{ "run_date": "2026-03-22T00:44:43.404343", "methodology": { "split": "60/40 time-ordered IS/OOS", "fomc_excluded": true, "trace_corruption_excluded": "2025-10-27 to 2026-02-17", "n_permutations": 1000, "rng_seed": 42, "ic_method": "Spearman rank correlation", "wr_test": "one-tailed permutation (observed >= shuffled)", "ic_test": "two-tailed permutation (|observed| >= |shuffled|)" }, "signals": { "A1": { "signal": "A1: MM SKEW", "what": "IC of (MM gamma skew at 10AM: above_spot vs below_spot) vs 3H ES return. Claimed IC=-0.620 (contrarian: more gamma above spot \u2192 market falls). Sign=-1 (we use -skew as the signal so positive IC = contrarian works).", "n_oos": 134, "signal_metric": -0.04086464187426591, "null_baseline": -0.0017588311275770682, "p_value": 0.624, "is_metric": "IC=-0.0790", "oos_metric": "IC=-0.0409, WR=47.8%", "is_ic": -0.07897366946569886, "oos_ic": -0.04086464187426591, "oos_wr": 0.47761194029850745, "is_n": 199, "oos_n": 134, "gap_pp": 0.038109027591432955, "down_months": 0.54, "verdict": "FAILS" }, "A2": { "signal": "A2: GAMMA CONTRARIAN", "what": "IC of (total gamma at 10AM, all participants) vs 3H ES return. High gamma \u2192 expect fade/reversal. Signal = -total_gamma (contrarian). Claimed IC=-0.088.", "n_oos": 134, "signal_metric": 0.048691440905087674, "null_baseline": -0.002832871772557686, "p_value": 0.562, "is_metric": "IC=0.0599", "oos_metric": "IC=0.0487, WR=46.3%", "is_ic": 0.059856860057865086, "oos_ic": 0.048691440905087674, "oos_wr": 0.4626865671641791, "is_n": 199, "oos_n": 134, "gap_pp": 0.011165419152777412, "down_months": 0.42, "verdict": "FAILS" }, "A3": { "signal": "A3: GAMMA TILT (MM pct_above > 0.75)", "what": "Win rate when MM gamma tilt > 75% (more gamma above spot) \u2192 bullish 3H. Claimed WR 62-84% at various thresholds, morning only.", "n_oos": 35, "signal_metric": 0.6571428571428571, "null_baseline": 0.5749142857142857, "p_value": 0.182, "is_metric": "WR=55.3% (N=47)", "oos_metric": "WR=65.7% (N=35)", "down_months": 0.5555555555555556, "threshold_scan_oos": { "thr=0.55": "WR=63.8% N=47", "thr=0.6": "WR=65.9% N=44", "thr=0.65": "WR=70.0% N=40", "thr=0.7": "WR=66.7% N=36", "thr=0.75": "WR=65.7% N=35", "thr=0.8": "WR=61.3% N=31", "thr=0.85": "WR=58.6% N=29" }, "verdict": "FAILS" }, "A4": { "signal": "A4: FADE RETAIL (Customer Gamma Skew)", "what": "IC of (-cust_skew) vs 3H ES return. Fade customers: if cust_skew > 0 (customers long gamma above spot), expect negative return. Claimed IC=-0.126. PREVIOUSLY FLAGGED DEAD (IC=+0.024 on clean data). Verifying here.", "n_oos": 134, "signal_metric": -0.0049842352029658915, "null_baseline": 0.0004724307144392044, "p_value": 0.954, "is_metric": "IC=0.0296", "oos_metric": "IC=-0.0050, WR=50.7%", "is_ic": 0.029581523758841682, "oos_ic": -0.0049842352029658915, "oos_wr": 0.5074626865671642, "is_n": 199, "oos_n": 134, "gap_pp": 0.03456575896180757, "down_months": 0.48, "verdict": "FAILS" }, "A5": { "signal": "A5: CUM MM (Cumulative MM Gamma Direction)", "what": "IC of (MM total gamma at 10AM) vs 3H ES return. Positive MM gamma \u2192 market bullish (or contrarian?). Claimed IC=+0.120.", "n_oos": 134, "signal_metric": -0.021728477572651925, "null_baseline": -0.0037611039552940804, "p_value": 0.799, "is_metric": "IC=-0.0436", "oos_metric": "IC=-0.0217, WR=56.7%", "is_ic": -0.043567839195979896, "oos_ic": -0.021728477572651925, "oos_wr": 0.5671641791044776, "is_n": 199, "oos_n": 134, "gap_pp": 0.02183936162332797, "down_months": 0.58, "verdict": "FAILS" }, "A6": { "signal": "A6: INST FLOW DOMINANCE (firm+bd+mm gamma / retail gamma)", "what": "IC of (institutional gamma dominance ratio at 10AM) vs 3H ES return. High inst dominance \u2192 bullish. Claimed WR=71.6%, IC=+0.174. Note: proxy uses TRACE gamma ratio, not DTE-based classification.", "n_oos": 134, "signal_metric": -0.02561381764589785, "null_baseline": 0.007011569984132416, "p_value": 0.76, "is_metric": "IC=-0.0024", "oos_metric": "IC=-0.0256, WR=56.7%", "is_ic": -0.002402923709456373, "oos_ic": -0.02561381764589785, "oos_wr": 0.5671641791044776, "is_n": 199, "oos_n": 134, "gap_pp": 0.023210893936441476, "down_months": 0.64, "verdict": "FAILS" }, "A7": { "signal": "A7: MM GAMMA ASYMMETRY (mm_gamma above / total_mm)", "what": "IC of (mm_above / |mm_total|) vs 3H ES return. More MM gamma above spot \u2192 bearish (contrarian). Sign=-1. Claimed OOS IC=+0.133, Sharpe=1.78.", "n_oos": 134, "signal_metric": -0.0039052904715681075, "null_baseline": -0.003631349058406367, "p_value": 0.959, "is_metric": "IC=-0.0433", "oos_metric": "IC=-0.0039, WR=44.0%", "is_ic": -0.043337901629358924, "oos_ic": -0.0039052904715681075, "oos_wr": 0.44029850746268656, "is_n": 199, "oos_n": 134, "gap_pp": 0.039432611157790814, "down_months": 0.44, "verdict": "FAILS" }, "B1": { "signal": "B1: VIX RATIO MOMENTUM (VIX9D/VIX drops fast \u2192 long ES)", "what": "VIX9D/VIX ratio declining quickly \u2192 long ES next hour. Claimed 64% WR hourly, N=571.", "n_oos": 17, "signal_metric": "N/A \u2014 insufficient intraday VIX history", "null_baseline": "N/A", "p_value": null, "is_metric": "N/A", "oos_metric": "N/A", "down_months": "N/A", "verdict": "INSUFFICIENT DATA (vix_intraday_log.csv has only 17 trading days)", "note": "vix_intraday_log.csv only covers 17 days starting 2026-03-04. Need multi-year intraday VIX data. Cannot test without proper data source." }, "B2": { "signal": "B2: CALM VIX + DXY RISING", "what": "WR when VIX_930 < 18 AND DXY daily return > 0 \u2192 long ES day. Claimed 67% WR, N=149. NOTE: using daily DXY (not hourly) as proxy.", "n_oos": 63, "signal_metric": 0.38095238095238093, "null_baseline": 0.5094920634920634, "p_value": 0.996, "is_metric": "WR=44.2% (N=95)", "oos_metric": "WR=38.1% (N=63)", "down_months": 0.2972972972972973, "verdict": "FAILS", "caveat": "Using daily DXY returns instead of 1H DXY \u2014 may miss intraday signal" }, "B3": { "signal": "B3: DXY UP + ES RESILIENT", "what": "WR when DXY daily return > 0 \u2192 ES positive day. Claimed 67% WR, N=48. NOTE: using daily resolution, not intraday 'ES resilient' filter.", "n_oos": 96, "signal_metric": 0.4479166666666667, "null_baseline": 0.5030833333333333, "p_value": 0.964, "is_metric": "WR=49.0% (N=151), IC=-0.1232", "oos_metric": "WR=44.8% (N=96), IC=-0.1137", "down_months": 0.39344262295081966, "verdict": "FAILS", "caveat": "B3 requires intraday DXY to properly capture 'ES resilient' filter \u2014 daily data is an approximation" }, "C1": { "signal": "C1: FACTORMINER F080 (daily signal)", "what": "IC of F080 normalized signal vs next-day ES return. Claimed IC=0.260, Sharpe=2.12, N=267.", "n_oos": 114, "signal_metric": 0.06722828618294488, "null_baseline": 0.0020125554363014117, "p_value": 0.506, "is_metric": "IC=0.0314, N=170", "oos_metric": "IC=0.0672, Sharpe=1.33, N=114", "f080_z_oos_ic": 0.06722828618294488, "f080_z_p": 0.49, "down_months": 0.4, "verdict": "FAILS" }, "C2": { "signal": "C2: FACTORMINER F042", "what": "F042 signal vs next-day ES return. Claimed IC=0.109, Sharpe=1.62, N=267.", "n_oos": 0, "signal_metric": "N/A", "null_baseline": "N/A", "p_value": null, "is_metric": "N/A", "oos_metric": "N/A", "down_months": "N/A", "verdict": "INSUFFICIENT DATA", "note": "No factorminer-f042-daily.csv file exists in data/. Cannot test without source data." }, "C3": { "signal": "C3: P/C OI RATIO CONTRARIAN", "what": "Extreme P/C OI ratio \u2192 contrarian next day. Claimed Sharpe=1.42, N=86.", "n_oos": 0, "signal_metric": "N/A", "null_baseline": "N/A", "p_value": null, "is_metric": "N/A", "oos_metric": "N/A", "down_months": "N/A", "verdict": "INSUFFICIENT DATA", "note": "No P/C OI ratio data file found in data/. Files checked: []" }, "C4": { "signal": "C4: MAX PAIN PINNING", "what": "Price pins near max pain strike", "n_oos": 0, "signal_metric": "N/A", "null_baseline": "N/A", "p_value": null, "is_metric": "N/A", "oos_metric": "N/A", "down_months": "N/A", "verdict": "INSUFFICIENT DATA \u2014 boolean index did not match indexed array along axis 0; size of axis is 318 but size of corresponding boolean axis is 213" }, "D1": { "signal": "D1: MONDAY LONG (buy open, sell close)", "what": "WR of Monday RTH day return > 0. Claimed 69% WR, N=72.", "n_oos": 64, "signal_metric": 0.640625, "null_baseline": 0.50778125, "p_value": 0.009, "is_metric": "WR=63.4% N=93", "oos_metric": "WR=64.1% N=64", "all_days_wr": { "Monday": "WR=63.7% N=157", "Tuesday": "WR=53.2% N=158", "Wednesday": "WR=51.1% N=135", "Thursday": "WR=45.2% N=155", "Friday": "WR=56.2% N=144" }, "monday_by_year": { "2023": { "wr": 0.8095238095238095, "n": 42 }, "2024": { "wr": 0.5, "n": 52 }, "2025": { "wr": 0.6153846153846154, "n": 52 }, "2026": { "wr": 0.7272727272727273, "n": 11 } }, "down_months": 0.5789473684210527, "verdict": "SURVIVES" }, "D2": { "signal": "D2: 2-DAY REVERSAL (2 down days \u2192 buy next day)", "what": "WR when previous 2 days both negative \u2192 next day positive. Claimed Sharpe=2.13. Inverse test: 2 up days \u2192 sell (should also work if reversal is real).", "n_oos": 64, "signal_metric": 0.46875, "null_baseline": 0.50596875, "p_value": 0.781, "is_metric": "WR=64.5% N=76", "oos_metric": "WR=46.9% N=64", "inverse_test_2up_short": "WR_short=54.9% N=71 p=0.8650 (WR_long=45.1%)", "down_months": 0.4318181818181818, "verdict": "FAILS", "note": "If reversal is real, 2-up \u2192 sell should ALSO work. Check inverse_test." }, "E1": { "signal": "E1: ORB BREAK THRESHOLD ANALYSIS", "what": "Analysis of OR15/OR30 range distributions and break threshold adequacy. Rule 20: WR% must always state threshold, horizon, null baseline. If effective break threshold < 7pts, flag ORB tree as potentially contaminated.", "n_total": 744, "or15_range_median_pts": 12.25, "or15_range_mean_pts": 14.901209677419354, "or15_range_p25": 9.0, "or15_range_p75": 17.75, "or30_range_median_pts": 16.5, "pct_or15_range_below_7pts": 0.09274193548387097, "pct_or15_range_below_10pts": 0.31451612903225806, "double_break_pct": 0.6075268817204301, "clean_break_pct": 0.3924731182795699, "n_oos": 296, "signal_metric": "N/A (analysis only)", "null_baseline": "N/A", "p_value": null, "is_metric": "N/A", "oos_metric": "N/A", "down_months": "N/A", "verdict": "FLAG: 9% of OR15 ranges are below 7pts \u2014 on these days, a 1-tick move outside the range counts as a 'break'. ORB decision tree break_direction_correct WR may be inflated on narrow-range days. Median OR15 range = 12.2pts (safe), but narrow days contaminate the distribution.", "note": "The 'break' in the ORB tree appears to be any move outside the OR, not a threshold-gated break. On days where OR < 7pts, the signal is noise." }, "E2": { "signal": "E2: GEX DIRECTION (gex_tier=DEEP_NEG predicts ORB break direction)", "what": "WR when gex_tier=DEEP_NEG that the ORB15 break direction is correct (or15_break_direction_correct=True). Claimed 68% WR. Testing if DEEP_NEG predicts direction vs just volatility.", "n_oos": 44, "signal_metric": 0.6818181818181818, "null_baseline": 0.5544545454545455, "p_value": 0.053, "is_metric": "WR=55.8% N=52", "oos_metric": "WR=68.2% N=44", "all_tiers_wr_oos": { "LOW_POS": "WR=52.3% N=44", "NEG": "WR=52.5% N=40", "MID_POS": "WR=56.5% N=69", "HIGH_POS": "WR=51.4% N=70", "DEEP_NEG": "WR=68.2% N=44", "EXTREME_POS": "WR=48.3% N=29" }, "deep_neg_avg_range_pts": 33.35227272727273, "other_tiers_avg_range_pts": 17.099206349206348, "down_months": 0.6538461538461539, "verdict": "MARGINAL (p<0.10)", "note": "DEEP_NEG avg OR15 range=33.4pts vs others=17.1pts \u2014 if DEEP_NEG has wider ranges, it may just predict volatility not direction." }, "E3": { "signal": "E3: BUY_PCT BEAR_WEAK (or15_buypct_cat)", "what": "WR when or15_buypct_cat=BEAR_WEAK that the day return is negative. Claimed 80% WR, N=15 OOS. LABELED HYPOTHESIS.", "n_oos": 15, "signal_metric": 0.6666666666666666, "null_baseline": 0.47420000000000007, "p_value": 0.111, "is_metric": "WR=58.8% N=34", "oos_metric": "WR=66.7% N=15", "avg_bear_weak_day_ret_oos": -0.4764321188533289, "down_months": 1.0, "verdict": "HYPOTHESIS (N too small \u2014 N=15 OOS)", "note": "N=15 OOS is below minimum threshold for verdict. Avg day return on BEAR_WEAK OOS: -0.48%." }, "F1": { "signal": "F1: BUY PRESSURE (buy_pct_30m at 10AM, from ES 1-min)", "what": "IC of buy_pct at 30-min mark (9:30-10:00) vs 3H forward return (10AM-1PM). Also: WR when buy_pct > P80 (51.24%). These are the core signals in the buy_pressure_service.py.", "n_oos": 300, "signal_metric": 0.1249427771165479, "null_baseline": -0.00041878083924529473, "p_value": 0.036, "is_metric": "IC=0.0003, WR_hi=51.0% N=104", "oos_metric": "IC=0.1249, WR_hi=64.4% N=59 p=0.0350", "high_buypct_wr_oos": 0.6440677966101694, "high_buypct_p": 0.035, "high_buypct_n": 59, "rvol_30m_ic_oos": 0.039025784297955744, "rvol_30m_p": 0.511, "down_months": 0.6, "verdict": "SURVIVES" } } }