LAB #3972 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD LONG — improves win rate from 32.5% to 55.7% (+23.2pp, n=115) — E8 Intelligence Research

Abstract

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD LONG — improves win rate from 32.5% to 55.7% (+23.2pp, n=115) SAME-WINDOW EFFECT: Applying the NZDUSD LONG-only filter would have blocked the two NZDUSD SHORT trades (wkjudas and rizzy5m) in the current window — one open, one closed as a loss. It would also have allowed the one NZDUSD LONG (wkjudas, 0.59515) which is currently marked [CLOSED/open] — ambiguous status, but if it was a win, that adds +1 win. Net effect: removes 2 shorts (1 loss, 1 unknown), adds 0 new trades, so win rate in this window improves from ~2/8 (25%) to ~3/6 (50%) — but sample is tiny and the open short could still close as a win, so the real delta is uncertain. OVER-TIME PROJECTION: Over 3–6 months, a +23.2pp win-rate jump on n=115 sounds large but is likely overfit to a specific regime (NZDUSD trending down during that backtest window). If the filter is simply "long-only on NZDUSD," it will fail when NZDUSD enters a Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com