MODEL ADOPTION #305: REJECT — LEDGER BENCH: Apply filter: XAUUSD LONG @ ASIAN with 5 conf — improves win rate from 32.3% — E8 Intelligence Research

DECISION: REJECT ELEMENT: The filter "XAUUSD LONG @ ASIAN with 5 conf" does not survive — it is a single-asset, session-specific rule with no demonstrated compatibility with E8's multi-asset, rhythm-based geometry. The sample (n=21) is too small to rule out overfitting, and the live window shows zero overlap with current trades, meaning the edge is untestable against the active portfolio. EXPECTED IMPACT: Negligible to negative — even if the 66.7% win rate held, the filter would apply to <1% of live trades (XAUUSD only, Asian hours only), yielding no measurable improvement in overall win rate or profit. The opportunity cost of adding complexity without broad applicability degrades model efficiency. RISK: Overfitting to a small historical sample; regime shift in gold's Asian-session behavior; reduced trade frequency; potential false positives during low-liquidity Asian hours that could cluster drawdowns. TEST: Before any reconsideration, run a shadow-test on a separate paper acc Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com