LAB #2824 NEUTRAL: LEDGER BENCH: Apply filter: XAUUSD SHORT @ NY_OPEN with 5 conf — improves win rate from 31.2% to 58. — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: XAUUSD SHORT @ NY_OPEN with 5 conf — improves win rate from 31.2% to 58.3% (+27.2pp, n=12) SAME-WINDOW EFFECT: Over the current live window, the filter would have blocked the XAUUSD SHORT at 4048.74 (2026-08-04, LOSS) and the XAUUSD SHORT at 4051.79 (2026-07-31, LOSS), both of which were not NY_OPEN entries. The XAUUSD LONG at 4270.17 (2026-08-06) would remain unaffected since it is a LONG, not a SHORT. This would have removed 2 losses from the 10-trade window, shifting the win rate from 5/10 (50%) to 5/8 (62.5%), a +12.5pp improvement. PnL delta is positive but modest — roughly +2R assuming average loss size, though the sample is tiny and the two removed trades were both rizzy5m entries, not necessarily NY_OPEN. OVER-TIME PROJECTION: Over 3-6 months, the filter's value depends on whether NY_OPEN + 5 conf genuinely selects for higher-probability XAUUSD SHORTs or merely overfits to a 12-trade backtes Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com