Swing Time in Division I Baseball: Bridging Bat Speed and Swing Length to Predict Hitter Success

Abstract

In Major League Baseball (MLB), with the new release of bat-tracking data, bat speed and swing length have recently emerged as crucial indicators for hitting performance. While higher bat speeds relate to increased power, they often necessitate longer swing lengths. Longer swings result in less contact and more strikeouts; thus, this creates a trade-off between swings. In the MLB this trade-off broadly favors bat speed, but in college, where getting on base is far more valuable than raw power, that same emphasis is suboptimal. However, because collegiate stadiums lack the high-speed optical tracking required to measure these two crucial metrics, analysis remains limited.

This thesis introduces "Swing Time", which encompasses the ratio of bat speed and swing length, but operationally measured through frame-by-frame video analysis at the collegiate level. Ultimately, Swing Time is defined as the amount of time it takes for the hitter to complete the process of a swing. It is expected that when a hitter devotes less time to the process of his swing, he will have more time to process the trajectory of a pitch. This precious extra time can mean the difference between a successful and failed at-bat. Using frame-by-frame video analysis from Synergy Baseball synchronized with Trackman pitch data, Swing Times for a sample of Division I collegiate hitters are manually collected. Statistical testing confirms that Swing Time is a robust, independent metric, and is relatively unaffected by game context, pitch characteristics or hitter archetype. Furthermore, Principal Component Analysis (PCA) of MLB data confirms that Swing Time captures the hypothesized bat speed-length tradeoff finding a meaningful negative correlation with season-level hitting performance; essentially suggesting that shorter Swing Times are associated with better performance at the plate. Finally, a ridge regression model as well as a single-hidden-layer neural network were trained demonstrating that Swing Time is a powerful predictive feature for future season-level hitting statistics, offering a scalable solution for collegiate player evaluation.