LAB #4897 NEUTRAL: LEDGER BENCH: Apply filter: ETHUSD SHORT — improves win rate from 31.9% to 50.0% (+18.1pp, n=74) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: ETHUSD SHORT — improves win rate from 31.9% to 50.0% (+18.1pp, n=74) SAME-WINDOW EFFECT: The filter only touches ETHUSD SHORT signals, and the current live window contains zero ETHUSD trades — every position listed is FX or metals (EURUSD, XAUUSD, AUDUSD, NZDUSD, GBPUSD, USDCHF, EURJPY, USDCAD). So over this specific window the filter would have changed nothing: same trades, same three closed losses, same two closed wins, same rejected/open set. The claimed +18.1pp is therefore entirely unverified against the live book and rests solely on the backtest sample of n=74. OVER-TIME PROJECTION: If the 31.9%→50.0% shift is real and not sample noise, it converts a losing ETHUSD-short sleeve into a roughly break-even-to-marginal one, which over 3-6 months is a modest but genuine drag reduction rather than an edge generator. The risk is that n=74 with a +18.1pp jump is exactly the profile of a curve-fit thres Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com