MLB Moneylines as Investment Assets

Abstract

In this paper, I apply prediction market theory to Major League Baseball (MLB) moneyline pricing.
Applying various machine learning models to a comprehensive data set of past game and player data, I calibrate probability estimates of teams’ chances to win games. With these probability estimates, I backtest profitable investment strategies using modified versions of the Kelly criterion staking strategy. Finally, I implement a profitable real-world betting strategy using the techniques developed herein over the first three months of the 2021 MLB season.