(Well, this is a must-bump. Very interesting post. -- Ross)
The draft is done, and the new draftees are looking forward to fame, fortune... and a lockout. While the immediate NFL future is uncertain, college football moves on. Every year brings a new crop of recruits, many of whom hope to follow this weekends draftees into the NFL. Looking back can often bring useful lessons, and each draft class offers such an opportunity. How did the draftees get to that point? What elevated them above all the others hoping for a chance at the NFL?
Now maybe the answer is easy. We know there are three must-have ingredients: God-given talent, the commitment to get better, and the right training. But how important are each of these factors? In particular, how important is that third factor? What effect do different college programs have on their players chances at the NFL? We know that some programs produce more draft picks than others. Why, however, is a tricker question. Are those schools just bringing in more talent at the recruiting stage? Or are some doing a better job of developing their recruits into NFL-ready athletes?
Good recruiting data across college football goes back to 2002. And we have draft information for every year. So, by matching between those two data sets, we can answer the questions above. We can identify the programs that do the best (and worst) job developing their players (at least over the past decade). Better yet, we can use this data to tell a prospective recruit exactly how much their NFL chances are affected by their choice of school. Whether this information is on the top of the recruiting packets or hidden from sight will depend on the school, but these are numbers every recruit should know before signing on that dotted line.
So come with me on a journey through the player development process in college football. We’ll find out who’s doing it right, and who’s doing it wrong. We’ll see how different conferences lead to different outcomes, and the impact play-style has on player development. There might even be a Rich Rodriguez cameo.
One of the great debates every recruiting season is over the importance of recruiting rankings. One one side are the recruitniks talking up the top names, and on the other side the "stars don't matter" crowd. And there's evidence both ways: schools with better recruiting classes do outperform their more recruit-challenged brethren, but certain schools consistently punch above their recruiting weight on the football field. So let's start by getting one thing straight, stars do matter, even if they aren't everything.
|| Percent drafted
|| Average draft position
|★★||4.9%|| 143 (5th rd)
|★★★||8.1%|| 124 (late 4th)
|★★★★||16.7%|| 107 (early 4th)
|★★★★★||38.0%|| 81 (3rd rd)
The likelihood of being drafted increase substantially with a player's star-ranking. And when drafted, higher rated recruits are drafter earlier. For instance, a 4-star recruit is over three-times more likely than a two-star to get drafted, and will be picked over a round earlier than those two-stars that do get drafted.
Ok, stars matter, but they certainly don't tell the whole story. Even among those bluest of the blue-chippers, the 5-star recruits, the large majority will never be drafted! Motivation, injury and life all play a big role, but the differences between programs must matter. Player development - its something Iowa fans have heard about as Kirk Ferentz is known as one of the best. But is it true? Can we see it in the numbers?
Choosing OSU, USC or Iowa almost doubles a recruit's NFL chances
We've calculated the fraction of each rank of recruits that are developed into draft picks by the average BCS program. That allows us to evaluate the player development of individual programs (or conferences) compared to this average. The Development Ratio is a simple way to measure the effect of a program on player development: take the number of recruits a program turned into draft picks and divide that by the number that an average BCS program would have produced from the same recruiting classes. For instance, let's say some college program brought in 20 4-star recruits, and 80 3-star recruits, and that 15 of them were drafted. The average BCS program, by the numbers above, would have had 10 of those recruits drafted. So our example program has a development ratio of 15/10 = 150%, very good!
Back to the original question - we know USC produces more draft picks then Stanford, but is that just because of all those 5-stars they bring, or does USC have a better development program as well? If I am a recruit with NFL aspirations, which schools will best help me fulfill that dream? How much does it matter? The answers - for the best, the worst and a selected few in between:
|Ranking||School||Recruits Drafted||BCS Expectation||Development Ratio|
Programs have a big impact, and the best developmental programs clearly separate themselves from the rest of the pack. And its interesting to see who those best programs are. Ohio State and USC are huge names, attract great recruits, and turn out even better NFL prospects. These statistics come almost entirely from the Pete Carrol and Jim Tressel eras and they show that both coaches deserve the credit they get - they didn't just skate by on the higher talent level those programs attract on name, they got the best out of that talent.
Iowa is a very different type of program than Ohio State and USC, but the rumours are true - Iowa is an elite developmental program. Kirk Ferentz and the Iowa program are transforming a moderate level of incoming talent into superior NFL-ready players out on the field. Recruits nearly double their chances of being drafted by putting on the black and gold. And on an Iowa-related note, it is hard not to compare the success at Iowa to the lack thereof at Iowa State. The Cyclones managed to sign up just a shade less talent than did the Hawkeyes, but did much, much less with it. Player development is something that can and should get better in Ames, although Iowa State faces an additional obstacle in this regard - membership in the Big 12.
The Big Ten is the premier player development conference
The differences in player development between some teams is clear, but how about the conferences? This is especially interesting in light of the growing recruiting advantage that southern teams have enjoyed over northern teams in recent years. The Big Ten has been indicted as less talented based on those trends, but also sometimes suggested to do a superior job in developing the talent it does attract. What do the numbers say?
|Rank||Conference||Recruits Drafted||BCS Expectation||Development Ratio|
The Big Ten is indeed the elite player development conference. Recruits that go to the Big Ten add 15% to their NFL chances. The SEC and the ACC did have more draft picks than the Big Ten over the time period in question (the additional team helps), but that was because they started with with more talented recruits (and simply more recruits). The moment players step on campus, the Big Ten starts closing the talent gap through its superior player development. And as a side note, the SECs development profile suffers by its addiction to oversigning. While few doubt that oversigning is advantageous for the programs that do it, the players are the ones that pay the price in the form of a lowered chance of turning their talent into an NFL career.
The second big story in the conference rankings is how badly the Big 12 develops its players. Easily the low outlier, recruits to the Big 12 take almost a 20% hit to their NFL chances when they pick the conference! Is this just a matter of too many weak programs, or is there something deeper in those numbers?
Play-style matters: Big Ten defenders rejoice, Big 12 defenders despair
One aspect of the modern Big 12 that has attracted both praise and criticism is the wide-open pass-heavy style of football that has spread through the conference. While entertaining for fans, the ubiquitous spread offenses and deemphasized run games depart significantly from the NFL norm. Could this the culprit for the subpar development of players? One way to answer this question is to look at conference Development Ratios for players categorized by type, i.e. development of offensive vs. defensive players, skill players vs. linemen.
The breakdowns for the Big Ten and the Big 12 stand out. First, while the Big 12 does an acceptable, albeit subpar, job of developing offensive prospects, the league is brutal when it comes to developing defenders and linemen. This seems like strong evidence for the play-style hypothesis, since those are exactly the areas one would expect to suffer under a passing-emphasis style. The Big Ten's development profile also reflects its reputation, but in a good way. The Big Ten does not struggle in any area of player development, but it truly shines at bringing along defenders and the men in the trenches. Smash-mouth football is alive and well in the Big Ten, and tangibly benefiting its players. I'm sure that Big Ten and Big 12 programs go head to head for more than a few prospects each year, and defenders in particular should heed these numbers - choosing the Big 12 means you are accepting a big hit to any NFL aspirations.
Wins and Losses vs. Recruiting Rankings
Finally, while recruits might be most interested in how programs develop for the draft, college football fans care more about wins and losses. As a way of visualizing this, I calculated a Win Ratio that is essentially similar to the Development Ratio used above, except based on wins. First, I determine the BCS average number of wins expected from a programs recruits, and then compare that to the actual number of wins achieved. I plotted the Development Ratio vs. the Win Ratio for every BCS program below - calling out certain programs - and it makes for some interesting viewing
The best programs inhabit the upper right of this chart. These aren't the best recruiting schools, these are the best programs - they turn the recruits they get into more wins and more draft picks than would an average BCS school. I think few would be surprised to see Ohio State, Iowa and Wisconsin in this area. USC deserves some praise as well. Although they didn't outperform their win expectation, that is only because its almost impossible given the incredible level of talent they have brought in over the past decade.
The bad programs sit in the lower left: the Washingtons, the Dukes (so bad its literally off the chart), the Iowa States and Illinois. Some of these programs have talent, others less so, but all of them are failing to get the results that they should out of that talent.
Finally, the lower right corner is particularly interesting. These programs are outperforming their talent in terms of wins and losses, but that isn't translating into NFL careers for the players. West Virginia is a great example of how this happens: these numbers mostly reflect the Rich Rodriguez era of the spread'n'shred and 3-3-5 defense. These were great systems for winning college football games, but a far cry from the NFL norm. As a result, while Rodriguez turned his talent into wins, he did not turn it into NFL draft picks. Texas Tech is a similar case, the pass-pass-pass offense that made Mike Leach famous evened the playing field against more talented opponents, but did not prepare players for the NFL.
There's a lot of data out there these days, and if you treat it right, it can teach you a lot. And sometimes you even "learn" something you already knew - like how good a coach Kirk Ferentz is.
Methods: Rivals star rankings were used to rank recruits. Recruits from the 2002-2008 recruting classes were matched against draft results from 2004-present to determine draft status. 2009 and 2010 recruiting classes were not included as most of those recruits are still ineligible to be drafted. The expected number of draft picks for a program was calculated by multiplying the number of 2/3/4/5 star recruits for that program by the BCS average number of draftees per 2/3/4/5 star recruit. The Development Ratio is then simply the actual number of their recruits drafted, divided by this expected number. Note that only drafted recruits are included in this analysis, not walkons or certain other players who joined the team in an irregular manner. The Win Ratio was calculated by first regressing total wins from 2002-2010 against each programs number of 2/3/4/5 star recruits. This regression was then used to produce an expected number of wins for each program given its recruits, and the ratio is then the actual number divided by this expectation. Conference affiliations and BCS membership were assigned as per the 2010 season as this was the configuration that held over most of the time period studied. This means that Nebraska and Colorado are in the Big 12, and TCU and Utah are not in the BCS.
Update: More programs, notes on methodology
First, a clarification of how the number of drafted recruits for each program is determined. This is NOT simply the total number drafted from 2004-present. This is the number of recruits in the program's 2002-2008 classes that were drafted. In particular, most 2004 and 2005 draftees will not count towards the drafted number because those players belong to recruiting classes prior to 2002. Every team is treated the same, but this why the draft numbers people calculate by summing their teams draftees from 2004-2011 might differ from the ones I report.
Second, a lot of people have noted that there is an additional factor at play - talent identification. If some teams are better at identifying talent, they will outperform their expectations because their recruits are systematically more talented than their ranking indicates. This is very difficult to separate out from player development using these numbers. My opinion is that it is a lesser factor - for instance USC's recruits almost average 4-stars, those guys aren't flying under anyone's radar. Also, the effect of talent identification will likely be strongest for 2-star recruits that have received the least attention overall. But this is a good point, and something to look into in the future. From a program's perspective it's not so important - whether its talent identification or player development the results are good (or bad) just the same.