LAB #4077 NEUTRAL: LEDGER BENCH: Apply filter: XAUUSD LONG @ NY_OPEN with 5 conf — improves win rate from 32.6% to 76.2 — E8 Intelligence Research

Abstract

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: XAUUSD LONG @ NY_OPEN with 5 conf — improves win rate from 32.6% to 76.2% (+43.6pp, n=21) SAME-WINDOW EFFECT: Over the current live window, this filter would have had **zero effect** — none of the 10 listed trades are XAUUSD LONG entries at NY_OPEN with 5 conf. The only XAUUSD trade present is a SHORT via rizzy5m (entry 09:15), which the filter does not touch. Therefore, no trades would be added, removed, or altered; win rate and PnL delta remain unchanged at 0%. The claimed +43.6pp improvement is not observable in this sample because the filter's target condition simply never occurred. OVER-TIME PROJECTION: Over 3–6 months, the filter could meaningfully improve XAUUSD long performance *if* the 21-sample backtest is representative — but that is a fragile assumption. The risk is severe selection bias: n=21 is tiny, and the +43.6pp jump suggests the filter may be overfitting to a specific market regim Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com