Historical Validation
Did the model work at prior cutoffs?
The lab avoids a single hero return number. Mean, median, hit rate, concentration, strict coverage, and valuation coverage must be read together.
Cutoffs
17
Top 25 mean excess
+27.6%
Top 25 median excess
+3.2%
Concentration caveat
AAPL appears in all 17 Top-K cutoffs.
Cutoff table
Median and concentration are intentionally adjacent to mean so the result is not over-claimed.
| Cutoff | Top-K | Mean | Median | Hit rate | N | Concentration | Strict coverage | Valuation coverage |
|---|---|---|---|---|---|---|---|---|
| 2022-Q2 | Top 25 | +27.6% | +3.2% | 53.6% | 25 | AAPL present | 61% | 18% |
| 2022-Q2 | Top 50 | +12.4% | -0.2% | 49.1% | 50 | AAPL present | 58% | 15% |
| 2022-Q2 | Top 100 | +6.8% | -4.0% | 45.8% | 100 | AAPL present | 52% | 11% |