Basket 28 ETFs, 5025 days (2006-09-29..2026-09-22), avg active 15.3/day. Live state: close>SMA200 & ROC60>0, inverse-vol weighted. OOS >= 2016-01-01, N perms = 300. Rule: REJECT if placebo p>=0.05 (trend state no better than random basket selection).
- observed OOS Sharpe 0.77 (full 0.72)
- null mean 0.88, p95 1.13
- empirical placebo p = 0.743
Verdict: REJECT
- SIGNAL -> the dual-trend state selects the right ETFs; REJECT -> the same number of random basket names does as well.