1st and a Mile: A Look at 1st year Defensive Coordinators

The 2010 college football regular season featured 23 new head coaches and staffs (defensive coordinators in parentheses): Florida State (Mark Stoops), UVa (Jim Reid) , Kansas (Carl Torbush), Texas Tech (James Willis), Cincinnati (Tim Banks / Jon Jancek), Louisville (Vance Bedford), USF (Mark Snyder), ECU (Brian Mitchell), Marshall (Chris Rippon), Memphis (Jay Hopson), Notre Dame (Bob Diaco), Akron (Curt Mallory), Buffalo (William Inge), Central Michigan (Joe Tumpkin), UNLV (Kraig Paulson), USC (Monte Kiffin), Kentucky (Steve Brown), Tennessee (Justin Wilcox), Vanderbilt (Bruce Fowler), UL-Monroe (Troy Reffert), Western Kentucky (vacant), La. Tech (Tommy Spangler) and San Jose State (Kent Baer).

In addition, there were 20 programs that unveiled new defensive coordinators: Georgia Tech (Al Groh), Georgia (Todd Grantham), Arizona (Bill Bedenbaugh / Greg Brown), Stanford (Vic Fangio), Cal (Clancy Pendergast), Illinois (Vic Koenning), Duke, (Jim Knowles), FAU (Kurt Van Valkenburgh), Nevada (Andy Buh), UTEP (Andre Patterson), FIU (Geoff Collins), Ball State (Jay Hood), UF (Teryl Austin), Miss. St. (Manny Diaz), Middle Tennessee St. (Randall McCray), Houston (Brian Stewart), Boise State (Pete Kwiatkowski), Eastern Michigan (Phil Snow), Western Michigan, and Texas A&M (Tim DeRuter). That's quite a bit of turnover. But such is the life in the modern day pay-for-play (no offense, Cam), admin-eat-coach world.

So how did the 1st year DCs do? We'll look at a statical comparison of the each BCS / mid-major team's defense, and then look at the change in defensive ranking from 2009 to 2010 to see who is working some magic and who might end up working the street corner.

Wait for it...and...the jump.

The table to the right is a composite of all 2010 BCS teams plus Boise State and Nevada that featured a new defensive coordinator for the 2010 season. The two columns are defensive F/+ rank and the defensive strength of schedule (FEI-based) to get a feel for the quality of offenses faced. The table has been sorted by Def F/+ ranking.

For reference, the F/+ metric is a combination of FEI and S&P ratings (give them a click). FEI is a Brian Fremau creation which adjusts for, in other things, garbage time and clock-kills. The S&P Ratings are based on 3 categories:

1. Success Rate - 50% of necessary yardage on 1st down, 70% on 2nd down, and 100% on 3rd and 4th downs.
2. Equivalent Points Per Play - an explosiveness measure derived from determining the point value of every yard line
3. Opponent Adjustments - Ultimately - this is a schedule-based adjustment designed to reward tougher schedules and punish weaker ones.

Boise State and Mississippi State headline the list, with Dan Mullen's team actually playing a tough schedule and Boise State playing one challenging opponent (VT) and giving away a game to a fraudulent team (Nevada). Having said that - earning the #1 Def. F/+ position while playing the 90th ranked offenses means they pummeled just about every team they played. The old adage here: Good teams destroy bad teams.

A simple average of Def. F/+ shows a 50th overall defensive ranking, or about the 42nd percentile of all 120 FBS teams. A ranking stronger than the median team number makes sense here; most are in good conferences that aim to hire accordingly. Florida State comes in at 11th out of 25 on this list with the 41st overall ranked defense. FSU jumped from 48th to 41st after their USC bowlgame performance.

Now let's compare how teams with new defensive coordinators compared to the year previous. (Defensive SOS unavailable for 2009 season).

The table to the left shows defensive F/+ ranking in 2009, 2010, the net change in ranking, and the amount of defensive starters returned for teams featuring a new defensive coordinator in 2010.

We see a very wide range in the net change. The mean of the teams is slightly positive (+2 ranks), the standard error is about 42 ranks. Basically, there is a wide variety of outcomes of defensive performances under a new defensive coordinator.

Stanford (+82), Illinois (+70), and Texas A&M (+56) showed amazing improvements under their respective first year DCs. FSU featured the 5th most improved defense (41 rank improvement) of the group, outperforming in-conference opponents Duke (-6), Georgia Tech (-13), and UVA (-70). Al Groh's 3-4 defensive implementation at GT is going to take some time, but you can't fault his defensive units from his days at UVa.

The correlation between starters returning and defensive performance has long been speculated. Recently, a study shows that a defensive unit's FEI ranking improves by +3.0 positions for each additional defensive starter that returns. Check out the MGOblog report here. Upper-classmen heavy defenses are in general bigger, stronger, and more versed in a particular system.

We can do our own correlation study here amongst first year DCs using everyone's favorite open-source statistical analysis software, R. The data is here. Feel free to follow along and mess around with a great piece of software!

> test=read.csv("test.csv",T); head(test)

We have to do a little prep work to get the data ready for testing. We must convert the "f." ranking to a normal random variable.

> test$f.norm=qnorm((test$f.-.5)/120)

Check. We will now do a Pearson Product-moment correlation test here between the normalized 2010 Def. F/+ ranking ("f.norm") and the # of defensive starters that returned for the 2010 season.

> cor.test(test$f.norm,test$starters)

	Pearson's product-moment correlation

data:  test$f.norm and test$starters 
t = -2.9484, df = 22, p-value = 0.00743
alternative hypothesis: true correlation is not equal to 0 
95 percent confidence interval:
 -0.7702325 -0.1640060 
sample estimates:
       cor 
-0.5321931

We see a nice (and significant) negative correlation. This makes sense with the theory: More starters returning, lower (better) Def. F/+ ranking.

But is there a relationship between the Net Change in F/+ and # of starters returning? That is, how much improvement in ranking under new DCs can be attributed to the number of starters they inherit? We can test this, too.

> cor.test(test$net,test$starters)

	Pearson's product-moment correlation

data:  test$net and test$starters 
t = 1.3561, df = 22, p-value = 0.1888
alternative hypothesis: true correlation is not equal to 0 
95 percent confidence interval:
 -0.1414964  0.6125210 
sample estimates:
      cor 
0.2777543

The relationship is positive (more starters returning inherited --> higher net improvement) as we might expect, but it is not significant.

Boise State returned 10 defensive starters from their 11th ranked D F/+ defense in 2009 and produced the top-ranked D F/+ defense this year. Conversely, East Carolina lost 9 defensive starters from their 2009 CUSA championship team and plummeted to 118th.

But what about a team like Georgia Tech? GT fired Dave Wommack and brought in former-UVA head coach Al Groh and his 3-4 system for 2010. However, even with 9 returning starters GT managed to drop 13 places, near the bottom 1/3 of FBS teams with new DCs.

Another aberration is FSU. Of teams with new defensive coordinators, FSU tied for returning the 2nd fewest starters with USF (ECU lost 8 starters heading into the 2010 season). However, even with only 4 returning starters FSU improved 47 spots (from 88th to 41st). ECU lost 9 starters to FSU's 8 and their DC (FSU's LB and DCIW Greg Hudson), and plummeted to near the bottom in the country.  There are a few things in play here for FSU, as TN readers know. FSU was not as bad, talent-wise, as 2009's 88th D F/+ ranking indicated. FSU also suffered from being out-coached and out-schemed.

So, what are the takeaways from this?

• There's a pretty big variance in how teams with new DCs fared. This can attributed to coaching ability, player receptivity, number of starters returned, quality of talent (not analyzed here).
• The # of starters and Defensive F/+ ranking are very nicely correlated, which corroborates previous research. However, improvements in Defensive F/+ ranking aren't very well explained by the number of starters returning. Upperclassmen leadership is important, but in a new system it can also be a detriment (i.e., old habits are hard to break).
• FSU under Mark Stoops demonstrated the 5th biggest leap amongst 1st-year DC teams, improving 47 spots.
• The 2011 FSU defense will feature 9 returning starters (compared to 2010's 4). According to the MGOblog analysis, that could result in a 15-rank improvement (~32nd). According to many of our opinions, that is probably extremely conservative, given that much of your youth is located along the defensive line and still physically maturing.
  

Sources:

-OneBarrelRum for vetting DC list

-Wikipedia for new coaches in 2010

-Ted Miller (@ESPN_Pac10) for info on Pac-10 teams

- Info on returning starters from Phil Steele

- For more on how to use R, check out this tutorial / walk-through made by Jim Elsner, PhD (Florida State University)