LAB #4895 NEUTRAL: LEDGER BENCH: Apply filter: XAUUSD LONG — improves win rate from 31.9% to 54.1% (+22.3pp, n=85) — 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 54.1% (+22.3pp, n=85) SAME-WINDOW EFFECT: The filter is XAUUSD LONG only, but every gold trade in the live window is XAUUSD SHORT (4393.01, closed loss). So the filter would have blocked zero losing trades and added zero winning trades in this window — the live book is unchanged. The claimed +22.3pp is therefore entirely invisible in the current sample; nothing in the window corroborates or contradicts it. Net PnL delta over this window: 0.00, with the caveat that the one gold loss remains. OVER-TIME PROJECTION: If the 54.1% win rate on n=85 is real and not a small-sample artifact, restricting XAUUSD to longs would meaningfully lift gold expectancy over 3-6 months — but n=85 with a 22pp swing is exactly the sample size where a handful of trades can manufacture a regime artifact (e.g., gold trending up over the test period). The bigger risk is opportunity Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com