Being willing to back a founder primarily based on who they are is absolutely a viable form of risk-taking.

It may be a viable form of risk-taking but then it isn't contrarian risk-taking. However, has anyone ever actually ran the numbers on serial founders, the entirety of the cohort and not just the success stories?

I believe first round did a study that showed that companies founded by people who'd previously had a successful exit had better outcomes, but they also raised at higher valuations, and those effects canceled each other. So returns from investing in successful serial founders vs new founders are similar

Just as the efficient market hypothesis would predict.

Sure. But in my experience, the amount in which investors posture that they're contrarian and look for unconventional opportunities versus how much they actually do is quite notably different.