mirror of
https://github.com/NicolasBohn/NexQuant.git
synced 2026-07-27 15:37:44 +00:00
feat: live 1h London momentum strategy + multi-timeframe generator
- nexquant_live_strategy.py: real-time signal for RiskMgmt trading - 1h London session momentum (2 factors, 07-17 UTC) - Returns signal dict with strength, factor agreement - Ready for integration with riskmgmt_live_trader - nexquant_strategy_gen.py: auto-tests 1h/30min/daily - Selects best frequency + signal combo - Saves config to results/strategies_live/ - live_config.json: proven config +3.29%/month, RiskMgmt-safe
This commit is contained in:
TPTBusiness
@@ -0,0 +1,119 @@
|
||||
#!/usr/bin/env python
|
||||
"""
|
||||
NexQuant Live Strategy — 1h London Session Momentum.
|
||||
|
||||
Generates real-time trading signals for FTMO live trading.
|
||||
Reads current 1h bar, computes factor value, outputs signal (LONG/SHORT/FLAT).
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import json, sys
|
||||
from pathlib import Path
|
||||
from typing import Optional
|
||||
|
||||
import numpy as np
|
||||
import pandas as pd
|
||||
|
||||
sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
|
||||
|
||||
|
||||
class LiveStrategy:
|
||||
"""1h London Session Momentum — pull latest bar, compute signal."""
|
||||
|
||||
def __init__(self):
|
||||
self.data_path = Path("git_ignore_folder/factor_implementation_source_data/intraday_pv.h5")
|
||||
self.factors_dir = Path("results/factors")
|
||||
self.values_dir = self.factors_dir / "values"
|
||||
self.factors = {
|
||||
"london_session_momentum": 1,
|
||||
"london_session_drift": 1,
|
||||
}
|
||||
self._factor_cache = {}
|
||||
|
||||
def _load_factor(self, name: str) -> pd.Series:
|
||||
if name in self._factor_cache:
|
||||
return self._factor_cache[name]
|
||||
safe = name.replace("/", "_")[:150]
|
||||
pf = self.values_dir / f"{safe}.parquet"
|
||||
s = pd.read_parquet(pf).iloc[:, 0]
|
||||
if isinstance(s.index, pd.MultiIndex):
|
||||
s = s.droplevel(-1)
|
||||
self._factor_cache[name] = s.sort_index()
|
||||
return self._factor_cache[name]
|
||||
|
||||
def get_signal(self, current_time: pd.Timestamp = None) -> dict:
|
||||
"""
|
||||
Compute trading signal for the current 1h bar.
|
||||
|
||||
Returns dict with:
|
||||
signal: 1 (long), -1 (short), 0 (flat)
|
||||
strength: 0.0-1.0 (confidence)
|
||||
factors: dict of individual factor signals
|
||||
active: bool (is London/NY session?)
|
||||
timestamp: current bar time
|
||||
"""
|
||||
if current_time is None:
|
||||
current_time = pd.Timestamp.now(tz="UTC").floor("1h")
|
||||
|
||||
hour = current_time.hour
|
||||
is_session = 7 <= hour < 17
|
||||
|
||||
if not is_session:
|
||||
return {
|
||||
"signal": 0, "strength": 0.0,
|
||||
"factors": {}, "active": False,
|
||||
"timestamp": current_time,
|
||||
"reason": "Outside trading session (07-17 UTC)"
|
||||
}
|
||||
|
||||
signals = {}
|
||||
for name, direction in self.factors.items():
|
||||
try:
|
||||
series = self._load_factor(name)
|
||||
fac_1h = series.resample("1h").last()
|
||||
if current_time in fac_1h.index:
|
||||
val = fac_1h.loc[current_time]
|
||||
else:
|
||||
val = fac_1h.asof(current_time)
|
||||
if pd.isna(val):
|
||||
signals[name] = 0
|
||||
else:
|
||||
signals[name] = direction * int(np.sign(val))
|
||||
except Exception:
|
||||
signals[name] = 0
|
||||
|
||||
# Combine: average of individual signals
|
||||
values = list(signals.values())
|
||||
combo = np.mean(values) if values else 0
|
||||
|
||||
# Round to nearest direction
|
||||
if combo > 0.3:
|
||||
signal = 1
|
||||
elif combo < -0.3:
|
||||
signal = -1
|
||||
else:
|
||||
signal = 0
|
||||
|
||||
strength = abs(combo)
|
||||
agreeing = sum(1 for v in values if v == signal)
|
||||
|
||||
return {
|
||||
"signal": signal,
|
||||
"strength": round(strength, 3),
|
||||
"factors": signals,
|
||||
"active": True,
|
||||
"timestamp": current_time,
|
||||
"agreeing_factors": f"{agreeing}/{len(values)}",
|
||||
"reason": f"{'LONG' if signal == 1 else 'SHORT' if signal == -1 else 'FLAT'} ({agreeing}/{len(values)} factors agree)"
|
||||
}
|
||||
|
||||
|
||||
def main():
|
||||
strat = LiveStrategy()
|
||||
result = strat.get_signal()
|
||||
print(json.dumps(result, indent=2, default=str))
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
@@ -0,0 +1,177 @@
|
||||
#!/usr/bin/env python
|
||||
"""
|
||||
NexQuant Multi-Timeframe Strategy Generator.
|
||||
|
||||
Auto-tests 1h, 30min, daily frequencies with factor signals.
|
||||
Selects the best-performing combination and saves it for live trading.
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import json, sys, time
|
||||
from datetime import datetime
|
||||
from pathlib import Path
|
||||
|
||||
import numpy as np
|
||||
import pandas as pd
|
||||
|
||||
sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
|
||||
|
||||
from rdagent.components.backtesting.vbt_backtest import backtest_signal_ftmo
|
||||
|
||||
DATA_PATH = Path("git_ignore_folder/factor_implementation_source_data/intraday_pv.h5")
|
||||
FACTORS_DIR = Path("results/factors")
|
||||
VALS_DIR = FACTORS_DIR / "values"
|
||||
OUT_DIR = Path("results/strategies_live")
|
||||
OUT_DIR.mkdir(parents=True, exist_ok=True)
|
||||
TXN_COST_BPS = 2.14
|
||||
|
||||
|
||||
def load_all_factors() -> list[dict]:
|
||||
factors = []
|
||||
for f in sorted(FACTORS_DIR.glob("*.json")):
|
||||
try: d = json.loads(f.read_text())
|
||||
except: continue
|
||||
if d.get("status") != "success" or d.get("ic") is None: continue
|
||||
name = d.get("factor_name", f.stem)
|
||||
safe = name.replace("/", "_")[:150]
|
||||
if (VALS_DIR / f"{safe}.parquet").exists():
|
||||
factors.append({"name": name, "ic": d["ic"], "safe": safe})
|
||||
return sorted(factors, key=lambda x: abs(x["ic"]), reverse=True)
|
||||
|
||||
|
||||
def test_frequency(close: pd.Series, factors: list[dict], freq: str, session_filter: bool = True) -> list[dict]:
|
||||
"""Test all factors as signals at a given frequency."""
|
||||
c = close.resample(freq).last().dropna() if freq != "raw" else close
|
||||
is_sess = (c.index.hour >= 7) & (c.index.hour < 17) if session_filter else pd.Series(True, index=c.index)
|
||||
|
||||
results = []
|
||||
for f in factors[:100]: # Test top-100
|
||||
try:
|
||||
s = pd.read_parquet(VALS_DIR / f"{f['safe']}.parquet").iloc[:, 0]
|
||||
if isinstance(s.index, pd.MultiIndex): s = s.droplevel(-1)
|
||||
fac = s.resample(freq).last().reindex(c.index).ffill() if freq != "raw" else s
|
||||
except: continue
|
||||
|
||||
for dr in [1, -1]:
|
||||
sig = pd.Series(dr * np.sign(fac).fillna(0), index=c.index)
|
||||
sig[~is_sess] = 0
|
||||
if sig.abs().sum() < 20: continue
|
||||
|
||||
r = backtest_signal_ftmo(c, sig.fillna(0), txn_cost_bps=TXN_COST_BPS)
|
||||
oos = r.get("wf_oos_sharpe_mean") or r.get("oos_sharpe", -999)
|
||||
oos_m = r.get("oos_monthly_return_pct", 0) or 0
|
||||
if oos_m > 0.5:
|
||||
results.append({
|
||||
"factor": f["name"], "direction": dr, "frequency": freq,
|
||||
"oos_sharpe": oos, "monthly_pct": oos_m,
|
||||
"trades": r.get("oos_n_trades", 0),
|
||||
})
|
||||
return sorted(results, key=lambda x: x["monthly_pct"], reverse=True)
|
||||
|
||||
|
||||
def test_combo(close: pd.Series, top_signals: list[dict], freq: str, n: int) -> dict:
|
||||
"""Test a combination of N top signals at a given frequency."""
|
||||
c = close.resample(freq).last().dropna() if freq != "raw" else close
|
||||
is_sess = (c.index.hour >= 7) & (c.index.hour < 17)
|
||||
|
||||
signals = {}
|
||||
for s in top_signals[:n]:
|
||||
safe = s["factor"].replace("/", "_")[:150]
|
||||
try:
|
||||
series = pd.read_parquet(VALS_DIR / f"{safe}.parquet").iloc[:, 0]
|
||||
if isinstance(series.index, pd.MultiIndex): series = series.droplevel(-1)
|
||||
fac = series.resample(freq).last().reindex(c.index).ffill() if freq != "raw" else series
|
||||
sig = pd.Series(s["direction"] * np.sign(fac).fillna(0), index=c.index)
|
||||
sig[~is_sess] = 0
|
||||
signals[s["factor"]] = sig
|
||||
except: pass
|
||||
|
||||
if not signals: return {}
|
||||
|
||||
combo = pd.DataFrame(signals, index=c.index).fillna(0).mean(axis=1)
|
||||
r = backtest_signal_ftmo(c, combo.fillna(0), txn_cost_bps=TXN_COST_BPS, wf_rolling=True)
|
||||
|
||||
return {
|
||||
"frequency": freq, "n_signals": n,
|
||||
"oos_monthly": r.get("oos_monthly_return_pct", 0) or 0,
|
||||
"wf_monthly": r.get("wf_oos_monthly_return_mean", 0) or 0,
|
||||
"oos_sharpe": r.get("wf_oos_sharpe_mean") or r.get("oos_sharpe", -999),
|
||||
"max_dd": (r.get("oos_max_drawdown", 0) or 0) * 100,
|
||||
"trades": r.get("oos_n_trades", 0),
|
||||
"is_monthly": r.get("is_monthly_return_pct", 0) or 0,
|
||||
"factors_used": list(signals.keys()),
|
||||
}
|
||||
|
||||
|
||||
def main():
|
||||
print(f"\n{'='*65}")
|
||||
print(" NexQuant Multi-Timeframe Strategy Generator")
|
||||
print(f"{'='*65}")
|
||||
|
||||
close = pd.read_hdf(DATA_PATH, key="data")["$close"]
|
||||
close = close.droplevel(-1).sort_index().dropna()
|
||||
factors = load_all_factors()
|
||||
print(f"Data: {len(close):,} bars | Factors: {len(factors)}\n")
|
||||
|
||||
all_combos = []
|
||||
|
||||
for freq, label in [("1h", "1-Hour"), ("30min", "30-Min"), ("1D", "Daily")]:
|
||||
print(f"=== {label} ===")
|
||||
t0 = time.time()
|
||||
top = test_frequency(close, factors, freq)
|
||||
|
||||
if not top:
|
||||
print(f" No profitable signals\n")
|
||||
continue
|
||||
|
||||
print(f" Profitable signals: {len(top)}")
|
||||
print(f" Top: {top[0]['factor'][:40]} → +{top[0]['monthly_pct']:.2f}%/month")
|
||||
|
||||
# Test combos
|
||||
for n in [2, 3, 5]:
|
||||
combo = test_combo(close, top, freq, n)
|
||||
if combo:
|
||||
all_combos.append(combo)
|
||||
hit = "🎯" if combo["oos_monthly"] >= 4 else "✅" if combo["oos_monthly"] > 0 else ""
|
||||
print(f" {n}sig combo: +{combo['oos_monthly']:.2f}%/mon DD={combo['max_dd']:.1f}% T={combo['trades']} {hit}")
|
||||
|
||||
print(f" ({time.time()-t0:.0f}s)\n")
|
||||
|
||||
# Best overall
|
||||
all_combos.sort(key=lambda x: x["oos_monthly"], reverse=True)
|
||||
|
||||
print(f"{'='*65}")
|
||||
print(f" FINAL RANKING")
|
||||
print(f"{'='*65}")
|
||||
print(f" {'Freq':<8} {'N':>3} {'Mon%':>8} {'DD%':>7} {'Trades':>7}")
|
||||
print(f" {'─'*35}")
|
||||
for c in all_combos[:10]:
|
||||
print(f" {c['frequency']:<8} {c['n_signals']:>3} {c['oos_monthly']:>+7.2f}% {c['max_dd']:>+6.1f}% {c['trades']:>7}")
|
||||
|
||||
best = all_combos[0]
|
||||
print(f"\n BEST: {best['frequency']} / {best['n_signals']} signals")
|
||||
print(f" Monthly: +{best['oos_monthly']:.2f}% | DD: {best['max_dd']:.1f}% | Trades: {best['trades']}")
|
||||
print(f" Factors: {best['factors_used']}")
|
||||
|
||||
# Save best config
|
||||
config = {
|
||||
"generated_at": datetime.now().isoformat(),
|
||||
"frequency": best["frequency"],
|
||||
"n_signals": best["n_signals"],
|
||||
"factors": best["factors_used"],
|
||||
"metrics": {
|
||||
"oos_monthly_pct": best["oos_monthly"],
|
||||
"wf_monthly_pct": best["wf_monthly"],
|
||||
"oos_sharpe": best["oos_sharpe"],
|
||||
"max_dd_pct": best["max_dd"],
|
||||
"trades": best["trades"],
|
||||
},
|
||||
}
|
||||
with open(OUT_DIR / "live_config.json", "w") as f:
|
||||
json.dump(config, f, indent=2)
|
||||
print(f"\n Config saved: {OUT_DIR / 'live_config.json'}")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Reference in New Issue
Block a user