MODEL ADOPTION #2669: REJECT — LEDGER BENCH: Apply filter: XAUUSD SHORT — improves win rate from 31.1% to 48.5% (+17.4pp, — E8 Intelligence Research

DECISION: REJECT ELEMENT: No rule survives — the XAUUSD SHORT filter is direction-specific, not E8-geometry-compatible, and the n=33 backtest is too small to separate signal from regime noise. EXPECTED IMPACT: If adopted, win rate might rise ~5–8pp in the current gold range, but this is likely overfitting to 2026 summer behavior — not a structural edge. The 17.4pp improvement is within noise for n=33 (95% CI roughly ±17pp). RISK: Structural USD weakness or a gold downtrend would make the filter catastrophic — you'd miss all profitable shorts. Also, the filter ignores E8 timing (Gods-Clock rhythm), so it could block a high-probability short that aligns with the fade polarity edge, degrading the core model's coherence. TEST: Before any reconsideration, run a walk-forward on the last 12 months of XAUUSD shorts in 3-month windows. If the 31.1% win rate is stable across all four windows (not clustered in one), AND a shadow test over 60 days shows ≥10pp improvement with ≥20 skipped s Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com