Universe: live 101-name sleeve universe. L/S deciles, hold=21, next-open, 10bps, sqrt(252). Split 2016-01-01. PASS bar: OOS Sharpe >= 0.50 AND bootstrap p < 0.05 AND residual Sharpe vs market > 0.
Arm Seg Ret Sharpe MaxDD CorrMkt ResidSh boot-p
A1 MAX-lottery IS -0.905644 -0.823183 -0.920399 -0.614818 NaN NaN A1 MAX-lottery OOS -0.903638 -0.907149 -0.925008 -0.438535 -0.490365 0.9982 A2 IVOL IS -0.909648 -0.853064 -0.918325 -0.528627 NaN NaN A2 IVOL OOS -0.934420 -1.072212 -0.950605 -0.376533 -0.725073 0.9996 A3 overnight-gap IS -0.301991 -0.127394 -0.621640 0.250519 NaN NaN A3 overnight-gap OOS -0.714233 -0.475392 -0.767272 0.023687 -0.500728 0.9364 A4 52w-high IS -0.844657 -0.607883 -0.919830 -0.602015 NaN NaN A4 52w-high OOS -0.832099 -0.558248 -0.853131 -0.387706 -0.158238 0.9676
ALL DONE
Reviewer note
- All four arms REJECT on the pre-registered beta-audit bar (OOS resid Sharpe <= 0), even though the L/S decile returns are significant.
- Sign read: A1 (MAX) and A2 (IVOL) L/S earn their return almost entirely as -market-beta (corrMkt ~ -0.5, bull-market sample): the short leg is short high-beta names, which lost. A3 (overnight-gap) has corrMkt ~ 0 and resid -0.50: no alpha. A4 (52w-high) has resid -0.16.
- Long-only controls (sanity, same harness, gross-beta): A1 0.83, A2 0.82, A3 0.64, A4 0.75 OOS Sharpe, ret ~2.3-2.8x � these are plain equity beta, not a new sleeve.
- Conclusion: none of the four documented effects survives as an orthogonal L/S factor on the live 101-name universe at 10bps. Parked; do not sleeve.