LAB #4137 NEUTRAL: LEDGER BENCH: Apply filter: USDCHF LONG — improves win rate from 32.6% to 39.7% (+7.2pp, n=146) — E8 Intelligence Research
Year: 2026
Type: preprint
Abstract: IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: USDCHF LONG — improves win rate from 32.6% to 39.7% (+7.2pp, n=146) SAME-WINDOW EFFECT: Over the current live window, the filter would have blocked all USDCHF LONG entries — but none of the listed trades are USDCHF LONG, so zero trades would have been affected. The claimed +7.2pp win-rate improvement is based on n=146 backtest trades, which is a small sample and likely regime-specific. Expected PnL delta in this window: ~0, since no USDCHF LONG signals fired. The filter adds no immediate value here and cannot be validated against these 10 trades. OVER-TIME PROJECTION: Over 3–6 months, the filter could genuinely improve win rate if USDCHF LONG has a persistent structural edge (e.g., carry or momentum asymmetry). However, the risk is overfitting to a short backtest window — 146 trades is roughly 5–8 weeks of data for one pair/direction, and the +7.2pp could be noise or a single volatile month. If the Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com
Author: Caldin, Andrew Stewart
Language: English
Subfield: Finance
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7222486972
Cites:
Cited by:
Related to: