LAB #3892 NEUTRAL: LEDGER BENCH: Apply filter: Pool density HIGH (>=90) — improves win rate from 32.5% to 34.0% (+1.5pp — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: Pool density HIGH (>=90) — improves win rate from 32.5% to 34.0% (+1.5pp, n=1807) SAME-WINDOW EFFECT: Applying the pool-density ≥90 filter would have excluded the two `wkjudas` entries (NZDUSD LONG, USDJPY SHORT) and likely the `rizzy5m` AUDUSD SHORT (if pool density was below threshold at 11:35). Of the remaining 7 trades, you have 2 wins (GBPUSD SHORT, EURUSD LONG) and 5 losses/opens — a raw win rate of ~28.6%, worse than the unfiltered 30% (3 wins/10). However, the filter would have removed the two `wkjudas` losses (NZDUSD, USDJPY) and the open AUDUSD, so net PnL impact is positive if those three were net-negative — but the small sample (n=7) makes the +1.5pp backtest edge statistically invisible here. OVER-TIME PROJECTION: Over 3–6 months, a +1.5pp win-rate gain on ~1,800 trades is real but marginal — roughly +27 extra wins per 1,800 trades. The risk is overfitting to a single density threshold Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com