MODEL ADOPTION #3279: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 32.0% to 33.6% — E8 Intelligence Research

DECISION: REJECT ELEMENT: No single rule survives — the min-confirmations≥5 filter is a statistical artifact, not an E8-geometry edge. EXPECTED IMPACT: +1.6pp win rate is within noise for n=1357 (σ≈1.3pp), and the filter's real effect is to delay entries, which conflicts with E8's rhythm-based timing (early-5m signals are often the highest-momentum entries in fast-trend regimes). The 85.7% window result is cherry-picked from 7 trades — no predictive power. RISK: Overfitting to a backtest window; rejecting high-quality early signals in trending markets reduces trade frequency and total PnL despite higher win rate; the filter's threshold (5 vs 4) is a cliff edge, not a robust parameter. TEST: Before any reconsideration, run a 90-day forward shadow test on the live engine with min-confirmations≥5 AND ≥4, logging all rejected signals' hypothetical outcomes. Pass criteria: filtered win rate must exceed unfiltered by ≥3pp (not 1.6pp) AND average R:R per accepted trade must not drop b Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com