A. Marginals blend (vol-scaled 10%, EW, OOS 2023+)
Pairwise leg corr (OOS):
pead_d rev_spy gld
pead_d 1.0 -0.0 0.1
rev_spy -0.0 1.0 0.1
gld 0.1 0.1 1.0
| Leg | Sharpe | Ret | MaxDD |
|---|---|---|---|
| pead_d | 1.14 | +44.77% | -10.29% |
| rev_spy | 0.79 | +32.49% | -7.69% |
| gld | 1.36 | +149.26% | -26.40% |
| blend | 1.84 | +52.19% | -5.48% |
Blend vs best leg: 1.84 vs 1.36, p=0.511 -> REJECT
B. BTC/ETH lead-lag (2017+, split 2020)
corr(BTC[t-1], ETH[t]) = -0.045 | corr(ETH[t-1], BTC[t]) = -0.069
| Variant | OOS Sharpe | OOS Ret | OOS MaxDD |
|---|---|---|---|
| trend BTC->ETH | -0.23 | -96.73% | -98.90% |
| reversal BTC->ETH | 0.23 | -59.97% | -95.20% |
| trend ETH->BTC | -0.38 | -93.34% | -95.79% |
| reversal ETH->BTC | 0.38 | +35.93% | -73.51% |
C. Crypto VIX-spike reversal (5bps, split 2020)
| Variant | OOS Sharpe | OOS Ret | OOS MaxDD | p_self |
|---|---|---|---|---|
| BTC spike-reversal | 0.02 | -29.86% | -63.74% | 0.4780 |
| ETH spike-reversal | -0.25 | -77.98% | -83.98% | 0.7582 |