# Decision

- Decision: `review`
- Best candidate: `momentum_60d (momentum, lookback=60)`
- Reason: bias pass=True, net Sharpe=0.409, net MaxDD=-29.51%

## Candidate Ranking

| Candidate | Bias | Gross Sharpe | Net Sharpe | Net MaxDD | TC Drag |
|---|---:|---:|---:|---:|---:|
| momentum_60d (momentum, lookback=60) | PASS | 0.534 | 0.409 | -29.51% | 1.4338% |
| vol_adj_momentum_60d (vol_adjusted_momentum, lookback=60) | PASS | 0.386 | 0.256 | -37.04% | 1.4444% |
| momentum_20d (momentum, lookback=20) | PASS | 0.266 | 0.046 | -37.45% | 2.4247% |
| mean_reversion_5d (mean_reversion, lookback=5) | FAIL | 0.093 | -0.341 | -38.96% | 4.8468% |

## Review Notes

- 这轮结果只证明 loop 跑通，不证明真实 alpha。
- 下一步应替换合成收益为 qsd-backtest 或真实 backtest 输出。
- Promotion 前必须加 hidden holdout 和与现有策略相关性检查。