ABSTRACT:
Major League Baseball (MLB), a professional baseball league in the U.S. and Canada, is one of the most popular sports leagues in North America. In 2006, PITCHf/x, a pitch tracking system, was introduced that allows measurements to be recorded and associated with every pitch thrown in MLB games. The system, which was installed in every MLB stadiums, records useful information for every pitch thrown in a game such as the initial velocity, plate velocity, release point, spin angle, spin rate, etc. These information are then used to classify the pitch type (e.g., fastball, curveball, changeup, knuckleball, etc.) according to an algorithm developed by MLB Advanced Media. Given these pitch type classifications, we developed a model that would predict the next type of pitch thrown by a given pitcher, using only data that would be available before he even stepped to the mound. We used data from three recent MLB seasons (2013-2015) to compare individual pitcher predictions based on multi-class linear discriminant analysis, support vector machines, and classification trees to lead to the development of a real-time, live-game predictor. Using training data from the 2013, 2014, and part of the 2015 season, our best method achieved a mean out-of-sample predictive accuracy of 66.62%, and a real-time success rate of over 60%.
Speaker: prof. Hien Tran (North Carolina State University, USA)
Contact persons: Antonio Marigonda and Marco Caliari
Date and place: May 3rd at 10,30 in room M.
