Believe in someone before the stats do. That’s when belief is actually worth something.

Backing a person after they have proved themselves is agreement, not belief. It costs nothing and changes nothing, since the evidence was already doing the work. The version that matters arrives early, when the numbers are unimpressive and the person is mostly wondering whether anyone sees anything in them at all.

Pick someone who is currently unremarkable at the thing they are trying and tell them specifically what you see. Then give them a chance slightly beyond their record. Most people who became good remember exactly who did that for them.