Predicting the NFL Draft Using Seven Numbers

by Bill Lotter

I know you’ve probably seen 17 mock drafts, so I’m gonna keep this short and sweet.  Back in February, I built models to predict where a player will be drafted and how well they will do in the NFL using combine data alone.  Surprisingly, both models did fairly well.  Time to put them to the test…and watch them get crushed by people named Todd and/or Mel.  After all, we’d expect experts, who have binged watch film, scouted players in person, and have inside knowledge of NFL teams to predict the draft much better than a handful of numbers.  But, let’s give it a shot.  Something I should note is that some players won’t be on here because they didn’t participate in the combine or didn’t participate in enough of the drills to make a meaningful prediction.  All the combine data was taken from http://nflcombineresults.com/.

Let’s start with CB’s, which, according to the model, is the easiest position to predict.  For every position, I’ll show the predicted draft order, as well as the predicted performance ranking.  More precisely, I’m trying to predict the total Approximate Value of a player within his first three years, a proxy for early career success.  Notice, this allows us to quantitatively say who had the best combine.  For the most part, the draft and performance rankings will line up because NFL teams, either implicitly or explicitly, do a good job at weighting the different combine events.  However, teams do tend to overweight the bench press (squats are where it’s at) and underweight a player’s weight.  Without further ado, here are the CB rankings…

bill.1

No surprise with Byron Jones on top, given his freakish performance.  But it’s interesting that Trae Waynes is so low.  Even though he had a great 40, he did relatively poorly in everything else.

In general, the combine has more predictive power for defensive positions.  Let’s look at OLB’s, DE’s, and DT’s next.

bill.2

bill.3

Familiar names with Vic Beasley and Alvin Dupree on top for OLB and DE respectively, but a lower projected prospect, Derrick Lott, takes the highest spot for DT’s with his very impressive 40, shuttle, and three cone for a guy his size.

Let’s round out the defense with positions for which the combine is less important:  ILB, FS, and SS.

bill.4

Notice my man Ibraheim Campbell out of Northwestern, aka Football U, with the top spot for SS (Go Cats!…sorry for the plug).

Alright, time to move onto offense.  We’ll start with some skill positions:  RB, WR, and TE.  The combine has some predictive power for RB’s and TE’s, but surprisingly doesn’t tell us much for WR’s.

bill.5 bill.8

Given his average numbers at the combine, Maxx Williams doesn’t even break the top-five lists for TE’s.  This goes to show you that there are some aspects of athleticism that can’t be measured by the combine drills alone.  I wouldn’t put particularly much stock in the predictions for these three positions.

Next up:  Quarterbacks.  This just in…Jameis Winston is going to drop to the seventh round!  Yeah not so much.  For Mariota and Hundley, the combine numbers are more meaningful and reflective of their style of play.

bill.6

Last but not least, let’s look at the big guys.

bill.7

One of the surprises from my analysis a couple months ago was that the combine actually has as much, if not more, predictive power for offensive lineman as skill positions, especially for centers.

All-in-all, the combine predictions give us what we might expect:  you can pick out a few players who look like they are going to be stars, but after that, it’s a crapshoot.  So as we go on throughout this week and into the next few years, I expect some of the top players to match the predictions here, but whether or not a guy is the 8th best OT or the 5th best is hard to tell from just the combine.  Neither numbers nor pure football intuition alone will give you the best predictive power. It’s the synergy of both that will lead to successful teams in the future.

About the author

harvardsports

View all posts

5 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *