MODEL ADOPTION #4607: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.6% to 33.1% — E8 Intelligence Research

DECISION: REJECT ELEMENT: No single rule survives — the +1.4pp win-rate gain is not tied to E8 geometry timing, and the filter (≥5 confirmations) is a generic latency threshold that conflicts with the model's rhythm logic (which already uses confirmation counts as a dynamic variable, not a static floor). EXPECTED IMPACT: Negligible to negative — the backtest gain is within noise (n=1902, +1.4pp ≈ 27 trades), and the filter would cut early winners that close before 5 confirmations, reducing frequency by 15–20% without a proven R-multiple improvement. The live window shows zero PnL delta on closed trades. RISK: Overfitting to a threshold that may not survive spread/volatility regime shifts; also, filtering on confirmations alone ignores the E8 node phase — a trade with 4 confirmations at a high-probability FADE node may outperform one with 6 confirmations at a weak node. TEST: Before any reconsideration, run a 12-month replay-bench on the exact E8-node/Gods-Clock signal set, spli Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com