MODEL ADOPTION #4587: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 6 — improves win rate from 31.5% to 35.6% — E8 Intelligence Research

DECISION: REJECT ELEMENT: No single rule survives — the ≥6 confirmation filter is a statistical artifact, not an E8-geometry-compatible edge. EXPECTED IMPACT: Win-rate gain of +4.0pp is likely overfit to a 919-sample backtest window; live frequency drop of 15–20% would reduce absolute profit even if win rate improves, because the model's edge is in FADE timing, not confirmation count. RISK: Filter delays entries past the Gods-Clock reversal window, systematically missing early-trend fades that E8 rhythm logic depends on; in low-volatility regimes, missed entries convert to net PnL loss despite higher win rate. TEST: Must pass a 12-month, 3-month-slice backtest where the +4.0pp appears in at least 3 of 4 slices AND the filtered model's total R-multiple (not win rate) exceeds unfiltered by ≥10% — otherwise reject. Also require a 30-day live shadow test where filtered portfolio's max drawdown is lower AND total PnL is not lower than unfiltered. Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com