当前硬结论:keep_P1
生成时间:2026-03-21 02:07 UTC | 口径:BTC/ETH/SOL perpetual, 15m, next-bar open, no-overlap, hold 8 bars, costs 6/10/15bps
三臂只比较:baseline_no_expiry、confirm_window_12、confirm12_entry24。
这轮直接把『无限等待』改成有时间预算的 replication 跑完了:如果 confirmWindow 或 confirm+entryWindow 不能在成本后同时改善失败率与 expectancy,就该尽快 park,而不是继续把它写成漂亮但不落地的 honesty 故事。
主要短板:改善还不够统一,暂时只算值得再给 1 次会改变 verdict 的检查。
| metric | value |
|---|---|
| chosen_variant | confirm12_entry24 |
| return_delta_6bps | 0.00 |
| failure_delta_6bps | -0.25 |
| retention_vs_baseline | 0.66 |
| positive_assets_6bps | 1.00 |
| positive_setups_6bps | 1.00 |
| positive_costs | 1.00 |
| mean_time_to_entry_bars | 7.22 |
| recommended_action | keep_P1 |
| why_now | 这轮直接把『无限等待』改成有时间预算的 replication 跑完了:如果 confirmWindow 或 confirm+entryWindow 不能在成本后同时改善失败率与 expectancy,就该尽快 park,而不是继续把它写成漂亮但不落地的 honesty 故事。 |
| main_weakness | 改善还不够统一,暂时只算值得再给 1 次会改变 verdict 的检查。 |
| split | variant | cost_bps_per_side | baseline_trades | variant_trades | trade_count_retention | baseline_return | variant_return | return_delta | baseline_failure | variant_failure | failure_delta | baseline_time_to_entry_bars | variant_time_to_entry_bars |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| train | confirm_window_12 | 6 | 490 | 355 | 72.45% | -0.05% | -0.02% | 0.03% | 49.18% | 27.61% | -21.58% | 1.0 | 4.0 |
| train | confirm_window_12 | 10 | 490 | 355 | 72.45% | -0.13% | -0.10% | 0.03% | 49.18% | 27.61% | -21.58% | 1.0 | 4.0 |
| train | confirm_window_12 | 15 | 490 | 355 | 72.45% | -0.23% | -0.20% | 0.03% | 49.18% | 27.61% | -21.58% | 1.0 | 4.0 |
| train | confirm12_entry24 | 6 | 490 | 333 | 67.96% | -0.05% | -0.06% | -0.01% | 49.18% | 29.73% | -19.45% | 1.0 | 6.7 |
| train | confirm12_entry24 | 10 | 490 | 333 | 67.96% | -0.13% | -0.14% | -0.01% | 49.18% | 29.73% | -19.45% | 1.0 | 6.7 |
| train | confirm12_entry24 | 15 | 490 | 333 | 67.96% | -0.23% | -0.24% | -0.01% | 49.18% | 29.73% | -19.45% | 1.0 | 6.7 |
| test | confirm_window_12 | 6 | 328 | 236 | 71.95% | -0.16% | 0.00% | 0.17% | 55.18% | 27.12% | -28.06% | 1.0 | 4.2 |
| test | confirm_window_12 | 10 | 328 | 236 | 71.95% | -0.24% | -0.08% | 0.17% | 55.18% | 27.12% | -28.06% | 1.0 | 4.2 |
| test | confirm_window_12 | 15 | 328 | 236 | 71.95% | -0.34% | -0.18% | 0.17% | 55.18% | 27.12% | -28.06% | 1.0 | 4.2 |
| test | confirm12_entry24 | 6 | 328 | 217 | 66.16% | -0.16% | 0.02% | 0.19% | 55.18% | 29.95% | -25.23% | 1.0 | 7.2 |
| test | confirm12_entry24 | 10 | 328 | 217 | 66.16% | -0.24% | -0.06% | 0.19% | 55.18% | 29.95% | -25.23% | 1.0 | 7.2 |
| test | confirm12_entry24 | 15 | 328 | 217 | 66.16% | -0.34% | -0.16% | 0.19% | 55.18% | 29.95% | -25.23% | 1.0 | 7.2 |
| split | setup | variant | baseline_trades | variant_trades | trade_count_retention | baseline_return | variant_return | return_delta | baseline_failure | variant_failure | failure_delta | variant_time_to_confirm_bars | variant_time_to_entry_bars |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| train | breakout_short | confirm_window_12 | 409 | 296 | 72.37% | -0.02% | 0.02% | 0.04% | 51.83% | 28.72% | -23.12% | 3.0 | 4.0 |
| train | fib_retest_long | confirm_window_12 | 20 | 17 | 85.00% | 0.31% | 0.06% | -0.25% | 30.00% | 17.65% | -12.35% | 4.1 | 5.1 |
| train | ema_psar_long | confirm_window_12 | 61 | 42 | 68.85% | -0.37% | -0.35% | 0.02% | 37.70% | 23.81% | -13.90% | 2.9 | 3.9 |
| train | breakout_short | confirm12_entry24 | 409 | 279 | 68.22% | -0.02% | -0.04% | -0.03% | 51.83% | 30.47% | -21.37% | 2.8 | 6.9 |
| train | fib_retest_long | confirm12_entry24 | 20 | 17 | 85.00% | 0.31% | 0.24% | -0.07% | 30.00% | 17.65% | -12.35% | 3.2 | 5.5 |
| train | ema_psar_long | confirm12_entry24 | 61 | 37 | 60.66% | -0.37% | -0.31% | 0.06% | 37.70% | 29.73% | -7.98% | 2.9 | 5.5 |
| test | breakout_short | confirm_window_12 | 286 | 206 | 72.03% | -0.20% | -0.10% | 0.10% | 59.09% | 30.10% | -28.99% | 3.3 | 4.3 |
| test | fib_retest_long | confirm_window_12 | 5 | 5 | 100.00% | -0.05% | -0.45% | -0.40% | 40.00% | 20.00% | -20.00% | 3.0 | 4.0 |
| test | ema_psar_long | confirm_window_12 | 37 | 25 | 67.57% | 0.10% | 0.95% | 0.85% | 27.03% | 4.00% | -23.03% | 2.6 | 3.6 |
| test | breakout_short | confirm12_entry24 | 286 | 195 | 68.18% | -0.20% | 0.03% | 0.23% | 59.09% | 30.77% | -28.32% | 3.2 | 7.3 |
| test | fib_retest_long | confirm12_entry24 | 5 | 4 | 80.00% | -0.05% | -0.12% | -0.07% | 40.00% | 50.00% | 10.00% | 2.8 | 5.8 |
| test | ema_psar_long | confirm12_entry24 | 37 | 18 | 48.65% | 0.10% | -0.02% | -0.12% | 27.03% | 16.67% | -10.36% | 2.6 | 7.1 |
| split | asset | variant | baseline_trades | variant_trades | trade_count_retention | baseline_return | variant_return | return_delta | baseline_failure | variant_failure | failure_delta |
|---|---|---|---|---|---|---|---|---|---|---|---|
| train | BTC-USD | confirm_window_12 | 162 | 120 | 74.07% | -0.06% | -0.04% | 0.02% | 50.62% | 30.00% | -20.62% |
| train | ETH-USD | confirm_window_12 | 160 | 104 | 65.00% | -0.11% | 0.08% | 0.19% | 53.12% | 25.00% | -28.12% |
| train | SOL-USD | confirm_window_12 | 168 | 131 | 77.98% | 0.03% | -0.08% | -0.11% | 44.05% | 27.48% | -16.57% |
| train | BTC-USD | confirm12_entry24 | 162 | 117 | 72.22% | -0.06% | -0.09% | -0.02% | 50.62% | 32.48% | -18.14% |
| train | ETH-USD | confirm12_entry24 | 160 | 97 | 60.62% | -0.11% | 0.12% | 0.23% | 53.12% | 24.74% | -28.38% |
| train | SOL-USD | confirm12_entry24 | 168 | 119 | 70.83% | 0.03% | -0.18% | -0.20% | 44.05% | 31.09% | -12.96% |
| test | BTC-USD | confirm_window_12 | 111 | 76 | 68.47% | -0.26% | -0.05% | 0.21% | 60.36% | 27.63% | -32.73% |
| test | ETH-USD | confirm_window_12 | 101 | 74 | 73.27% | -0.18% | 0.02% | 0.21% | 52.48% | 29.73% | -22.75% |
| test | SOL-USD | confirm_window_12 | 116 | 86 | 74.14% | -0.05% | 0.04% | 0.09% | 52.59% | 24.42% | -28.17% |
| test | BTC-USD | confirm12_entry24 | 111 | 72 | 64.86% | -0.26% | -0.02% | 0.24% | 60.36% | 30.56% | -29.80% |
| test | ETH-USD | confirm12_entry24 | 101 | 68 | 67.33% | -0.18% | -0.07% | 0.12% | 52.48% | 30.88% | -21.59% |
| test | SOL-USD | confirm12_entry24 | 116 | 77 | 66.38% | -0.05% | 0.15% | 0.20% | 52.59% | 28.57% | -24.01% |
| split | variant | median_time_to_confirm_bars | p75_time_to_confirm_bars | median_time_to_entry_bars | p75_time_to_entry_bars |
|---|---|---|---|---|---|
| train | confirm_window_12 | 2.0 | 4.0 | 3.0 | 5.0 |
| train | confirm12_entry24 | 2.0 | 4.0 | 5.0 | 8.0 |
| test | confirm_window_12 | 2.0 | 4.0 | 3.0 | 5.0 |
| test | confirm12_entry24 | 2.0 | 4.0 | 5.0 | 10.0 |