Predicting March Madness Results with Quantile Regression by JP Wong '24

The $1 Billion Model: Predicting March Madness Results with Quantile Regression and a Logistic Regression-Markov Chain Process by JP Wong ’24, Wednesday March 13, 1:10 – 1:50pm, North Science Building 015, Wachenheim, Statistics Colloquium

Abstract: Millions of people submit their predictions for the results of the Men’s NCAA March Madness Basketball tournament each year. In this talk, we will discuss two empirical methods for predicting the winners of the tournament from the pool of 64 participating teams using regular season performance statistics. We start by exploring the quantile regression method that compares the spread of scored points between each matchup of the tournament. We then introduce the Logistic Regression-Markov Chain model that uses regular season game data alongside estimation of home court advantage to rank each team. Lastly, the two methods will be compared with each other and to other existing models by their performance in predicting previous years’ tournament winners.