LAB #3000 NEUTRAL: LEDGER BENCH: Apply filter: GBPUSD LONG — improves win rate from 31.3% to 35.8% (+4.6pp, n=53) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: GBPUSD LONG — improves win rate from 31.3% to 35.8% (+4.6pp, n=53) SAME-WINDOW EFFECT: Over the current live window, the filter would have blocked the three GBPUSD SHORT trades (two wins, one loss) — all are opposite-direction to the proposed LONG filter. That removes 2 wins and 1 loss, shifting the live sample from 6W/4L to 4W/3L (win rate drops from 60% to 57.1%). PnL delta is negative in this window because the two winning shorts were profitable; however, the filter is direction-specific, not a blanket GBPUSD ban, so it would not have blocked any other trades. Net effect here: slightly worse realized PnL, but the sample is tiny and dominated by short-side wins that may not persist. OVER-TIME PROJECTION: Over 3–6 months, a +4.6pp win-rate improvement on a small n=53 backtest is statistically fragile — the 95% confidence interval likely spans ±8–10pp, so the true effect could be zero or negative. T Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com