The New York Racing Association, Inc. (NYRA) and the New York Thoroughbred Horsemen's Association (NYTHA) announced that a team led by Brendan Kumagai has been selected as the winner of the inaugural Big Data Derby competition, beating out 105 other submissions for first prize. Kumagai, together with Gurashish Bagga, Kimberly Kroetch, Tyrel Stokes, and Liam Welsh, took the $20,000 first prize with the submission, “Bayesian Velocity Models for Horse Race Simulation.”
Kumagai's team created a dynamic model that focused on horses' forward versus lateral speed and examined the results of sustained momentum and velocity within races. The team also studied jockey performance and how that would impact a horse's running style in addition to biometric data, leading to the possibility of calculating a horse's welfare and injury probability.
“We're honored to be named the winners of the inaugural Big Data Derby competition,” Kumagai said. “Our team primarily works in financial analytics and hockey statistics, so working with horse racing data has been a unique challenge which allowed us to apply our skills to a relatively new and unexplored domain.”
The Big Data Derby was launched with the goal of analyzing the vast amounts of data available to racing organizations, and to understand how the results of those studies could impact traditional methods of racing and training. The competition was sponsored NYRA and the NYTHA in partnership with the Kentucky Thoroughbred Association, Equibase, The Jockey Club, Breeders' Cup and the Thoroughbred Owners and Breeders Association (TOBA).
Other models in the competition shed light on injury prevention, jockey decision making metrics, race tactics, track bias and more. Kumagai, a Data Science intern at Zelus Analytics, was previously part of a team that won the 2022 Big Data Bowl offered by the National Football League. The competition was judged by data analyst Rob Bingel, Rhodes College Economics Professor, horseplayer and thoroughbred owner Marshall Gramm, and Craig Milkowski of TimeformUS.
An open notebook of user-created content and data can be viewed at: https://www.kaggle.com/competitions/big-data-derby-2022/code.
Not a subscriber? Click here to sign up for the daily PDF or alerts.
