If you recall, before the potential coaching change engulfed the world of Michigan fans (myself included), we were interested in whether or not the 3-3-5 defense could stop Big Ten-style run offenses.  There were various football people who insisted that it could if run properly and those, like Chris Spielman, who thought that it had too many “bubbles” (I fell into the latter camp, fwiw).  This diary is intended to address that issue. Obviously, the issue is more likely to be relevant if RR is retained (the wisdom of which I don’t mean to comment on here).

By "Big Ten-style" run offense, I mean, perhaps sloppily, any offense that does not primarily run out of the spread.  A majority of Big Ten teams now run out of the spread, of course, but the idea floating around toward the end of Michigan's regular season was that the 3-3-5 could not stop traditional Big Ten-style rushing attacks.  I accordingly looked for results from the 3-3-5 against run offenses similar to those used by Iowa, Wisconsin, etc. 

[Edit: I may not have been clear enough in defining "traditional Big Ten-style rushing attack."  I'll try to be more specific:  A running attack that mostly involves a QB under center and mostly involves the use of a fullback or at least one tight end on running plays.  Why am I using that definition?  I'm using it because it because I think that's what people mean when they say that the 3-3-5 can't stop Big Ten-style rushing attacks.  This criticism tends not to be very specific, so it is difficult to address it with a great deal of specificity.  It is difficult to catch the boogie man.]

For obvious reasons, WVU offers the best and most relevant example of a team that uses the 3-3-5.  I decided to look at WVU's success (or lack of it) against Pitt, Rutgers, and UConn from 2005-2009 as a measurement of the effectiveness of the 3-3-5 against Big Ten-style run offenses.  I think we can agree that those teams run using offenses that are similar to traditional Big Ten-style offenses or, at least, that they generally do not run out of the spread (or some exotic non-spread offense like the triple-option, wing-T, or lonely antelope*). 

Here, we can imagine the protest that WVU's 3-3-5 may work in the Big East but would not work in the Big Ten, as the Big East is inferior.  To this I have two responses:  First, WVU plays against Big East-caliber players when it plays the above teams but it does this with more-or-less that same caliber of player on its own side.   Off the top of my head, I'm fairly sure that Pitt, for one, nearly always out-recruits WVU, so WVU likely does not have a talent advantage in that match-up.  As to UConn and Rutgers, I believe that WVU typically out-recruits those schools, but certainly not to the degree that Michigan or any other top-echelon Big Ten School does.  I might be wrong on these assumptions and, frankly, I don't feel like looking it up.  The motivated reader is welcome to shed light here.  In any event, I hope we can agree that WVU does not have a significant talent advantage over Pitt, UConn, and Rutgers and so the “But it’s the Big East!” argument fails on that count. 

Second, it is possible that Big East coaches are inferior to Big Ten coaches so that Big Ten run offenses are more nuanced and difficult to defend than are Big East run offenses.  I have no answer to this objection.  I don't have the time or expertise to offer a meaningful reply.  WVU's 3-3-5 offers us the most relevant example of that defense, though, so we have to work with what we have.

With those issues addressed, below are the yards per carry (ypc) of Rutgers, UConn, and Pitt from '05-'09 for their entire seasons, not including their ypc against WVU in those years.  Below also are the ypc for each team against WVU (WVU ypc) for each of the relevant years and the difference between the teams' season ypc and WVU ypc for each year.

 

 

Rutgers	ypc	WVU ypc	difference
2009	3.7	1.9	-1.8
2008	4.0	2.7	-1.3
2007	4.8	4.6	-0.2
2006	4.7	4.7	0.0
2005	4.4	3.1	-1.3

The average difference in ypc between Rutgers’ season ypc and its ypc against WVU was -0.9 ypc for '05-'09.

 

UConn	ypc	WVU ypc	difference
2009	4.3	3.3	-1.0
2008	5.1	4.8	-0.3
2007	3.9	4.1	0.2
2006	4.6	2.9	-1.7
2005	4.2	0.3	-3.9

The average difference in ypc between UConn's season ypc and its ypc against WVU was -1.3 ypc for '05-'09.

 

Pitt	ypc	WVU ypc	difference
2009	4.9	5.0	0.1
2008	3.6	4.3	0.7
2007	3.7	3.0	-0.7
2006	4.1	0.1	-4.0
2005	3.1	3.9	-0.8

 

The average difference in ypc between Pitt’s season ypc and its ypc against WVU was -0.6 ypc for '05-'09.

 

Further Thoughts and Conclusion:

I realized as I was compiling these statistics that I did not account for sacks.  WVU averaged roughly 31 sacks per year from ’05-’09, a seemingly middling amount (and roughly what Michigan averaged for those years).  A brief review of the rest of the Big East’s sack totals for those years suggests that WVU was not a stand-out sack-wise in the conference for our time period.  It is none-the-less possible that, despite WVU’s average performance as a pass rushing defense, WVU had a disproportionate amount of sacks against UConn, Pitt, and Rutgers and that these sacks artificially lowered those teams’ ypc against WVU.  WVU recorded seven sacks for 50 yards in one of the games listed above.   

I am not inclined, half out of laziness, to go back and pick through the sacks-against totals for UConn, Pitt, and Rutgers against all non-WVU teams and then against WVU.  It strikes me as unlikely that those teams’ ypc were lowered by sacks against WVU any more than they were lowered by sacks against all other teams.  Again, though, the motivated reader is welcome to illuminate this issue.

UConn, Rutgers, and Pitt rushed for an average of 0.9 ypc less against WVU during our time period than they did against all other teams.   WVU’s worst performances were against Pitt in 2005 and 2008, when Pitt increased its ypc by 0.8 and 0.7 against WVU.  Other than that, WVU routinely held teams to their average ypc or less.  Twice (UConn 2005 and Pitt 2006), WVU stoned teams to the tune of a decrease of at least -3.9 ypc. 

As a comparison (and here my laziness shows up again), Alabama’s very strong rush defense of 2009 held its Division I opponents to 1.6 ypc less than those teams rushed for during their entire seasons (here, though, I did not subtract those teams' ypc against Alabama from their total ypc, so the above number is artificially deflated).   The fact that WVU’s performance against UConn, Rutgers, and Pitt does not match Alabama’s (deflated) performance of 2009 shows us that WVU was not dominant.  Despite this, I am inclined to say that WVU’s performance was pretty good or, if you like, A-okay. 

Despite the imperfections in the above analysis, I now believe that the 3-3-5 can succeed against traditional Big Ten offenses.  Of course, replicating WVU’s success would require recruiting appropriate players for the 3-3-5, developing them, and then using the 3-3-5 at least as well as Jeff Casteel does at WVU. 

 

 

*There is, of course, no such offense as the wing-T.