Bayesian Velocity Models for Horse Race Simulation

Placement: 1st out of 108

The 2022 Big Data Derby, organized by the NYRA and NYTHA, provided data scientists with a previously untouched source of horse racing tracking data.

In Tyrel Stokes, Kimberly Kroetch, Gurashish Bagga, Liam Welsh, and I's submission for the Big Data Derby, we focused on developing a model to predict the probability that any given horse will finish in any possible place (i.e., 1st, 2nd, 3rd, etc.) dynamically at any point in the race. To do so, we developed Bayesian B-spline models to predict the forward and lateral velocity of each horse and used these models to simulate forward a race from any given situation.

Our Kaggle submission linked above describes our approach in further detail.

Brendan Kumagai
Brendan Kumagai
Data Scientist

Data scientist with a passion for hockey and science.