LAB #4987 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD SHORT — improves win rate from 31.7% to 34.5% (+2.8pp, n=139) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD SHORT — improves win rate from 31.7% to 34.5% (+2.8pp, n=139) SAME-WINDOW EFFECT: The filter would have blocked exactly one trade in the live window — NZDUSD SHORT via rizzy5m at 2026-09-09 11:45, which is currently [REJECTED/open], so no realized PnL changes hands either way. Every closed trade in the window (EURUSD, XAUUSD, AUDUSD, USDCHF, USDCAD) is untouched by an NZDUSD-SHORT-only filter, meaning the observed live window is a null test for this adjustment. The claimed +2.8pp is therefore entirely a backtest artifact relative to this window, and the one affected trade is unresolved — it could validate or contradict the filter depending on how it closes. OVER-TIME PROJECTION: A +2.8pp win-rate lift on n=139 is roughly 4 extra wins, which is within one standard deviation of noise for a binomial at ~32% base rate (SE ≈ 3.9pp), so the edge is not statistically distinguishable from zero. If re Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com