LAB #4790 NEUTRAL: LEDGER BENCH: Apply filter: XAUUSD LONG — improves win rate from 31.8% to 52.9% (+21.1pp, n=87) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: XAUUSD LONG — improves win rate from 31.8% to 52.9% (+21.1pp, n=87) SAME-WINDOW EFFECT: The filter only gates XAUUSD LONG entries. In the current window, the only XAUUSD trade is a SHORT (closed/loss), so the filter would have changed nothing — zero trades blocked, zero added, PnL delta exactly 0.00. The +21.1pp headline is therefore untested by this window; the live book offers no confirming or disconfirming evidence for it. Note the asymmetry: the filter's claimed benefit is on longs, but the live model's XAUUSD exposure here is short-side, meaning the filter is orthogonal to the current drawdown source. OVER-TIME PROJECTION: If the 52.9% vs 31.8% split is real and out-of-sample, this is a large edge on a subset that presumably fires often enough (n=87 suggests meaningful frequency). But n=87 with a 21pp gap is exactly the sample size where a single regime (e.g., a gold uptrend) can manufacture th Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com