LAB #2379 HARMFUL: LEDGER BENCH: Apply filter: BTCUSD LONG @ OFF_SESSION with 5 conf — improves win rate from 31.2% to — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: BTCUSD LONG @ OFF_SESSION with 5 conf — improves win rate from 31.2% to 100.0% (+68.8pp, n=10) SAME-WINDOW EFFECT: Over the current 10-trade window, the filter would have blocked all 10 trades — none are BTCUSD LONG @ OFF_SESSION. Therefore, the live PnL would be unchanged (zero delta), and the win-rate improvement of +68.8pp is an artifact of a tiny, single-instrument, single-session subset (n=10) that has zero overlap with the actual live trades shown. The claimed 100% win rate is statistically meaningless at this sample size and does not generalize to the broader portfolio. OVER-TIME PROJECTION: Over 3–6 months, this filter would likely reduce trade frequency dramatically (only BTCUSD longs in off-session hours with confidence ≥5). Given the GTX setup data shows 186 losses vs 111 wins in the last 30 days across all setups, a narrow filter could either avoid many bad trades or simply eliminate mos Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com