Russ Goodman has served Central College since 2002, teaching at all levels of the curriculum. He especially enjoys learning about sports analytics, origami, algebraic geometry,
and mathematics in pop culture. Russ is also an assistant coach for the Central College women’s soccer team. In his free time, he enjoys spending time with his beautiful wife and two daughters, running, watching soccer, and enjoying all kinds of music.
Perfect Football Recruit
Who is the “perfect” football recruit for our small, liberal arts college? My student researcher and I embarked on finding an answer to this question in the fall of 2019, pre-COVID-19, and persevered through the pandemic to create a model that quantifies the likelihood of a football recruit matriculating at our institution.
Through conversations with the head football coach, we refined the project goals and focused on acquiring the data we would need. In our project, as in any data project, acquiring and cleaning our data set with over 1,300 recruits over the past seven years was painstaking, but critical work. After those initial steps, my student created a map with an analysis of the football staff’s recruiting success in our state, where a vast majority of our students come from.
We ran a regression analysis to predict the likelihood of a recruit committing to our school based on factors such as: number of campus visits, home town, alumni parent/sibling, and number of wins by our program in the prior year. Our commitment prediction model correlated moderately well (r 2 = 0.45) with our test data and intriguingly told us that total number of campus visits and having an alumni sibling are the most significant factors in a recruit’s commitment.
The final step in the project was to develop a user interface for the coaches. The interface allows coaches to interact with a map that provides geographic information on past recruiting efforts. Additionally, coaches can input the factors for a current recruit and obtain a probability of success in convincing the recruit to commit. Planned future work on this exciting data project includes using more test data to develop greater modelaccuracy, improving the model via the use of more variables, and continuing to strive to fully characterize the “perfect” football recruit.