Universe 20 style/asset ETFs | top-k=3 EW | monthly rebalance | 5bps/flip | IS<2016-01-01/OOS>= | Bonferroni bar p < 0.0125 over 4 signals.
OOS benchmarks: EW universe Sharpe 0.83 | SPY B&H 0.88
| signal | IS SR | OOS SR | OOS vs EW | placebo p | candidate |
|---|---|---|---|---|---|
| mom63 | 0.43 | 0.53 | -0.30 | 0.7060 | - |
| mom126 | 0.54 | 0.46 | -0.38 | 0.8260 | - |
| rev21 | 0.27 | 0.24 | -0.59 | 0.9920 | - |
| lowvol63 | 0.76 | 0.84 | +0.00 | 0.1320 | - |
Candidates: 0/4
Reading
- Beating EW is the bar that matters: cross-sectional rotation among near-identical betas must add selection, not just beta.
- A raw p<0.05 leg is expected by chance over this many trials; only below-Bonferroni counts. No candidate -> rotation adds nothing.