Predicting the future of free agent receivers and tight ends in the NFL
Keywords:National Football League, Prediction, Linear Regression, Recursive Partitioning Trees
While the NFL Draft is an important way to add new talent in the National Football League, free agency is the primary method for teams to acquire veteran players. Veteran players cost teams more money to sign, making it critical for teams to ensure that the free agents they sign are worth their higher salaries. We focus on predicting the future salary, performance, and value of free agents at the wide receiver and tight end positions.
These two positions have recently gathered attention as the NFL has transitioned to more passing-oriented offenses. We use player’s physical attributes, college performance, and NFL performance to date to create regression and tree models that predict the likelihood that a player is signed, how much they will cost, and how productive the player will be in the future. We find that there are differences between the predictors of salary and the predictors of future performance, which suggests teams are not efficiently evaluating free agents at these positions.