MODEL ADOPTION #4081: REJECT — LEDGER BENCH: Apply filter: Phase distance TIGHT (<=2deg) — improves win rate from 32.5% t — E8 Intelligence Research

DECISION: REJECT ELEMENT: No single rule survives — the ≤2° phase-distance filter is too tight, regime-dependent, and unproven against E8's own cycle-anchoring logic. EXPECTED IMPACT: Win-rate gain of +5.0pp on n=596 is likely overfit to a narrow window; real-world effect would be ≤+1.5pp after trade-frequency loss (20–30% fewer signals), with net profit flat or negative due to missed high-R trades near boundaries. RISK: Cycle detector drift (session volatility, re-anchoring lag) silently excludes valid wkjudas/rizzy entries; 2° threshold is below E8's noise floor — small calibration shifts cause erratic filter behavior, degrading model coherence. TEST: Must pass a 12-month walk-forward (3-month blocks) with the filter applied to ALL signal types, requiring ≥+3.0pp win-rate improvement AND ≥+0.5R net profit per block, plus a phase-distance distribution test showing winners cluster ≤2° and losers >2° in at least 8 of 12 months. Until then, keep live model unfiltered. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com