LAB #2114 NEUTRAL: LEDGER BENCH: Apply filter: Session = LONDON_OPEN — improves win rate from 32.0% to 32.8% (+0.7pp, n — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: Session = LONDON_OPEN — improves win rate from 32.0% to 32.8% (+0.7pp, n=232) SAME-WINDOW EFFECT: Over the current 10-trade window, the LONDON_OPEN filter (roughly 07:00–09:00 UTC, i.e., 09:00–11:00 CEST) would have removed the 2026-08-03 GBPUSD SHORT (11:00 CEST, borderline) and the 2026-07-31 XAUUSD SHORT (09:55 CEST, just inside), but kept the 2026-07-31 EURJPY SHORT (10:35) and most others. Removing the two losses (−1) and keeping the wins would flip the window from 4W/6L to 4W/4L, raising win rate from 40% to 50% — a +10pp swing, far larger than the backtest's +0.7pp, suggesting the filter's effect is highly sample-sensitive. Net PnL delta would be positive (avoiding two full losses), but the small n means this is noise, not signal. OVER-TIME PROJECTION: Over 3–6 months, the +0.7pp improvement (32.0→32.8%) is marginal and likely within sampling error for n=232 (standard error ~3.3pp). The filte Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com