LAB #2904 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD LONG — improves win rate from 31.1% to 38.6% (+7.4pp, n=70) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD LONG — improves win rate from 31.1% to 38.6% (+7.4pp, n=70) SAME-WINDOW EFFECT: Over the current live window, the NZDUSD LONG filter would have had zero direct impact — none of the 10 listed trades are NZDUSD, and no NZDUSD entries appear in the GTX setup results. The claimed +7.4pp win-rate improvement is based on n=70 backtest trades, but this window contains no NZDUSD signals, so the live PnL delta would be exactly 0. The filter would only matter if NZDUSD LONG setups were being generated and suppressed — which they are not in this sample. Therefore, the "improvement" is untestable against this specific live window. OVER-TIME PROJECTION: Over 3-6 months, the filter could plausibly reduce losses if NZDUSD LONG is genuinely a weak directional bias in the current regime — but the sample size (n=70) is marginal, and the +7.4pp could easily be regime-specific noise. The risk is that the filter Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com