Zero-Knowledge Proofs of Stock-Picking Skill

Abstract

The conventional wisdom is that you must reveal something about how you pick stocks in order to prove that you have stock-picking skill. In this paper I show that, prior to executing any trades, it is possible to prove you have stock-picking skill without revealing any additional information about your underlying trading signal. Here is how the protocol works. The evaluator presents you with a sequence of paired return data sets, one real and the other suitably randomized. A profitable trading signal will only be able to predict the cross-section of returns in the real data set. So by repeatedly using your trading signal to identify the real data set, you can prove that you have stock-picking skill without revealing anything else about your underlying signal. This protocol represents a zero-knowledge proof of stock-picking skill—i.e., a proof which reveals nothing except for the validity of your claim. Zero-knowledge proofs allow any skilled stock picker to advertise his ability without fear of his trading signal getting scooped. As a result, they have important implications for how the active-management industry is organized.