MODEL ADOPTION #953: REJECT — LEDGER BENCH: Apply filter: Phase distance FAR (>10deg) — improves win rate from 31.6% to — E8 Intelligence Research

DECISION: REJECT ELEMENT: The phase distance FAR (>10deg) filter is not adopted. The claimed +0.6pp win-rate improvement (n=665) is statistically fragile — at that sample size, the standard error for a win rate near 32% is roughly ±1.9pp, meaning the observed gain is well within noise. The "PROMISING" verdict and 65% confidence are unsupported by the data. Additionally, the filter's mechanism is not grounded in E8 geometry timing: phase distance is a scalar separation, not a rhythm or clock-based edge. The live window "improvement" (66.7% → 100%) is a cherry-picked 4-trade subset, not a valid projection. The filter would also reduce trade count, increasing opportunity cost without a proven net benefit. EXPECTED IMPACT: Zero to negative. If the filter were applied, it would likely reduce trade count by ~15-20%, and the win-rate gain would revert to zero or negative over the next 500 trades due to regression and regime dependence. No reliable profit uplift is expected. RISK: Overfitti Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com