Pitching strategy evaluation via stratified analysis using propensity score

Abstract Recent measurement technologies enable us to analyze baseball at higher levels of complexity. There are, however, still many unclear points around pitching strategy. are two elements that make it difficult measure the effect a First, most public datasets do not include location data where catcher demands ball, which is essential information obtain battery’s intent. Second, there confounders associated with pitching/batting results when evaluating We here clarify attempts specific location, e.g., inside or outside. employ causal inference framework called stratified analysis using propensity score evaluate effects while removing confounding factors. use pitch-by-pitch dataset Japanese professional games held in 2014–2019, includes ball. The reveal an outside attempt more effective than one minimize allowed run average. In addition, shows regardless magnitude estimated batter’s ability, and proportion pitched for pitcher/batter. Our provides practical insights into selecting strategy runs.