Strategic swinging model: building a model that optimizes swinging decisions in baseball

The most important battle in baseball is the battle between the pitcher and the batter, as hitting the baseball hard and far will drastically change the outcome of a game. In this research, we are attempting to build the strategic swinging model that can help Major League Baseball (MLB) hitters decide whether they should swing at a pitch before it has been thrown. We used random forest classifiers to output a probabilistic prediction of pitch type and pitch location, and estimated how well the hitter would like the pitch based on his past batting data. We evaluated the model by calculating how much better it performed compared to the scenario in which the batter did the opposite. The model outperformed any random swinging strategy. However, the decisions of most above average hitters are better than the model's decisions.