Counterfactual Optimization of Baseball Pitch Sequences and Estimation of Its Impact on Season-Level Statistics
Researchers have developed a Transformer-based machine learning model to analyze Major League Baseball pitch sequences using MLB Statcast data. The model predicts pitch outcomes and generates counterfactual sequences to estimate the impact of optimizing setup and final pitches on season-level statistics. The study suggests that strategic pitch sequencing could lead to significant improvements, such as over a 1.0 increase in K/9, and offers practical insights into effective pitch locations and the role of middle-velocity pitches. AI
IMPACT Demonstrates AI's potential to uncover complex strategic advantages in sports analytics beyond traditional methods.