LAB #4998 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD LONG @ NY_OPEN with 6 conf — improves win rate from 31.7% to 82.8 — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD LONG @ NY_OPEN with 6 conf — improves win rate from 31.7% to 82.8% (+51.0pp, n=29) SAME-WINDOW EFFECT: This filter would have blocked the NZDUSD SHORT on 2026-09-09 (rizzy5m, entry 0.58415) — but that trade was already REJECTED, so no live PnL changes in the current window. The filter is direction-specific (LONG only), and every NZDUSD trade in this window is SHORT, so the net effect on the listed trades is zero. The claimed +51pp improvement is therefore untested by this window and cannot be corroborated by it. OVER-TIME PROJECTION: If the 82.8% win rate holds on n=29, this is a high-value conditional edge — but n=29 is far too small to trust, and a +51pp jump from a single session/time filter is the classic signature of overfitting to a handful of lucky NY_OPEN sessions. Over 3-6 months, expect regression toward the base rate (likely 40-55%), with the real risk being that the filter suppres Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com