Every hypothesis the system has proposed, and how it tested out. The AI proposes; a deterministic runner decides true/false with the same statistics (two-proportion z · Benjamini-Hochberg FDR · out-of-sample) used across the project. Nothing here changes the live strategy until it is manually promoted.
—
—
Passed the stats (z-test + FDR + out-of-sample). NOT auto-deployed — promotion is a manual decision.
| validated | Strongest-trend trades produce higher net R PASSED: match +0.270R vs rest +0.158R (Δ+0.113R), p=0.000, FDR=sig, OOS consistent, n=1369 trend_strength top_pct 40 → higher net Rproposed 2026-06-21 · tested 2026-06-22 · manual-seed |
| validated | Widest opening ranges produce higher net R PASSED: match +0.266R vs rest +0.160R (Δ+0.106R), p=0.000, FDR=sig, OOS consistent, n=1384 or_range top_pct 40 → higher net Rproposed 2026-06-21 · tested 2026-06-22 · manual-seed |
| validated | Trades with an opening range above 25 points produce higher net R than those at or below 25 points PASSED: match +0.317R vs rest +0.195R (Δ+0.121R), p=0.001, FDR=sig, OOS consistent, n=208 or_range > 25 → higher net Rproposed 2026-06-21 · tested 2026-06-20 · ai_eval |
Tested but not significant, underpowered, or not consistent out-of-sample.
| inconclusive | Short-side trades produce higher net R than long-side trades INCONCLUSIVE: match +0.200R vs rest +0.205R (Δ-0.005R), p=0.990, FDR=ns, OOS consistent, n=1446 side == short → higher net Rproposed 2026-06-21 · tested 2026-06-20 · ai_eval |
Tested and contradicted (effect opposite the claim).
| rejected | Entries at the 10 ET hour produce higher net R than entries at any other hour REJECTED (opposite of claim): match +0.195R vs rest +0.239R (Δ-0.044R), p=0.031, FDR=sig, OOS consistent, n=2844 entry_hour == 10 → higher net Rproposed 2026-06-21 · tested 2026-06-20 · ai_eval |
The Breakout · GC opening-range research. Historical and simulated results for educational purposes only — not financial advice.