I actually wrote this post almost exactly two years ago, and have no idea why it just got cross-posted to LessWrong. I mainly like my post because it covers how the Kelly criterion sort-of applies to markets where you have to predict a bunch of things but don’t know what you’re actually going to learn. It’s also a more theoretical take on the subject. [EDIT: oh, also it goes through the proof of equivalence between a market of Kelly bettors and Bayesian updating, which is kind of nice and an interesting parallel to logical induction]
I actually wrote this post almost exactly two years ago, and have no idea why it just got cross-posted to LessWrong. I mainly like my post because it covers how the Kelly criterion sort-of applies to markets where you have to predict a bunch of things but don’t know what you’re actually going to learn. It’s also a more theoretical take on the subject. [EDIT: oh, also it goes through the proof of equivalence between a market of Kelly bettors and Bayesian updating, which is kind of nice and an interesting parallel to logical induction]