Equal-weight long ['BTC', 'ETH', 'SOL', 'XRP', 'LINK', 'AVAX', 'DOGE', 'ADA'] | target vol 40% | w = clip(target/exp(forecast), 0, 1.5) | trend gate = close>SMA100 | IS < 2025-01-01 <= OOS | no costs
| variant | IS Sharpe | IS maxDD | OOS Sharpe | OOS ret | OOS maxDD |
|---|---|---|---|---|---|
| unsized EW (beta) | +2.12 | -30.1% | -0.28 | -51.9% | -71.8% |
| vol-target HAR | +1.49 | -21.3% | -0.21 | -24.9% | -54.5% |
| vol-target HAR+flow | +1.42 | -21.9% | -0.17 | -23.6% | -54.2% |
| trend + vol-target HAR | +1.54 | -12.5% | +0.19 | +2.8% | -23.5% |
| trend + vol-target HAR+flow | +1.47 | -12.3% | +0.23 | +4.5% | -23.3% |
The only difference between the HAR and HAR+flow rows is the sizing forecast. A real gain = higher OOS Sharpe or lower maxDD at equal exposure policy.