| tf | L | H | basis | turn/bar | full SR | OOS SR | b/e bps | net@1bp | net@2bp |
|---|
| 60 | 1 | 6 | close | 0.88 | +1.69 | +1.68 | 1.85 | +0.87 | +0.05 |
| 60 | 1 | 6 | lw | 0.88 | +1.42 | +1.60 | 1.58 | +0.78 | -0.04 |
| 60 | 1 | 6 | hlmid | 0.87 | +1.35 | +1.50 | 1.49 | +0.68 | -0.15 |
| 30 | 2 | 6 | close | 0.67 | +1.80 | +1.63 | 1.36 | +0.46 | -0.71 |
| 30 | 2 | 6 | lw | 0.67 | +1.51 | +1.51 | 1.16 | +0.34 | -0.84 |
| 30 | 2 | 6 | hlmid | 0.67 | +1.39 | +1.39 | 1.06 | +0.21 | -0.97 |
| 60 | 1 | 3 | close | 1.30 | +1.85 | +1.75 | 1.35 | +0.52 | -0.72 |
| 60 | 1 | 3 | lw | 1.30 | +1.56 | +1.61 | 1.15 | +0.37 | -0.87 |
| 60 | 1 | 3 | hlmid | 1.30 | +1.45 | +1.51 | 1.07 | +0.27 | -0.98 |
| Year | bars | gross SR | gross bps/bar | b/e bps | net@1bp SR |
|---|
| 2020 | 772 | +2.38 | +2.311 | 2.55 | +1.46 |
| 2021 | 1762 | +3.11 | +2.545 | 2.81 | +2.01 |
| 2022 | 1755 | +0.95 | +1.055 | 1.20 | +0.16 |
| 2023 | 1746 | +1.19 | +0.834 | 0.95 | -0.05 |
| 2024 | 1752 | +1.74 | +1.497 | 1.72 | +0.74 |
| 2025 | 1737 | +1.75 | +1.839 | 2.11 | +0.92 |
| 2026 | 1309 | +1.38 | +1.646 | 2.00 | +0.69 |
| bps one-way | full SR | OOS SR | OOS ret |
|---|
| 0.5 | +1.23 | +1.27 | +71.3% |
| 1 | +0.78 | +0.87 | +42.3% |
| 1.5 | +0.32 | +0.46 | +18.3% |
| 2 | -0.14 | +0.05 | -1.7% |
| 3 | -1.05 | -0.76 | -32.1% |
| 4 | -1.96 | -1.58 | -53.2% |
| K | turn/bar | full SR | b/e bps | net@1bp | net@2bp |
|---|
| 5 | 0.94 | +1.24 | 2.35 | +0.91 | +0.47 |
| 10 | 0.94 | +1.62 | 2.24 | +1.02 | +0.40 |
| 20 | 0.88 | +1.69 | 1.85 | +0.87 | +0.05 |
| 30 | 0.79 | +1.58 | 1.88 | +0.47 | -0.44 |
| 40 | 0.69 | +1.50 | 1.67 | +0.31 | -0.63 |
| Universe | names | turn/bar | full SR | b/e bps | net@1bp | net@2bp |
|---|
| live 101 | 101 | 0.88 | +1.69 | 1.85 | +0.87 | +0.05 |
| all cached | 112 | 0.78 | +1.53 | 1.91 | +0.97 | +0.16 |
- Grid trials: 50 (L x H x sign, 60m).
- White Reality Check p (best of 50 survives data-mining): 0.315 (t_obs 1.93, best
R1_6).
- Winner net@1bp OOS — SR 0.87, PSR 0.92, DSR(K=50) 0.00, haircut SR 0.00, MinTRL infy (skew 1.20, kurt 29.2).
Read: if the b/e survives on VWAP/mid, the sign is positive every year, and the Reality-Check p is small, the winner is a real (if thin) edge. If it collapses on VWAP or the RC p is large, it was bounce or the max of a 50-cell grid.