LAB #4599 HARMFUL: LEDGER BENCH: Apply filter: Session = NY_OPEN — improves win rate from 31.5% to 38.1% (+6.6pp, n=805 — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: Session = NY_OPEN — improves win rate from 31.5% to 38.1% (+6.6pp, n=805) SAME-WINDOW EFFECT: Applying the NY_OPEN filter would have excluded all 10 live trades listed, as every entry timestamp falls between 09:56 and 15:08 +02:00 — which corresponds to London/early NY overlap, not the defined NY_OPEN session (typically 14:30–21:00 +02:00). The filter would have removed 4 winners (USDCHF, USDCAD, GBPUSD, NZDUSD) and 2 losers (AUDUSD SHORT, GBPUSD LONG) — netting out to a loss of +2 wins and -2 losses, leaving only the 2 remaining losers (USDJPY, AUDUSD LONG) and 2 rejected/open trades. Expected PnL delta is negative: you would have forfeited roughly +4R of realized gains while avoiding only -2R of losses, a net -2R hit in this window. The backtest improvement (+6.6pp) appears to be a session-mislabeling artifact — your live entries are not actually in NY_OPEN. OVER-TIME PROJECTION: Over 3-6 months,

Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com