MODEL ADOPTION #2632: REJECT — LEDGER BENCH: Apply filter: Phase distance TIGHT (<=2deg) — improves win rate from 31.0% t — E8 Intelligence Research

Abstract

DECISION: REJECT ELEMENT: No single rule survives — the phase-distance TIGHT (≤2°) filter is a statistical artifact of n=441 and the live window's 5W/0L outcome is overfit noise. EXPECTED IMPACT: Win-rate gain would likely shrink to +1–2pp out-of-sample, but trade frequency drops 15–20%, and the filter removes high-reward entries that later converge — net profit likely flat or negative. RISK: Phase-distance is derived from Gods-Clock rhythm parameters; tightening to 2° increases sensitivity to parameter drift, causing regime-dependent signal loss. It also disproportionately kills rizzy5m/rizzy7m signals (mean-reversion entries often start loose-phase), degrading model diversity. TEST: Before any adoption, run a 90-day walk-forward on 1.5°/2°/2.5°/3° thresholds across all three entry models, requiring ≥500 signals per arm, and demand the 2° filter beat unfiltered by ≥3pp win-rate AND ≥+0.5R average R:R with no model-specific signal starvation. Current evidence fails this bar. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com