LAB #2839 NEUTRAL: LEDGER BENCH: Apply filter: GBPUSD LONG — improves win rate from 31.2% to 34.6% (+3.5pp, n=52) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: GBPUSD LONG — improves win rate from 31.2% to 34.6% (+3.5pp, n=52) SAME-WINDOW EFFECT: Applying a GBPUSD LONG-only filter would have blocked all 4 GBPUSD SHORT trades in the current window — 2 of which were wins (1.34407 and 1.34461) and 1 loss (1.34508), plus 1 open/rejected. That removes 2 wins and 1 loss from the sample, shifting the live win rate from ~5/9 (55.6%) to ~3/6 (50%) — a net loss of 1 win. PnL impact is negative in this window because the filter would have excluded profitable shorts while adding no new longs (no GBPUSD LONG signals appear in the live list). The backtest's +3.5pp win-rate gain is not reproduced here; it appears to be a directional artifact, not a robust edge. OVER-TIME PROJECTION: Over 3–6 months, a GBPUSD LONG-only filter would systematically remove all GBPUSD SHORT trades from the engine. If the backtest's improvement is real, it implies the model's GBPUSD shorts are Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com