MODEL ADOPTION #2963: REJECT — LEDGER BENCH: Apply filter: Min confirmations >= 5 — improves win rate from 31.3% to 32.9% — E8 Intelligence Research

DECISION: REJECT ELEMENT: No single rule survives — the +1.6pp win-rate gain is not attributable to confirmation count alone, and the filter conflicts with E8's timing geometry (which already encodes confluence via rhythm phase). EXPECTED IMPACT: Win-rate gain would be ~0.0–0.5pp in live, not 1.6pp, because the backtest's n=1431 includes regime-specific noise; the filter removes both losers and winners (as Merlin noted), and the net PnL delta is flat-to-slightly-positive at best. Profit impact: negligible, possibly negative if it blocks high-volatility breakouts that E8 timing already validates. RISK: The filter would reject entries where E8's geometry is strong but confirmation count is low (e.g., early trend reversals with 3–4 confirmations). This degrades the model's ability to catch fast moves, increasing missed-trade opportunity cost and reducing total profit despite a marginal win-rate bump. Also, overfitting to the 2026-08-04 window is likely — the two rejected trades were Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com