LAB #1818 NEUTRAL: LEDGER BENCH: Apply filter: USDCHF LONG @ LONDON_OPEN with 6 conf — improves win rate from 31.9% to — E8 Intelligence Research

Abstract

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: USDCHF LONG @ LONDON_OPEN with 6 conf — improves win rate from 31.9% to 63.2% (+31.2pp, n=19) SAME-WINDOW EFFECT: Over the current live window, none of the 10 listed trades are USDCHF LONG at London open, so this filter would have changed nothing in the actual trade set. The claimed +31.2pp win-rate improvement comes from n=19 backtest samples — statistically fragile (95% CI roughly ±22pp), meaning the true effect could be anywhere from +9pp to +53pp. PnL delta in this window: zero, because no qualifying signal fired. The filter only matters if it would have blocked or added trades; here it neither adds nor removes any. OVER-TIME PROJECTION: Over 3–6 months, if the effect is real, you'd see fewer USDCHF trades but a higher hit rate — likely net positive expectancy if the filter is selective (6 conf is high). However, the risk is overfitting to a small sample: 19 trades is not enough to distinguish a Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com