MODEL ADOPTION #4022: REJECT — LEDGER BENCH: Apply filter: Phase distance TIGHT (<=2deg) — improves win rate from 32.5% t — E8 Intelligence Research
Abstract
DECISION: REJECT ELEMENT: No element survives — the phase-distance filter (≤2°) is not compatible with E8 geometry timing as a standalone rule, and its backtest edge is not proven against live entry conditions. EXPECTED IMPACT: Win-rate improvement is likely illusory — the +5.2pp on n=595 is within noise for a 40–60% trade-frequency reduction, and the filter would likely exclude high-R:R trades that carry the model's net profit. Net effect: lower total PnL, higher variance per trade. RISK: Overfitting to a specific regime (trending vs. ranging); look-ahead bias if phase distance is computed with future data; live entries (e.g., GBPUSD LONG at 1.35301) already show the filter would have excluded winners, not just losers — degrading the model's core fade polarity edge. TEST: Before any reconsideration, run a 30-day shadow test on the live engine logging both filtered and unfiltered signals with outcomes, plus a walk-forward split of the 595 trades into 3 sequential chunks — requi Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com