LAB #4890 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD LONG — improves win rate from 31.9% to 54.5% (+22.7pp, n=121) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD LONG — improves win rate from 31.9% to 54.5% (+22.7pp, n=121) SAME-WINDOW EFFECT: The filter only touches NZDUSD LONG signals, and the current window contains exactly one NZDUSD trade — a SHORT (rizzy5m, 2026-09-09) that was already rejected. So zero live trades change: no entries added, no losses avoided, no PnL delta. The +22.7pp headline is therefore untested by this window and cannot be credited or debited here. Note the asymmetry worth flagging: the filter is defined on LONG, while the engine's recent NZDUSD behavior is short-biased. OVER-TIME PROJECTION: If the 54.5% vs 31.9% split is real and stable, NZDUSD longs go from a coin-flip-losing leg to a modest edge, which over 3-6 months is worth real expectancy — but only in proportion to how often NZDUSD LONG signals actually fire, which the current window suggests may be rare. The dominant risk is n=121 with a 22.7pp jump: that is a smal Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com