"does that mean black people are more suited to be piano players"
Maybe I'm misunderstanding things, but if some statistical factor was identified with predictive power, does that not literally mean that generally they'd be more suited to be piano players?
This isn't some argument about philosophically what constitutes a piano player. If this were true, you'd expect on average that this group has a higher proportion of piano players, and you could say heuristically that, yes, they are more suited to be piano players.
Caveat: this does not necessarily mean you should base decisions solely off of whether they belong to the group, as it depends on how strong said predictor actually is.