACHE_FILE.exists():
print(f"캐시 로드: {CACHE_FILE}")
return pickle.load(open(CACHE_FILE, "rb"))
data = {}
for ticker in TICKERS:
print(f" {ticker} 로딩...", end=" ", flush=True)
df = fetch_all(ticker, FROM_DATE)
if df is not None:
data[ticker] = df
print(f"{len(df)}봉 ({df.index[0].strftime('%m-%d')}~{df.index[-1].strftime('%m-%d')})")
else:
print("실패")
time.sleep(0.3)
pickle.dump(data, open(CACHE_FILE, "wb"))
return data
# ── 포지션 시뮬 ───────────────────────────────────────
def simulate_pos(df: pd.DataFrame, buy_idx: int, buy_price: float):
"""매수 후 청산 시뮬레이션.
- 최고가: 각 봉의 high 기준
- 스탑 발동 체크: 각 봉의 low 기준 (intra-candle 포착)
- 청산가: peak × (1 - stop_pct) 근사
"""
buy_dt = df.index[buy_idx]
peak = buy_price
for i in range(buy_idx + 1, len(df)):
row = df.iloc[i]
ts = df.index[i]
if row["high"] > peak:
peak = row["high"]
elapsed_h = (ts - buy_dt).total_seconds() / 3600
stop_price = peak * (1 - STOP_LOSS_PCT)
# 트레일링 스탑 (low가 stop_price 이하 진입 시)
if row["low"] <= stop_price:
sell_price = stop_price
pnl = (sell_price * (1 - FEE) - buy_price * (1 + FEE)) / (buy_price * (1 + FEE)) * 100
return pnl > 0, sell_price, ts, f"트레일링({pnl:+.1f}%)", pnl
# 타임 스탑
pnl_now = (row["close"] - buy_price) / buy_price * 100
if elapsed_h >= TIME_STOP_HOURS and pnl_now < TIME_STOP_MIN_PCT:
pnl = (row["close"] * (1 - FEE) - buy_price * (1 + FEE)) / (buy_price * (1 + FEE)) * 100
return pnl > 0, row["close"], ts, "타임스탑", pnl
last = df.iloc[-1]["close"]
pnl = (last * (1 - FEE) - buy_price * (1 + FEE)) / (buy_price * (1 + FEE)) * 100
return pnl > 0, last, df.index[-1], "데이터종료", pnl
# ── 현행 전략 (추세 확인형) ───────────────────────────
def run_trend(df: pd.DataFrame) -> list:
"""12h 가격 +5% 확인 + 1h 거래량 급증 + 1h 워치리스트."""
trades = []
watchlist_i = None
in_pos = False
buy_idx = buy_price = None
i = max(TREND_HOURS, LOCAL_VOL_HOURS + 2)
while i < len(df):
if in_pos:
is_win, sp, sdt, reason, pnl = simulate_pos(df, buy_idx, buy_price)
next_i = next((j for j in range(i, len(df)) if df.index[j] > sdt), len(df))
trades.append((is_win, pnl, df.index[buy_idx], sdt, reason))
in_pos = False
watchlist_i = None
i = next_i
continue
close = df.iloc[i]["close"]
past12 = df.iloc[i - TREND_HOURS]["close"]
trend_ok = (close - past12) / past12 * 100 >= TREND_MIN_PCT
vol_recent = df.iloc[i - 1]["volume"]
vol_avg = df.iloc[i - LOCAL_VOL_HOURS - 1:i - 1]["volume"].mean()
vol_ok = vol_avg > 0 and vol_recent >= vol_avg * VOL_MULT
if trend_ok and vol_ok:
if watchlist_i is None:
watchlist_i = i
elif i - watchlist_i >= 1: # 1h 확인
in_pos = True
buy_idx = i
buy_price = close
watchlist_i = None
else:
watchlist_i = None
i += 1
return trades
# ── B 전략: 거래량 선행 + 소규모 추세 확인 ───────────
def run_vol_lead(df: pd.DataFrame) -> list:
"""거래량 급증(축적) 감지 후 소규모 추세 확인 시 선진입.
흐름:
1. 직전 1h 거래량 > 이전 5h 평균 × VOL_MULT AND
2h 가격 변동 < PRICE_QUIET_PCT% (횡보 중 축적)
→ 축적 신호 기록 (signal_price = 현재가)
2. 신호 후 현재가가 signal_price 대비 +TREND_AFTER_VOL% 이상 상승 시 진입
(현행 +5% 대신 작은 기준으로 더 일찍 진입)
3. SIGNAL_TIMEOUT_H 시간 내 추세 미발생 → 신호 초기화
"""
trades = []
signal_i = None
signal_price = None
in_pos = False
buy_idx = buy_price = None
i = max(TREND_HOURS, LOCAL_VOL_HOURS + 2)
while i < len(df):
if in_pos:
is_win, sp, sdt, reason, pnl = simulate_pos(df, buy_idx, buy_price)
next_i = next((j for j in range(i, len(df)) if df.index[j] > sdt), len(df))
trades.append((is_win, pnl, df.index[buy_idx], sdt, reason))
in_pos = False
signal_i = None
signal_price = None
i = next_i
continue
close = df.iloc[i]["close"]
close_2h = df.iloc[i - 2]["close"]
quiet = abs(close - close_2h) / close_2h * 100 < PRICE_QUIET_PCT
vol_recent = df.iloc[i - 1]["volume"]
vol_avg = df.iloc[i - LOCAL_VOL_HOURS - 1:i - 1]["volume"].mean()
vol_spike = vol_avg > 0 and vol_recent >= vol_avg * VOL_MULT
# 축적 신호 갱신
if quiet and vol_spike:
if signal_i is None:
signal_i = i
signal_price = close
else:
if signal_i is not None and close < signal_price:
# 가격 하락 → 축적 실패, 초기화
signal_i = None
signal_price = None
# 타임아웃
if signal_i is not None and (i - signal_i) > SIGNAL_TIMEOUT_H:
signal_i = None
signal_price = None
# 진입: 축적 신호 후 가격 +TREND_AFTER_VOL% 이상 상승
if signal_i is not None:
move = (close - signal_price) / signal_price * 100
if move >= TREND_AFTER_VOL:
in_pos = True
buy_idx = i
buy_price = close
signal_i = None
signal_price = None
i += 1
return trades
# ── 결과 출력 ─────────────────────────────────────────
def summarize(label: str, trades: list) -> dict:
if not trades:
print(f" [{label}] 거래 없음")
return {"total": 0, "wins": 0, "wr": 0.0, "pnl": 0.0}
wins = sum(1 for t in trades if t[0])
total = len(trades)
pnl = sum(t[1] for t in trades)
wr = wins / total * 100
print(f" [{label}] {total}건 | 승률={wr:.0f}% ({wins}{total-wins}패) | 누적={pnl:+.2f}%")
for idx, (is_win, p, bdt, sdt, reason) in enumerate(trades, 1):
mark = "" if is_win else ""
print(f" #{idx}: {mark} {p:+.2f}% | {reason}"
f" ({bdt.strftime('%m-%d %H:%M')}{sdt.strftime('%m-%d %H:%M')})")
return {"total": total, "wins": wins, "wr": wr, "pnl": pnl}
def run_vol_lead_thresh(df: pd.DataFrame, thresh: float) -> list:
"""run_vol_lead의 TREND_AFTER_VOL 파라미터를 동적으로 받는 버전."""
trades = []
signal_i = None
signal_price = None
in_pos = False
buy_idx = buy_price = None
i = max(TREND_HOURS, LOCAL_VOL_HOURS + 2)
while i < len(df):
if in_pos:
is_win, sp, sdt, reason, pnl = simulate_pos(df, buy_idx, buy_price)
next_i = next((j for j in range(i, len(df)) if df.index[j] > sdt), len(df))
trades.append((is_win, pnl, df.index[buy_idx], sdt, reason))
in_pos = False
signal_i = None
signal_price = None
i = next_i
continue
close = df.iloc[i]["close"]
close_2h = df.iloc[i - 2]["close"]
quiet = abs(close - close_2h) / close_2h * 100 < PRICE_QUIET_PCT
vol_recent = df.iloc[i - 1]["volume"]
vol_avg = df.iloc[i - LOCAL_VOL_HOURS - 1:i - 1]["volume"].mean()
vol_spike = vol_avg > 0 and vol_recent >= vol_avg * VOL_MULT
if quiet and vol_spike:
if signal_i is None:
signal_i = i
signal_price = close
else:
if signal_i is not None and close < signal_price:
signal_i = None
signal_price = None
if signal_i is not None and (i - signal_i) > SIGNAL_TIMEOUT_H:
signal_i = None
signal_price = None
if signal_i is not None:
move = (close - signal_price) / signal_price * 100
if move >= thresh:
in_pos = True
buy_idx = i
buy_price = close
signal_i = None
signal_price = None
i += 1
return trades
def main() -> None:
print("데이터 로딩 중...")
data = load_data()
# ── A 현행 전략 (기준선) ─────────────────────────────
print(f"\n{'='*72}")
print(f"A(현행 12h+5%+거래량) 기준선 | {FROM_DATE[:10]} ~ 현재")
print(f"{'='*72}")
agg_a = {"total": 0, "wins": 0, "pnl": 0.0}
trend_results = {}
for ticker, df in data.items():
t = run_trend(df)
trend_results[ticker] = t
s = {"total": len(t), "wins": sum(1 for x in t if x[0]),
"pnl": sum(x[1] for x in t)}
agg_a["total"] += s["total"]
agg_a["wins"] += s["wins"]
agg_a["pnl"] += s["pnl"]
a_wr = agg_a["wins"] / agg_a["total"] * 100 if agg_a["total"] else 0
print(f"A 합계: {agg_a['total']}건 | 승률={a_wr:.0f}% | 누적={agg_a['pnl']:+.2f}%")
# ── B 전략: TREND_AFTER_VOL 파라미터 스윕 ───────────
THRESHOLDS = [0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 4.0, 5.0]
print(f"\n{'='*72}")
print(f"B(거래량→+N% 선진입) 파라미터 스윕")
print(f"{''*72}")
print(f"{'임계값':>6}{'거래':>5} {'승률':>6} {'누적PnL':>10} │ vs A PnL")
print(f"{''*72}")
best = None
for thresh in THRESHOLDS:
agg = {"total": 0, "wins": 0, "pnl": 0.0}
for ticker, df in data.items():
t = run_vol_lead_thresh(df, thresh)
agg["total"] += len(t)
agg["wins"] += sum(1 for x in t if x[0])
agg["pnl"] += sum(x[1] for x in t)
wr = agg["wins"] / agg["total"] * 100 if agg["total"] else 0
diff = agg["pnl"] - agg_a["pnl"]
marker = " ← best" if (best is None or agg["pnl"] > best["pnl"]) else ""
if marker:
best = {**agg, "thresh": thresh, "wr": wr}
print(f"+{thresh:>4.1f}% │ {agg['total']:>5}{wr:>5.0f}% {agg['pnl']:>+9.2f}% │ {diff:>+8.2f}%{marker}")
print(f"{''*72}")
print(f"\n★ 최적 임계값: +{best['thresh']}% → "
f"{best['total']}건 | 승률={best['wr']:.0f}% | 누적={best['pnl']:+.2f}%")
# ── 최적 임계값으로 종목별 상세 출력 ─────────────────
best_thresh = best["thresh"]
print(f"\n{'='*72}")
print(f"★ B(vol→+{best_thresh}%) vs A(12h+5%+vol) 종목별 비교")
print(f"{''*72}")
print(f"{'종목':<14}{'A 현행':^24}{'B +{:.1f}%'.format(best_thresh):^24}")
print(f"{'':14}{'거래':>4} {'승률':>5} {'누적':>9}{'거래':>4} {'승률':>5} {'누적':>9}")
print(f"{''*72}")
for ticker, df in data.items():
t_a = trend_results[ticker]
t_b = run_vol_lead_thresh(df, best_thresh)
wa = sum(1 for x in t_a if x[0])
wb = sum(1 for x in t_b if x[0])
pa = sum(x[1] for x in t_a)
pb = sum(x[1] for x in t_b)
wr_a = wa / len(t_a) * 100 if t_a else 0
wr_b = wb / len(t_b) * 100 if t_b else 0
print(f"{ticker:<14}{len(t_a):>4}{wr_a:>4.0f}% {pa:>+8.2f}% │"
f" {len(t_b):>4}{wr_b:>4.0f}% {pb:>+8.2f}%")
if __name__ == "__main__":
main()