MOO 9:30 → EXIT 9:55 · ADV 2M+ · v3 Interactive Dashboard
MOO 9:55 Audit + Backtest + 6M Forecast
Equity Curve (based on current controls)
Critical Audit Findings
Linear (fixed size)
Exponential (compounding)
Daily P&L Histogram
Drawdown
Monte Carlo Forecast — Live on Selected Filters
Bootstrap from CURRENT filtered daily P&L. Recomputes on every filter change.
6M Projection Linear — Percentile Bands + Sample Paths
6M Projection Compound — Percentile Bands + Sample Paths
Final Equity Distribution (both modes)
Look-ahead Bias Audit
Проверка каждого фактора: использует ли он данные после момента решения (9:28-9:29)?
Grade Drift Over Years
Full Audit Report
Trades per Day Analysis
Trades per Day Distribution
No-Trade Days by Month
M8 Audit — Errors, Look-aheads, Inconsistencies
Deep integrity check of M8 model: look-ahead bias, pattern discovery overfit, logic inconsistencies, Timeline Pro integration risks.
Novel Patterns — Advanced Edges ★
Systematic exploration of DOW, month, consecutive gaps, VIX regime, SPY/QQQ divergence, 52w position, turn-of-month, PM range. Combined with M7 filters.
Post-Earnings R+1 / R+2 / R+3 Patterns
2,081 earnings events × 5 future days. Looking for MOO→close patterns on post-earnings days with multi-factor filters.
Independent Audit — 3 Perspectives
Каждый аудитор проверяет стратегию с разной стороны: Risk Manager (tail risk), Statistician (mathematical rigor), Practitioner (real-world execution).
Example Trading Days — Manual Verification
Реальные дни с конкретными тикерами, PnL, применёнными фильтрами и скипами. Проверить можно вручную через Datum/Finviz.
3-Model Comparison — Design Trading Strategy for 2024
M1 (Raw) vs M2 (Filtered V2) vs M3 (Best 2024 — all audit findings applied). Both full period and 2024 calendar year.
★★★ LIVE MODEL (trading 2026-04-20): M10 Refined + MKT @ 9:30:00 (2 bps)
User fires MKT orders sharp at 9:30:00, positions filled by 9:30:59. Opening print = ~2 bps spread.
- Live Sharpe expected: 4-5 (backtest 8.73 raw, 4.55 conservative)
- Avg daily: $169 on $100K BP (CAGR 42.6%)
- Annual: $35-42K expected
- Entry cost 2 bps (opening auction spread) + Exit 5 bps + SHORT locate $0.02/share
- Max DD -$1.3K · Max streak 4 days · WR days 76.3%
- M11 MKT-5bps/8bps/10bps variants показаны для sensitivity
Model 10 REFINED (MOO baseline, Sharpe 8.94)
M10 Honest (3 audit fixes: direction from pm_close, exit spread, locate) + only 2 validated SHORT rules.
- SHORT S1: gap_pm -3..-1% AND prev_day<-2% (momentum continuation) — N 3,697 Sharpe 5.90
- SHORT S12: earnings D+1 AND prev<-5% (post-earnings capitulation) — N 59 Sharpe 8.99
- LONG: all M10 Honest logic (direction from pm_close, no lookahead)
- Sharpe raw 8.94 | conservative 4.65 | CAGR 44.1%/year
- Train OOS 8.63 → Test OOS 10.39 (Test BETTER than Train — robust!)
- Max DD -$1,258 (vs M10 Honest -$2,680) | Max streak 4 days
- WR days 76.3% | 22,582 trades | 28.6/day | Avg daily $175
Model 9d — β-excess SHORT filter
One additional hard skip: SHORT trades where stock gaps DOWN significantly MORE than SPY-beta predicts (bounce pattern). Excess gap = gap_pm − beta × SPY_gap.
- SKIP SHORT if excess_gap_spy < -3% (capitulation → mean-reversion UP is likely)
- Sharpe 8.84 (vs M8 8.28, +0.56)
- Train 9.76 → Test OOS 9.70 — only 0.6% degradation (vs M8 11%)
- Compound $100K → $486K (vs M8 $449K)
- 2024 Sharpe 10.08
- Removes ~1,200 bad SHORT trades (WR 45.5%, Sharpe -0.92 pocket eliminated)
★★ Model 8 — M7 + hard skips
4 HARD SKIP rules from pattern research. Same sizing as M7. Result: Sharpe up, drawdown down, overfitting reduced.
- SKIP April entirely (worst month, tariff risk)
- SKIP Monday SHORT (Sharpe 0.36)
- SKIP Consecutive gap DN SHORT (Sharpe -0.25)
- SKIP Tech lags SHORT (QQQ<SPY-0.3, Sharpe -0.46)
- Sharpe raw 12.17 (vs M7 11.65), Conservative 11.29
- Train→Test degradation 1% (vs M7 11%) — robustness 10x better
- Trades/day 17.5 (vs M7 18.6, tighter focus)
★ Model 7 — M6 + SHORT
LONG+SHORT both directions. SHORT restricted to A+/A grades (B+ shorts WR 43% excluded). Same filters, same sizes, same cap 20/day.
- +25% PnL vs M6 ($124.6K vs $99.6K)
- Max DD Linear improved 12x (-$1K vs -$12K)
- Kurtosis 2.08 (near-normal) vs M6 3.74
- Train/Test degradation 11% (M6 was 18%)
- Trades/day 18.6 (LONG 32% / SHORT 68%)