Pitch Prediction using Hidden Markov Model
The objective of this project is to predict the next pitch of individual pitchers in real time using a Hidden Markov Model
Hidden Markov Model Theory
To follow
Melling (2017) attempted to predict the next pitch using a Hidden Markov Model, but achieved only "pretty poor accuracy" and acknowledged that "adjustments could likely be made to improve performance." For his prediction, he used the Viterbi algorithm, but in this project we want to use the Forward algorithm, as it has been successfully used by Ötting (2021) to predict play calls in the NFL.
- Micah Melling, Predicting the Next Pitch: A Markov Chain Approach (2017) - https://www.baseballdatascience.com/predicting-the-next-pitch-a-markov-chain-approach/
- Marius Ötting, Predicting play calls in the National Football League using hidden Markov models (2021) - https://arxiv.org/abs/2003.10791