Ohio State nav bar

Student Research Winner: Brooke Williams on “Predicting Free Throws in the NBA”

Abstract

The goal of the research for this paper is to identify predictive factors of free throw attempts in NBA basketball games with the intention to develop a model predicting the number of Free Throws (FT) each team will take in an upcoming game assuming there is a significant correlation between Free Throw Attempts (FTA) in a game and winning that game. An exploratory examination of possible variables correlated with a team’s FTA provides two worthy variables: FTA for the home team and Personal Fouls (PF) of the opposing, visiting team. This linear regression equation for free throw attempts, FTA = intercept (11.7) + FTA Team* (0.501) + PF Opnt* (-0.014), is the resulting model after training on 80% of the 2018-2019 NBA season data. However, it is important to note that these two variables do have a strong correlation with each other, and this possibly explains the negative coefficient for the season average opponent’s personal foul of -0.014. Additionally, a naïve model was created to get a benchmark with which to compare the model results. In the end, the model’s predictive power is weak and tends to predict around the average value of the data it is trained on, as opposed to the wider range of actual values. The conclusion from this research is that possessions per minute, a team’s personal fouls and an opponent’s free throw attempts are not strong indicators of free throws in an upcoming game. Ultimately none of the variables considered and analyzed within this research show significant predictive power of free throw attempts for a team in an upcoming game.

Sports and Society Statement

One of the biggest issues plaguing the world of sports today is the idea of paying collegiate athletes. The research I conducted explores the concept of athletic COA stipends, the first step in paying collegiate athletes, and their potential negative impact on women's and men's non-revenue sports.‪ This project meets the criteria of researching a sport and society issue via sports analytics, with the eventual end-goal of contributing to the up and coming data-analysis-driven industry within legal sports betting. This research paper attempts to identify predictive factors of free throw attempts in NBA basketball games in order to develop a model that predicts the number of free throws each team will take in an upcoming game. The hypothesis is that there is a significant correlation between free throw attempts in a game and winning that game; if this correlation is significant, then this predictive FTA model could inform on an upcoming game’s outcome.