Statistical relationship between NFL combine success and success during rookie contract

This week marks the first major step in the NFL offseason: the scouting combine. Over the next few days, many viral videos will emerge as players show how fast they can run, how much they can lift, and how shifty they can move between cones. Indeed, it is fun to watch super athletes do super athletic things, especially when the super athlete is a very large human.

However, these drills are not evidence of one’s ability to play football… which… uh… is what teams actually care about. While true, there is a rather meaningful correlation between one’s overall athleticism [measured used Kent Lee Platte’s (@MathBomb on Twitter) relative athletic score calculation often using combine scores] and the approximate value (a measure of overall play quality) a player generates during their rookie contract.

For those of us that are less statistically inclined, Pearson’s R is a correlation estimating the degree to which a score on ‘Variable Y’ trends up or down depending on one’s score on ‘Variable X.’ In this case, we see positive relationships between RAS and approximate value for all defense positions, implying better players tend to be more athletic (as we might expect).

Additionally, in examining these graphs, we can see that the ‘R’ scores vary considerably between positions, ranging from .02 for defensive tackles (DT) to .19 for strong safeties (SS). This implies players at some positions may be especially benefitted by being highly athletic, which greater R values reflecting positions that are especially benefitted.

If we wanted to get more technical, R^2 is the squared calculation of R and is thought to summarize the “variance explained in Variably Y by Variable X.” This essentially means a score of .19 on R (the score for SS players) allows us to predict 4% of the approximate value a player will accrue during their rookie deal based only on their combine performance; a small amount for sure, but likely a meaningful one given the stakes. By comparison, a squared score on the .02 R for DTs would be .004 variance explained, which essentially tells us nothing.

For a team like the New York Jets who are linked to both DT and SS players, the aggregate of this data would inform us that more might be learned about the SSs than the DTs over the next few days. This may be something to keep in mind as players ‘rise’ and ‘fall’ based on their performances.