Videos by OutKick
One exercise I thought would be interesting is to study the colleges from a pure analytics perspective to see how “NFL ready” their student-athletes are when drafted. I measured this in two ways: how many games they started and how many games they played in, out of the maximum possible they could have played in during their NFL careers.
I looked at a 5 year period from 2010-2014 and grabbed every player drafted, the games he played in and the games he started, and began the analysis. Eventually, as the two tables below will show, I came up with statistics for each school which ranked the percentage of games started and the percentages of games played vs the NFL average, to see which schools were above or below average.
The next step was to run a regression analysis to look at games started per year (GS/Yr) vs predicted. To get the predicted figure, my regression was on all NFL draft picks from 2004-2010 (I set the cut off at 2010 because the NFL average career length is 5 years and I did not want to go beyond that number). I was intrigued to learn that the relationship between NFL draft pick and games started is even stronger than first expected. Certain positions are more correlated than others. For instance, the location of where a QB was drafted (pick 1-256) alone explains over 40% of the number of games they will start. And that’s factoring in 100% of the unknowns, such as injuries, starters, systems, coaches, performance, etc.
I filtered only for schools who saw 3+ players drafted by NFL teams over the last 5 years. This left us with 1,143 players from 107 different schools to analyze. Below are the results, first for games started, and second for games played. GOA is games played over average, and SOA is starts over average.
The next table shows the results of the regression analysis. I also included the ranking from the prior analysis on the far right, to see how the two were similar in many cases but different in others. Just a note that a reason I used games started rather than games played was the correlation was stronger between the games started and position drafted than the games played and position.
Discussion of Results
The reason Central Michigan moves into first place is because of the success of their two studs: Antonio Brown and Eric Fisher. Brown being a 6th round pick but starting in 43 total games since 2010 (a rate of 8.6 per year) is well above the expectation for a 6th round pick. On the other end of the spectrum, offensive lineman Eric Fisher was drafted #1 overall, and has started 29 of 32 possible games, which is still above average, even for the #1 overall pick. (Note that O-Linemen have a very strong correlation of draft position to games started. So if we averaged Fisher against just his roster position’s average, he’d fall much closer to average than he does when averaging against all draft positions.)
Kansas State remains one of the worst programs at providing the NFL with usable talent. In the last 5 years, they’ve had two players drafted in the 2nd round (RB Daniel Thomas and LB Arthur Brown) who have 3 starts combined. They had a 4th round WR (Chris Harper) drafted in 2013 with zero starts, a 5th round DB (Joshua Moore) drafted in 2010 with zero starts, and last year had a 6th round offensive lineman drafted (Tavon Rooks) who also has zero starts. Ironically, the one player with more starts combined than those other 5 is their 7th round RB (Bryce Brown) who earned 7 starts so far in the NFL. So while Brown is well above average, the rest of the players are well below average.
As you probably have notice with all of my draft analysis, there is very little precision with anything in the NFL. So much happens thru interaction and reaction rather than specific, measurable results which are 100% true. Some of these players don’t start games not because they are not good players, but because of a myriad of reasons: Injuries, drafted in a position to a team who already has a great starting veteran, etc. But the simple fact is, in the NFL, if you can play, you will start eventually. The “win now” culture is not going to let a good player sit on the bench for months if they are better than the player in front of them. A rookie QB is typically the lone exception.
One Final Analysis
I decided to look at conferences to see how they stacked up. Once again, I looked at both variables: the variance of GS/Yr as well as true % of games started. The results are below:
As previously stated, this is not an exact science. But it is consistent across all schools. Players are drafted to fill different voids on different teams with different expectations and different responsibilities. So rarely will there be a true apples to apples comparison across the entire NFL. We work with what we can to see if there is anything to uncover.
One thing we certainly can see is there is plenty of potential NFL starters to be found on various campuses around the country. Smaller conference schools are developing players more than capable of playing and making impacts in the NFL. Guys like Alfred Morris, Jason Kelce, Bruce Miller, Charles Clay, Khalil Mack, Muhammad Wilkerson, Mike Iupati, Antonio Brown and on and on rank in the top 100 for games started vs predicted (based on draft position) and come from C-USA, American and the MAC. Meanwhile, there are plenty of schools in the power conferences whose players don’t end up catching on in the NFL, such as Kansas State, West Virginia, Michigan, Wake Forest and Oregon State.
The reasons for this are surely better explained by scouts and people closer to those programs. This is just one way to look at it from a numbers perspective. All teams hope that a player they draft will start/play for them at some point, or they wouldn’t be drafting them. Sometimes they are forced to start players they drafted but don’t like or who aren’t ready, because of injuries or other factors. So simply looking at starts alone is far from a perfect way to measure success of players to the NFL. It’s simply one aspect.
But I think the bottom line in talent development is as follows: Much like we see popular power conference school with average records make bowl games only to be beaten by strong, small conference schools with excellent teams, we see the same thing with the players themselves transitioning to the NFL. While a simple analysis like this shouldn’t stop a team from drafting a player, it should be known by every team how successful those colleges are at developing players who did translate to the NFL. When certain schools have so many players who do nothing in the NFL, it should be looked at with a more critical eye by these teams.