""" test_gold_fork.py — Definitive D_LIQ_GOLD replication test on agent-forked code. Tests the D_LIQ_GOLD engine using the PROVEN run_backtest() path from exp_shared_AGENT_fork.py (exact copy of agent-modified exp_shared.py). Gold target: ROI=181.81%, DD=17.65%, Calmar=10.30, T=2155 Tolerance: ROI ±2pp, Trades ±5 Specific agent claims verified here: A. Orchestrator ceiling_lev 6→10 (set_esoteric_hazard_multiplier) B. MC red: proportional 0.80x vs hard 5.0x cap C. exp_shared rolling vol_p60 vs static D. float32 casting, lazy OB loading RUNS ON: exp_shared_AGENT_fork.py (do not touch production exp_shared.py) """ import sys, time from pathlib import Path ROOT = Path(r"C:\Users\Lenovo\Documents\- DOLPHIN NG HD HCM TSF Predict") sys.path.insert(0, str(ROOT / 'nautilus_dolphin')) # ── Import the FORKED exp_shared (not the live one) ─────────────────────────── import importlib.util, types def load_fork(fork_path: Path, module_name: str): spec = importlib.util.spec_from_file_location(module_name, fork_path) mod = importlib.util.module_from_spec(spec) sys.modules[module_name] = mod spec.loader.exec_module(mod) return mod print("Loading forked exp_shared...") exp = load_fork( ROOT / 'nautilus_dolphin' / 'dvae' / 'exp_shared_AGENT_fork.py', 'exp_shared_fork' ) from nautilus_dolphin.nautilus.proxy_boost_engine import create_d_liq_engine, LiquidationGuardEngine GOLD_ROI = 181.81 GOLD_TRADES = 2155 GOLD_DD = 17.65 GOLD_CALMAR = 10.30 TOL_ROI = 2.0 # ±2pp TOL_TRADES = 5 # ±5 def run_test(label: str, factory, extra_kwargs=None): print(f"\n{'='*60}") print(f" TEST: {label}") print(f"{'='*60}") t0 = time.time() result = exp.run_backtest(factory, label, extra_kwargs=extra_kwargs) elapsed = time.time() - t0 roi = result['roi'] trades = result['trades'] dd = result['dd'] calmar = result.get('calmar', 0.0) roi_ok = abs(roi - GOLD_ROI) <= TOL_ROI trade_ok = abs(trades - GOLD_TRADES) <= TOL_TRADES print(f"\n ROI: {roi:+.2f}% (gold={GOLD_ROI:.2f}% diff={roi-GOLD_ROI:+.2f}pp) {'✓' if roi_ok else '✗'}") print(f" Trades: {trades} (gold={GOLD_TRADES} diff={trades-GOLD_TRADES:+d}) {'✓' if trade_ok else '✗'}") print(f" DD: {dd:.2f}% (gold={GOLD_DD:.2f}%)") print(f" Calmar: {calmar:.2f} (gold={GOLD_CALMAR:.2f})") print(f" Time: {elapsed:.0f}s") passed = roi_ok and trade_ok print(f"\n VERDICT: {'PASS ✓' if passed else 'FAIL ✗'}") return {'label': label, 'roi': roi, 'trades': trades, 'dd': dd, 'calmar': calmar, 'elapsed': elapsed, 'pass': passed} if __name__ == '__main__': print("\n" + "="*60) print(" D_LIQ_GOLD REPLICATION — AGENT FORK TEST") print(f" Using: exp_shared_AGENT_fork.py") print(f" Gold: ROI={GOLD_ROI}% T={GOLD_TRADES} DD={GOLD_DD}% Calmar={GOLD_CALMAR}") print("="*60) print("\nWarming up JIT...") exp.ensure_jit() print(" JIT ready.") results = [] # Test 1: D_LIQ_GOLD — perfect maker (closest to gold certification) results.append(run_test( "D_LIQ_GOLD perfect-maker (fork)", lambda kw: create_d_liq_engine(**kw), extra_kwargs={ 'sp_maker_entry_rate': 1.0, 'sp_maker_exit_rate': 1.0, 'use_sp_slippage': False, } )) # Test 2: D_LIQ_GOLD stochastic (production mode, 0.62 fill) results.append(run_test( "D_LIQ_GOLD stochastic 0.62 (fork)", lambda kw: create_d_liq_engine(**kw), )) # ── Summary ──────────────────────────────────────────────────────────────── print(f"\n{'='*60}") print(" FORK TEST SUMMARY") print(f"{'='*60}") n_pass = sum(1 for r in results if r['pass']) for r in results: status = 'PASS ✓' if r['pass'] else 'FAIL ✗' print(f" {status} {r['label']:<40} ROI={r['roi']:+.2f}% T={r['trades']} Calmar={r['calmar']:.2f}") print(f"\n {n_pass}/{len(results)} tests passed") if n_pass == len(results): print("\n ✓ AGENT CHANGES VERIFIED — D_LIQ_GOLD REPLICATES ON FORK") else: print("\n ✗ REPLICATION FAILURE — agent changes compromise gold standard")