Is the SEC faster?

So we've established that the SEC and ACC are producing more NFL rushing yards than the other conferences. There's also been a long established rumor of SEC speed. I thought I'd dig into this a little. One way to model the quality of athletes in a region is to look at total population, as defined in the US Census. There are 1700 NFL football players and roughly 300M Americans. In theory (although soon to be disproven), there should be one NFL player for every 175,000 Americans. To model this, I've constructed a very crude model of conference recruiting regions (notice the BE in pink):

Some assumptions:

1) Programs receive the vast majority of their players from the regions on the map

2)Florida is 50/50 SEC/ACC. I realize that the ACC has more schools but UF seems to be getting the lions share of the recruits

3) States where one program dominates (Ohio, Penn) are listed in a single conference's region even though another conference (Big East) will try to poach players. I'm assuming the best players (future NFLers) are generally going to the top programs in the state. In Ohio, I think this is a very reasonable assumption (who was the last player to diss OSU for Cincinnati?)

You could spend all day poking holes in this model. Essentially it assumes that all the players on these teams come from the listed Geos. While this is largely true for small time programs (read Michigan State) and geographically gifted major programs (Florida and Texas), it certainly less true for the national, regionally challenged programs such as UM. Still, its a starting point. If we look at census data for the given regions, we find a surprising result:

US Census Data: Total Population

The Big 10 has the largest recruiting region! This means that, by statistics, the Big 10 should recruit and train the most NFL players. However, we've already shows that this isn't really the case, especially in regards to impact NFL RBs. So what's missing? First, college football does not recruit across all age and gender demos. It focuses on Males, 17-23. So lets look at the population breakdown among this group:

US Census Data: Males 17-23

Even with this cut, the Big 10 should be producing more talent than the rest of the country. (I guess the theory about the aging upper Midwest is somewhat apocryphal). So why isn't the Big 10 producing 23.5% of NFL running backs? This is where you get into some controversial territory. It seems NFL ability isn't distributed equally to all races. There's a great post on this topic in the Minneapolis city pages (http://blogs.citypages.com/blotter/2008/12/nfl_a_study_in.php). Essentially NFL starting line ups very quite a bit by race and position:

Quarterback (32 starters)

26- White (81 percent)
6- Black (19 percent)

Wide Receiver (64 starters)

59- Black (92 percent)
3- White (5 percent)
1- Other (2 percent)

Running Back (32 starters)

32- Black (100 percent)

Fullback (32 starters)

21- Black (66 percent)
10- White (31 percent)
1- Other (3 percent)

Tight End (32 starters)

18- White (56 percent)
13- Black (41 percent)
1- Other (3 percent)

Offensive Line (160 starters)

79- White (49 percent)
72- Black (45 percent)
8- Other (5 percent)

Defensive Line (119 starters)

94- Black (79 percent)
21- White (18 percent)
4- Other (3 percent)

Defensive Back (128 starters)

123- Black (96 percent)
4- White (3 percent)
1- Other (1 percent)

Linebacker (105 starters)

78- Black (74 percent)
23- White (22 percent)
4- Other (4 percent)

Placekicker (32 starters)

32- White (100 percent)

Punter (32 starters)

31- White (97 percent)

1- Black (3 percent)

At the running back position, 100% of week 15 starters in the 2008 football season were African American (sorry McGuffie). If you believe that African Americans are more likely to have the physical and mental disposition to play running back in the NFL, you should examine young African American demographics to understand the conferences relative strengths at this position (I'm making no inferences to starting left tackles here -- Jake Long). So if we breakdown the census data by African American males ages 17-23 we see that:

US Census Data: African American Males 17-23

Wow. Suddenly the numbers look ugly for the Big 10. ~26% of this group lives in SEC country. In fact, Georgia has more young African American men than California. So I think this begins to explain the SEC speed myth. The SEC has more men likely to make NFL running backs than the Big 10. Now is this analysis perfect: not even close. According to this, the Pac 10 should be the worst, and we know that's not true (although take away recruiting from LA and Oakland and you might see a pretty bleak picture). There's really a lot more nuance to this discussion that can be answered in this blog. The ACC, for instance, performs better than the numbers above would indicate (some tweaking of the FL model might even this up a little though).

In reality, recruiting is not so geographically confined as this analysis permits. Still, it does explain why RR and crew are spending a lot of time looking for "speed" players in the Southeast (I'm assuming that speed is euphemistically used instead of African American). Demographically, there's simply a higher concentration of the population there. Eventually I'll get around to linemen recruiting and hopefully we can explain why the Big 10 produces 2x as many NFL centers as the SEC.

On the 20 hour a week nonsense

I'm going to spend 10 minutes writing a response to the Freep article, b/c that's all it really deserves. By this time next week, we'll be far more consumed by the performance, or lack thereof, of our QBs in the Western game than in the practice time leading up to the game.

I spent a year doing NCAA D-1 athletics. It was on a rowing team that cost my university big bucks to run and generated 0 revenue. With no financial incentives to be good, we still spent far more than 20 hours a week in season and 8 hours a week out of season working on the team. Having said that, we broke no NCAA rules. The rules provide a kind of Bill Clintonesque state of interpretation (depends on what practice is).

Practice is essentially time spent doing physical activity with a coach watching. After practice, we'd be handed workout regimens and tapes to review. These would double the actual time spent with coaches. Some players also had mandatory study time assigned as well. I can well imagine a 10am-10pm schedule that involves only 4 hours of "NCAA practice time"

Its something like this:

10-12: Trainer time (not practice)

12-1: Team Lunch (not practice)

1-2: Reviewing team tapes (not practice)

2-6: Beaten into a pulp by merciless coaches making 7 figures (practice)

6-7: Shower and see trainer again (not practice)

7-8: Dinner (not practice)

8-10: More tape review/study hall (not practice)

This is a 12 hour day, with only 4 hours counted as practice time. It is also copacetic with the Freep report. The UM staff will put together a schedule like this, deliver it to the NCAA and all will be fine. The only alternative is to ban all voluntary workouts at every school. Then players will sue the NCAA for limiting their ability to pursue professional careers. They'd have a strong anti-trust case for this. My guess is this whole thing gets swept under the rug after a perfunctory investigation.

How bad is Tom Lemming?

This is kind of fun: break down the 2009 RB draft class by (Scout.com) star rankings out of high school. Some observations:

1) There were as many 2 stars as 5 stars drafted in the first round.

2) There were more 3 stars drafted than 4 stars.

3) In the later rounds (4-7), 2 stars dominated the draft boards.

This analysis is kind of painful to do, since it requires wading through the mounds of historical scout.com rankings, but it proves a point. Rankings are a pretty poor predictor of college performance. Would anyone trade Kevin Grady (5 star) for Donald Brown (scout 2 star) at this point?

The RBs in the NFL

While the ESPN data is great for getting a directional understanding of BCS schools at producing NFL quality talent, it doesn't really indicate the quality of the NFL players. To understand this, we'll break down 2008 NFL stats by conference. NFL.com lists player stats for the last season (http://www.nfl.com/stats/categorystats?tabSeq=1&statisticPositionCategory=RUNNING_BACK&season=2008&seasonType=REG). This analysis uses the top 150 yardage earners in this year during the regular season. If we cross reference this with the ESPN.com data to get college affiliation (filling in the missing college stats for Deuce and others), we can correlate conferences with total production:

A couple of interesting findings:

1) When compared to the number of bodies, the BCS schools are producing more yardage and TDs than non-BCS schools. This analysis suggests that these schools have more impactful running backs than the smaller schools. Still, more than 30% of the yards in the NFL are coming from athletes that were (presumably) ignored by the BCS. I

2) The supposed SEC speed advantage shows up here. The Big 10 and SEC have a similar number of backs in the league (13.7 for Big 10, 13.2 for SEC), yet the Big 10 produces 55% fewer yards. Ouch. The TD situation is better, suggesting that Big 10 backs are finding niches as short yardage bruisers (see TJ Duckett).

3) The comparison becomes really interesting on a school by school breakdown:

A few more observations:

1) The Big 10 will miss Glenn Mason. Without his stable of backs, the Big 10 is a RB black hole. Even so, 5 SEC schools have better numbers than the top Big 10 school.

2) Wisconsin's reputation for producing RBs is total bunk. They don't show up at all. Ditto Steve Spurrier (but he doesn't really have a run 1st offense)

3) Michigan RBs don't produce in the NFL. Not too surprising with Chris Perry and Mike Hart being out last 2 NFL running backs.

We'll keep on digging through this. Next up: the WRs.

Do BCS schools produce more NFL Talent?

Its been noted that a number of the impact players in the NFL came from schools that don't typically recruit top flight high school talent. Year after year, players from no-name schools (ie non-BCS) are taken in the first rounds of the NFL draft. Just looking at the 2009 draft we see the following names off the board in the first 2 rounds:

1 (16): Larry English, Northern Illinois

2 (47): Mike Mitchell, Ohio

2 (52): David Veikune, Hawaii

2 (57): Paul Kruger, Utah

2 (59): Martin Sherrod, Troy

2 (61): Sean Smith, Utah

That's 6 guys that the NFL scouts thought pretty highly of who were (presumably) missed by the 65 recruiting staffs at BCS schools. The question then becomes, what is the impact of the BCS players in the NFL vs. the "rest" of the college football world.

To do this analysis, lets start with some basic facts/assumptions:

1) There are 119 D-1A schools (I refused to learn what FBS and FCS are)

2) There are 65 BCS Schools.

3) ~55% of players play for BCS schools. This assumes that every school has filled up all of its scholarship spots. We'll ignore D-II and D-III for this to make the math more reaslistic.

4) We'll ignore the "Randy Moss" effect. There are some players too toxic for BCS programs that will have an NFL future. The BCS schools didn't miss these players, they just chose to ignore them to save the embarrassment of having them on the roster. My intuition tells me that this effect is becoming less and less prevalent as the pressure on coaches mounts (Google "The Fullmer Cup" for a fuller discussing on this topic)

So, based on this, if one were to randomly assign the "best" high school players to Div-1A schools, 55% would be drafted out of the BCS, 45% of the dregs of Div-1 (The MAC, MWC, Conf USA, Notre Dame, etc.)

To figure out if the BCS, with all of its advantages ($$$, TV exposure, etc.) is finding and developing players , lets grab a data source that lists nfl rosters by college. ESPN offers such a source (http://espn.go.com/nfl/players?search=). A couple of comments on the ESPN data:

1) its horribly inaccurate. A lot of players are missing from the data when listed by college, including Deuce McAllister, Le'Ron McClain, Michael Pittman, etc.

2) It includes far more players than are active on NFL rosters (~2200 players vs. 1700 active on 53 man rosters)

3) There are some weird elements to the data. Only 2 OG are listed and there are 5 dBs listed (I'd guess CB/S tweeneers). Kudos to Steve Hutchinson for being only one of 2 OG that ESPN considers worthy of listing.

4) we'll fix the errors in some subsequent analysis by cross referencing with NFL.com data

The first analysis is to simply count the players, by position and conference to see of the BCS schools are better at recruiting NFL talent than the rest of the country. This analysis is below:

Instantly a couple of observations can be made:

1) Overall, the BCS is a little better than the rest of the world. Instead of the 55% of NFL rosters one might expect from a random distribution, the BCS accounts for 62.5% of NFL rosters. This is actually pretty low considering their advantages in recruiting. In a subsequent post, I'll break down star rankings and (hopefully) show that the BCS is getting far more than 62.5% of the top flight high school recruits (at least as the folks at scouts and rivals rank them). You'd expect that there would be a huge disparity in talent. In fact, this analysis suggests that the BCS schools are just slightly better than average in judging (and developing) NFL talent.

2) Some positions clearly project better than others from High School to the NFL. Assuming the BCS is getting their choice of the "best" high school players, one can correlate high school ranking to NFL ability. The position with the best correlation appears to be DT and LB, where more than 70% of players are from BCS rosters. I'd postulate that these positions are mostly influenced by attributes that can be measured in high school (Size and strength).

3) Offensive skill positions are essentially the worst for the BCS schools. WR, QBs, and RBs are nearly statistically indeterminate. For all the energy devoted to landing top flight talent here, most BCS schools (the USCs and Florida notwithstanding) would be better off entering a player lottery than in signing the players they do (same talent at a fraction of the recruiting $). An NFL quality wide receiver is almost equally as likely to have come from a BCS school as not. QBs and RBs are not much better. Essentially the qualities that project an offensive skill player to the NFL are largely hard to distinguish at the High school level.

4) Punters are statistically more likely to come from non BCS schools. This is hard to justify. The data is a little skewed due to the small sample size, so it might be a mistake to read too much into it. There could be a "Zoltan Mesko effect" here. Perhaps the punters from weak schools get more game experience and are more likely to succeed in the NFL, but this may just be creating insights from statistical noise.

In the next few posts I'll dig into this data a little more, by looking at NFL stats to see if BCS players account for more production than non BCS players. Maybe they are a disproportionate amount of NFL yards and tackles?