You're falling into the same trap. You think someone is better when in all probability they got luckier. And you think that picking someone you think is better will statistically get you a better person when that's not true if credentials are highly luck based. It's like buying lotto tickets from a gas station that's produced more winners in the past.
I'm saying your measurements aren't going to be any better than whatever percent of "skill" is randomness. CEO candidate 2 went to a better school but you don't know that it was his grandpa's alma matter who pulled some strings. Candidate 3 might be a multimillionaire but it's because he got a settlement from a malfunctioning ski lift when he was 10.
A lot of the "signals" companies use to determine if somebody would be a good leader or employee are 100% luck based. You could statistically pick a worse candidate every time that looks betters if you're looking at the wrong signals.