MODEL ADOPTION #841: REJECT — LEDGER BENCH: Apply filter: Phase distance FAR (>10deg) — improves win rate from 31.5% to — E8 Intelligence Research

Abstract

DECISION: REJECT ELEMENT: The +0.5pp win rate improvement from a 10° phase distance filter is statistically fragile (n=659, confidence 65) and the live window "100% win rate" projection is based on 2 removed losers — a sample too small to distinguish signal from noise. The filter also risks rejecting winning trades in trending regimes where phase distance compresses below 10° but the fade still resolves profitably. Over-optimization to a single threshold is likely. EXPECTED IMPACT: Negligible or negative. Removing 10–15% of signals for a +0.5pp win rate gain does not guarantee net PnL improvement — reduced trade frequency may lower absolute returns, and the filter's edge may vanish in regimes where phase distance is structurally low (e.g., high-volatility breakouts). The backtest edge is too thin to justify adding a hard cutoff. RISK: The filter will reject valid trades during compressed-phase trending moves (e.g., strong momentum fades that snap back quickly). This degrades the mod Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com