Historical performance simulation — 60 stocks, 252 trading days, updated nightly.
| Quartile | N | Win % | Avg Ret | Score Range |
|---|---|---|---|---|
| Q1 (Low) | 63 | 58.7% | +0.425% | 0.528–0.764 |
| Q2 | 63 | 41.3% | +0.108% | 0.765–0.796 |
| Q3 | 63 | 58.7% | +0.678% | 0.796–0.824 |
| Q4 (High) | 63 | 42.9% | -0.416% | 0.824–0.864 |
The backtest period split into 4 equal windows. Consistent performance across all periods suggests genuine edge rather than luck in one stretch.
| Period | Dates | N | Win % | Avg Ret | Total | SPY Total |
|---|---|---|---|---|---|---|
| Period 1 | 2025-06-12 – 2025-09-11 | 63 | 49.2% | +0.072% | +4.5% | +9.4% |
| Period 2 | 2025-09-12 – 2025-12-10 | 63 | 50.8% | +0.309% | +19.5% | +4.9% |
| Period 3 | 2025-12-11 – 2026-03-13 | 63 | 58.7% | -0.001% | -0.1% | -3.3% |
| Period 4 | 2026-03-16 – 2026-06-12 | 63 | 50.8% | +0.414% | +26.1% | +11.9% |
2,000 simulations randomly resampling the same return pool. If the actual total return falls in a high percentile, the pick order (skill) contributed meaningfully — not just the underlying stock returns.