We’ve already gotten early word that Mississippi State wide receiver Da’Runnya Wilson improved his shockingly slow 4.85 40-yard dash time from the combine, timing between 4.71 and 4.78 at his pro day. Auburn running back Peyton Barber ran the forty in a time of 4.59 seconds at his pro day, besting his 4.64 time at the combine.
While Bill Lotter of the Harvard Sports Analysis Collective found that the combine doesn’t matter much for wide receivers, the forty is the drill most predictive of running back prospect NFL success.
I’ve done work predicting running prospect success about as well as NFL draft position by simply plugging into a logistic regression equation a running back’s forty time, weight, rushing yards per game, and receiving yards per game. But which forty time should we use?
The combine provides the only uniform, controlled environment with electronic timing where prospects are on a level playing field. But there is also a credible theory that a prospect who re-runs the forty at his pro day probably knows that the combine doesn’t represent his best possible time, either due to an off day, injury, or other factor.
I decided to try and answer the combine-versus-pro-day debate by looking at running backs who ran the forty at both events from 2000-2013, and see which time gave the most predictive results for that cohort. The answer is extremely important for running back prediction because the difference in forty times was not insignificant, with pro day times on average over 0.08 seconds faster than the prospects’ combine times.
First, let’s take a look at the 74 running backs that ran at both events from 2000-2013.
The cohort that ran at both events weren’t as highly touted, as prospects, as the entire group of running backs who ran at the combine. But there isn’t a dramatic difference in mean and median draft positions.
|Mean Draft Pos||Median Draft Pos|
|All Running Backs||146.2||137.5|
|Combine & Pro Day||173.3||175.0|
The list of biggest improvers in the 2000-2013 cohort in pro day forty times includes hits (LeGarrette Blount and Dion Lewis) and misses (Montee Ball).
|Name||Weight||Combine 40||Pro Day 40||Diff|
What did we find when we compared the statistical significance of combine and pro day forty times for the entire group?
Without getting too technical, I ran logistic regressions a few different ways to see the difference of statistical significance, or p-values, for combine and pro day forties for the entire cohort. Lower p-values are better, meaning the independent variable (in this case the forty time) is more likely to be predictive.
|Forty Only||40 + Wt + Prod||40 + Wt + Prod + Draft Pos|
|Pro Day 40||0.466||0.586||0.899|
You can see that the combine forty time was more statistically significant than the pro day time in predicting NFL success in each regression equation.
This was a smallish cohort for testing, so the results should be viewed in a context of some doubt. Plus, none of the p-values for the combine or pro day times came in below the 0.10 threshold for any of the equations.
This isn’t the definitive word, but for now it seems logical to view pro day times incredulously, and possibly fade prospects with vastly improved results.