Pain Trade Monitor — signal backtest
4,308 reconstructed asset-sessions, 2025-10-02 → 2026-09-25 (420d history, 12 assets). hit = % of signals whose sign matched the forward return; mean_bp = mean signal-aligned return (bp); t_adj = t-stat ÷ √horizon (rough overlap correction; ≥2 / ≤−2 highlighted). Pooled across correlated assets — effective sample is smaller than n. No costs. Calibration evidence, not a strategy.
A1. Pain-direction (trapped shorts ⇒ up, trapped longs ⇒ down) — reconstructed
| method | h | subset | n | hit | mean_bp | t | t_adj |
|---|
| v12 | 1 | all | 4053 | 48.75 | -9.30 | -1.42 | -1.42 |
| v12 | 1 | top tercile | 1434 | 49.02 | -16.78 | -1.54 | -1.54 |
| v12 | 3 | all | 4030 | 48.59 | -25.97 | -2.12 | -1.23 |
| v12 | 3 | top tercile | 1430 | 47.76 | -45.09 | -2.20 | -1.27 |
| v12 | 7 | all | 3989 | 49.03 | -53.02 | -2.74 | -1.04 |
| v12 | 7 | top tercile | 1419 | 47.99 | -80.72 | -2.42 | -0.91 |
| v13d | 1 | all | 4204 | 49.22 | 2.17 | 0.32 | 0.32 |
| v13d | 1 | top tercile | 1433 | 49.34 | -6.20 | -0.50 | -0.50 |
| v13d | 3 | all | 4180 | 48.37 | -3.08 | -0.25 | -0.14 |
| v13d | 3 | top tercile | 1424 | 46.70 | -45.59 | -2.08 | -1.20 |
| v13d | 7 | all | 4132 | 50.05 | 4.33 | 0.22 | 0.08 |
| v13d | 7 | top tercile | 1412 | 48.58 | -44.21 | -1.31 | -0.50 |
| v13 | 1 | all | 4046 | 49.33 | -1.32 | -0.20 | -0.20 |
| v13 | 1 | top tercile | 1436 | 49.09 | -9.09 | -0.78 | -0.78 |
| v13 | 3 | all | 4024 | 48.33 | -18.08 | -1.44 | -0.83 |
| v13 | 3 | top tercile | 1436 | 47.49 | -32.81 | -1.52 | -0.88 |
| v13 | 7 | all | 3988 | 49.62 | -19.21 | -0.98 | -0.37 |
| v13 | 7 | top tercile | 1436 | 49.37 | -85.46 | -2.56 | -0.97 |
A2. Cross-section: long top-3 / short bottom-3 by pain signal, daily
| method | h | days | mean_bp | hit | t_adj |
|---|
| v12 | 1 | 358 | 16.67 | 51.68 | 1.05 |
| v12 | 3 | 356 | 69.66 | 55.34 | 1.37 |
| v12 | 7 | 352 | 138.97 | 59.09 | 1.22 |
| v13d | 1 | 358 | 29.74 | 53.63 | 1.79 |
| v13d | 3 | 356 | 81.41 | 57.58 | 1.49 |
| v13d | 7 | 352 | 105.39 | 53.69 | 0.84 |
| v13 | 1 | 358 | 33.53 | 52.51 | 2.06 |
| v13 | 3 | 356 | 55.85 | 56.46 | 1.18 |
| v13 | 7 | 352 | 93.29 | 52.84 | 0.76 |
A3. Next-session return by regime label
| method | regime | n | fwd1_bp | up% |
|---|
| v12 | New longs (P↑ OI↑) | 1544 | 34.12 | 46.50 |
| v12 | New shorts (P↓ OI↑) | 490 | -38.19 | 45.92 |
| v12 | Short covering (P↑ OI↓) | 442 | 11.53 | 48.64 |
| v12 | Long profit-take (P↓ OI↓) | 1820 | -11.92 | 45.49 |
| v13d | Neutral (P→ / OI→) | 625 | -9.02 | 45.92 |
| v13d | New longs (P↑ OI↑) | 898 | 38.64 | 46.88 |
| v13d | New shorts (P↓ OI↑) | 960 | -18.42 | 46.88 |
| v13d | Short covering (P↑ OI↓) | 785 | 22.98 | 46.75 |
| v13d | Long profit-take (P↓ OI↓) | 1028 | -11.72 | 44.84 |
| v13 | Neutral (P→ / OI→) | 625 | -9.02 | 45.92 |
| v13 | New longs (P↑ OI↑) | 898 | 38.64 | 46.88 |
| v13 | New shorts (P↓ OI↑) | 960 | -18.42 | 46.88 |
| v13 | Short covering (P↑ OI↓) | 785 | 22.98 | 46.75 |
| v13 | Long profit-take (P↓ OI↓) | 1028 | -11.72 | 44.84 |
B. Published logbook calls (52 sessions, 2026-07-16 → 2026-09-25)
| signal | h | n | hit | mean_bp | t | t_adj |
|---|
| offside (published) | 1 | 581 | 47.16 | -27.35 | -1.54 | -1.54 |
| taker−50 (published) | 1 | 612 | 49.35 | 9.78 | 0.54 | 0.54 |
| offside (published) | 3 | 573 | 49.91 | -30.65 | -0.81 | -0.47 |
| taker−50 (published) | 3 | 600 | 48.67 | -18.74 | -0.50 | -0.29 |
| offside (published) | 7 | 543 | 44.38 | -188.37 | -2.85 | -1.08 |
| taker−50 (published) | 7 | 564 | 46.99 | -95.99 | -1.47 | -0.56 |
| rich funding ⇒ down | 1 | 201 | 44.78 | -172.42 | -4.46 | -4.46 |