BTC 2013-01-01..2026-09-20 | held window [t,t+1] | prior completed factor session | IS<2021-01-01/OOS>= | 15bps/flip | lag-shift placebo 2..30.
PRIMARY = UUP (dollar) prior-session return < 0 -> long BTC.
| leg | exposure | IS net SR | OOS net SR | OOS B&H | net DD | OOS IC | dir-part-IC | flips |
|---|---|---|---|---|---|---|---|---|
| UUP high | 50% | 0.91 | -0.03 | 0.60 | -79% | -0.010 | +0.008 | 1772 |
| UUP low * | 46% | 0.74 | -0.31 | 0.60 | -88% | -0.010 | -0.008 | 1751 |
| GLD high | 53% | 0.50 | -0.35 | 0.60 | -95% | -0.000 | +0.016 | 1787 |
| GLD low | 46% | 1.04 | 0.26 | 0.60 | -67% | -0.000 | -0.016 | 1784 |
| TLT high | 52% | 1.31 | -0.38 | 0.60 | -83% | -0.008 | +0.007 | 1820 |
| TLT low | 47% | 0.17 | 0.29 | 0.60 | -94% | -0.008 | -0.007 | 1821 |
| SPY high | 56% | 1.16 | -0.08 | 0.60 | -84% | -0.031 | -0.022 | 1753 |
| SPY low | 44% | 0.37 | 0.03 | 0.60 | -91% | -0.031 | +0.022 | 1750 |
| VIX high | 41% | 0.40 | 0.22 | 0.60 | -93% | +0.033 | +0.008 | 1786 |
| VIX low | 58% | 1.12 | -0.27 | 0.60 | -87% | +0.033 | -0.008 | 1777 |
| TNX high | 49% | 0.57 | 0.06 | 0.60 | -82% | +0.013 | -0.004 | 1785 |
| TNX low | 50% | 0.98 | -0.19 | 0.60 | -77% | +0.013 | +0.004 | 1790 |
| MOM365 high (control) | - | - | 0.89 | 0.60 | - | - | - | - |
Placebo (UUP low, lag-shift null)
- OOS obs -0.31 vs null mean -0.02/p95 0.50 -> p 0.931
- IS obs 0.74 vs null mean 0.80 -> p 0.483
- DSR(K=12) OOS net vs B&H 0.60: 0.01 | haircut -0.31
Verdict: REJECT
- PRIMARY (UUP low): OOS net-vs-BH -0.91, placebo p 0.931, IS edge -0.81, dir-partial-IC -0.008, net -0.31 vs MOM365 0.89
- REJECT -> the anti-dollar / macro channel does not time BTC at the daily horizon net of the placebo + momentum controls; factor feeds stay research infrastructure. No live wiring.