Live 3-sleeve book (flagship/pead/crypto) | monthly rebalance | trailing 63d vol | IS<2023-06-01/OOS>= | 252 annualization.
| sleeve | OOS Sharpe | OOS ret | OOS maxDD |
|---|
| flagship | 0.89 | +62.5% | -28.1% |
| pead | 1.02 | +44.7% | -12.4% |
| crypto | 0.67 | +57.9% | -27.8% |
| scheme | IS Sharpe | OOS Sharpe | OOS ret | OOS maxDD | p vs equal |
|---|
| equal (live) | 1.56 | 1.29 | +61.9% | -13.6% | - |
| invvol | 0.99 | 1.44 | +64.1% | -15.8% | 0.460 |
| invvol_capped | 0.98 | 1.47 | +66.8% | -15.8% | 0.386 |
| erc | 1.37 | 1.11 | +52.2% | -16.1% | 0.659 |
| scheme | flagship | pead | crypto |
|---|
| equal | 33% | 33% | 33% |
| invvol | 32% | 50% | 20% |
| invvol_capped | 34% | 49% | 20% |
| erc | 26% | 36% | 39% |
- best vol scheme
invvol_capped: OOS Sharpe 1.47 vs equal 1.29; OOS ret +66.8% vs +61.9%; p 0.386
- Vol-weighting tilts capital toward the low-vol sleeves (flagship) and away from crypto, which is the return engine; equal-weight keeps the return. Unless a vol scheme clears the OOS Sharpe+return+ significance bars, the live equal-weight allocator is unchanged (
compute_weights('invvol') stays a stub).