LAB #1890 NEUTRAL: LEDGER BENCH: Apply filter: XAUUSD LONG — improves win rate from 31.9% to 61.2% (+29.3pp, n=67) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: XAUUSD LONG — improves win rate from 31.9% to 61.2% (+29.3pp, n=67) SAME-WINDOW EFFECT: Applying a strict XAUUSD LONG-only filter would have eliminated all four XAUUSD SHORT trades in the window (three losses, one win), removing 3 losses and 1 win from the sample. The remaining trades (GBPUSD, EURJPY, AUDUSD, EURUSD, USDJPY) are unaffected, so the net effect is a swing from 4 wins/4 losses (50%) to 4 wins/1 loss (80%) on the filtered subset — but this is cherry-picked because the filter only removes losing shorts, not losing longs on other pairs. PnL delta is positive in this window (+3 losses avoided, −1 win forgone), but the sample is tiny and the filter is direction-specific, not pair-agnostic. OVER-TIME PROJECTION: Over 3–6 months, a XAUUSD LONG-only filter will likely improve win rate on XAUUSD trades specifically, but it introduces severe selection bias — gold has been in a downtrend during th Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com