What NFL Combine Numbers Matter For Fantasy Football? Here's What the Research Says.

Combine numbers only tell us so much. Forty times are more important for WRs than RBs. College production might be most important for RBs.

The defensive linemen, linebackers, and defensive backs already did their drills at the NFL Scouting Combine, and we will see the quarterbacks, wide receivers, and tight ends today and the running backs and offensive linemen on Sunday. Can we glean any information to help us win our fantasy leagues from the Combine numbers?

Let’s jump into the research.

I will look at the 2019 paper “Identifying Under Identifying Undervalued Players in Fantasy Fantasy Football” by Christopher D. Morgan, Caroll Rodriguez, Korey MacVittie, Robert Slater, and Daniel W. Engels of Southern Methodist University.

The SMU professors analyzed data from 2012-2018, including NFL Scouting Combine results, and compared it to RB and WR fantasy performance. They used lasso regression, random forest regression, and XGBoost to build their model, training the model with 2013-2016 data and testing the model with 2017 and 2018 data. Only lasso and XGBoost models produced reliable results. The XGRegression had the most compelling results for the greatest number of variables.

The results showed that for running backs,

20-yard shuttle, vertical jump, and bench press are the three most important Combine measurements—in that order.

Running Back Scores

The lasso regression the SMU professors completed included very few measurements from the Combine, and lasso models show results in terms of coefficients, while the XGBoost regression showed results in terms of F-score. I labeled them in terms of well they fell in rank of feature importance. Bench press was the most important of the strength and athletic measurements in both models.

But college yards per carry was more important than any of those measures. Maybe the best NFL RB would be the one with a high college yards per carry and a good bench, vert, and 20-yard shuttle? I may look into doing some analysis of that hypothesis this coming week.