How Much Should The D Improve? Not Enough

Submitted by Brian on July 12th, 2011 at 12:17 PM

Let's take some small sample sizes and extrapolate wildly. It will be fun. Here's Bill Connolly breaking down expected improvement from teams that return varying numbers of defensive starters:

So Cincinnati returns 11 defensive starters. That's probably a good thing, right? But how good? And how much can a bad defense improve in one offseason just because of experience? Let's take a look, shall we?

Average Change In Def. F/+, Last Three Years
N Avg Chg in
Def. F/+

1 1 -12.4%
2 4 -10.9%
3 10 -8.4%
4 32 -2.1%
5 53 -1.1%
6 69 -0.5%
7 85 1.1%
8 56 1.5%
9 37 4.2%
10 9 6.0%
11 3 5.4%

So basically, if you return between five and eight starters, you are likely not going to change much, but three or fewer is a problem, and nine or more is a good thing.

F/+ is Connolly's advanced metric; it's play-based instead of drive-based like FEI. Don't be fooled by the % symbol—the metric is percentage based and from context it's clear the difference is meant to be added to the score, not multiplied. Since the best defenses are around +17% and the worst around –13%, 6% is about a fifth of the entire scale.

Michigan is, unsurprisingly, right at the bottom of that scale at 115th. They were 12% worse than an average defense down-to-down. The good news is they return 9-ish starters, losing Greg Banks, James Rogers and Jonas Mouton while reacquiring Troy Woolfolk. (They also lose Ray Vinopal and Obi Ezeh, but Ezeh had been replaced and Michigan should get JT Floyd back so let's call it a wash.)

The numbers are thin at both ends of the spectrum but, hey, extrapolating wildly from small sample sizes. Doing so says Michigan's defense will storm forward from 115th nationally to…



I have no source for this, unfortunately.

But wait! Our sample sizes are not small enough and our extrapolation is not making out with other nubile young extrapolations in front of a television camera. Bill added a second factor, the previous year's defense, and finds that a defense with an F/+ under –10% that returns nine starters should expect (for a given confidence level that is not high at all) to improve by 8.6%, which would see them get to…



You might be able to argue that Mike Martin wasn't right and the team was even younger than the average team that returns nine starters and GERG is rubbing stuffed animals on the faces of other stuffed animals at a tearful tea party and for the first time in a long time they'll just run one damn defense per year and that they should expect to improve even more. You're probably setting yourself up for disappointment. Like installing the spread 'n' shred, digging out of a hole this big is a multi-year project.



July 12th, 2011 at 12:26 PM ^

Mattison/GERG > Magee/Borges

Therefore, based on this simplistic analysis, hopefully the progression on defense will outpace the regression on offense


July 12th, 2011 at 12:54 PM ^

I also haven't seen the sun rise tomorrow, but I'm pretty comfortable in predicting that it will.

Besides, we also haven't seen the defense improve.  It might get worse.

/checks math

I'm sorry, I'm being told that there is no mathematically feasible way that the defense could be worse.  My apologies.



July 12th, 2011 at 1:07 PM ^

I literally have watched hours of footage of our defense from last year. And I have seen the spring game a bunch of times. 

There is nothing like a freshman to sophmore jump in both experience and physical ability. Having almost 1.5 years of college weight room training and coaching has a direct and immediate impact on a players ability to do one thing. Tackle.
I consider myself the tackling expert 'round these parts, and trust me when I say sophmore >>>>> freshman with respect to this stat. 

The key to the defense this year will be, I think, Craig Roh and Jibreel along with a headhunting Cam Gordon/Jake Ryan coming of the edge. Our corners are able to cover for about 3-4 seconds, but if we can blitz the edge and contain, force things into Martin and Demens and get a sack or two we can make a huge impact directly on two things. 
1. letting our safeties edge back a bit, not being committed early to run support
2. Letting our corners play pass first. 

RVB and BWC (Big Will Campbell) will be key. they must stand tall at the point of attack and force the runner to choose to cut back or bounce out.


July 12th, 2011 at 1:16 PM ^

"RVB and BWC (Big Will Campbell) will be key. they must stand tall at the point of attack and force the runner to choose to cut back or bounce out."

From what my people (everyone) have been telling me for 3 years, BWC will have no trouble at all standing tall at the point of attack.  So I guess we are good?  I am off to buy my Rose Bowl tix.

One Inch Woody…

July 12th, 2011 at 2:09 PM ^

GERG is notorious for doing two things: A.) Turning Syracuse into a loser school after a period of medicore-to-good teams and B.) Turning the Michigan defense into a pile of crap, but crap that is positioned nicely so as to give opposing wide receivers 20 yards of running room. It's not like Michigan all of a sudden is recruiting worse talent than Minnesota for god's sake so what else can you blame a terrible defense on other than injuries and coaching scheme.

With the return of a four-man front, we get something which most season-to-season statistics cannot account for and that's some (hopefully adequate) pressure on the quarterback and linebackers that act as linebackers and not little bags of feathers running into the brick wall that is a freed up wisconsin offensive lineman. With less time for a quarterback to pass, he is undoubtedly going to make worse throws, so our secondary should improve. That, and two natural safeties (Marvin Robinson and Carvin Johnson) should make passing a hell of a lot harder than last season.

A final thing to look at is time of possession and how the offense relates to defense. I would argue that given a young, inexperienced defense, forcing them to play for a large portion of the game can only magnify the negative defensive statistics. For example, if our offense did not score within 5 plays of every possession during the 2010 Illinois game, our defense would be on the field for less time and therefore the opponent could not rack up as many points/yards. So a "regression" in the offense is almost a blessing in disguise, and something I look forward to. However, the conversion efficiency per yard (points per yard) of our team over the whole season was TERRIBLE. Ranked 9th in the Big Ten (Maize and Brew did just for fun statistical analysis of this). So if we improve that and tone down our offense, the defense should become insanely better.



July 12th, 2011 at 3:47 PM ^

Not another TOP acolyte...

Look, the reason the defense spent so much time on the field last year is because they couldn't stop anyone. Teams ran long drives against them, burning clock and running up the score. Sure, scoring less quickly would have given the opponent fewer drives - but it would also have given our offense fewer drives. So the final score may have been lower, but the result would have been the same. Do you honestly think any coach would look at a 5 play TD drive and say "damn, I wish it had taken us 20 plays to get that TD"? (Except maybe at the very end of the game/half).

Also, you neglect to note that a major driver of our poor yards-to-points ratio on offense was due to the lousy defense and awful field goal kicking. The offense got the ball off a kickoff almost every drive, requiring 75 yards/TD even if the offense was perfect. And if they got 55 of those yards, they'd miss a field goal that most teams would make. Does the offense need to improve consistency? Sure. But I rather doubt Borges looks at the RR system and says, "damn, those guys just got way too many big plays - we should stop doing that".


July 12th, 2011 at 3:53 PM ^

To quote Greg Mattison,


It's great if the O score quickly: "I hope we score in 3 plays every time they're out there. Because our job is going to be to get off the field in three plays." As long as the offense isn't turning the ball over, they'll help the defense. The defense earns their own rest by getting off the field on third downs. "Your job is to do what you're supposed to do on third down to get off the field." They have started installing the nickel/3rd down package, because stopping third downs is so crucial.


July 12th, 2011 at 5:20 PM ^

There are two ways to use TOP here:

(1)  The "TOP acolyte":  Claiming that having an offense score quickly hurts the defense's effectiveness (e.g. by wearing them out) and therefore hurts the team's chances to win.  Mattison refutes this concept in his quote above.

(2)  Claiming that having an offense score quickly skews a defense's statics.  This has a basis in statistical fact (having an offense score quickly generally gives each side a greater number of possessions per game, which tends to inflate certain stats). 

The poster above appears to be in group (2), but you derided him as if he were in group (1).


July 12th, 2011 at 5:27 PM ^

Eh, maybe it's 2, but he also stated that an offense that scored less quickly would be a "blessing in disguise" that he "looked forward to". In other words, he's arguing that increased offensive TOP is a good thing in general, not just statistically. That seems to point to 1.

Agree however that 2 is a real thing, at least as long as you're looking at possession sensitive stats ("total defense", "scoring defense"). However, Michigan's D was still ranked near the bottom of possession insensitive stats (FEI, for example), so they still would have been bad even if the offense scored less quickly.


July 12th, 2011 at 4:17 PM ^

As Fremeau (the F in FEI) notes in the latest HTTV, Michigan had way too high a percentage of three outs to be an elite offense. In other words, the offense was a mixed bag. There is no point in suggesting big plays or scoring too quickly were a problem.

One Inch Woody…

July 12th, 2011 at 5:34 PM ^

Right, I should have clarified in my post, but I simply meant to say that because our defense was so terrible, the opponent's TOP magnified the statistics. There could have been other teams in the country whose defense was just as poor as ours, but their offense took up more time and so the statistics showed that their defense only allowed 400 ypg. Oregon is a good example of team that has a really explosive offense and whose defense is actually capable of forcing an opponent to go 3-and-out. (We could force 4 downs occasionally, but teams like Miss State would just go for it on 4th down and convert anyway). Their defense, as a result, couldn't have all of these yards racked up on it because they'd get off the field fairly quickly and give their offense a little more time to score.

And I totally agree about the field goals. Maybe our points per yard gained wouldn't be so absolutely terrible if we had been able to convert field goals.


July 12th, 2011 at 4:30 PM ^

"THE scheme... THE scheme... The scheme...." No one man is more important than "The scheme".  I agree with your posting.

I'd assume that a vast majority of the defenses researched did not change their schemes.  A 3-3-5 in the big 10 is horrible as we all know, so I think by simply changing to the 4-3 will be an automatic improvement that's not reflected in the math thats in this article.  However, this could very well be negatated by the fact that it indeed is "new" to all of these players, but hopefully that's were Mattison the Great comes in and learns em quick.

One Inch Woody…

July 12th, 2011 at 5:38 PM ^

Yup, and since a 3-3-5 is essentially a magician's 3-4, it should be very good against the run. However, lining up linebackers less than a yard away from the nose tackle thereby forcing them to commit to being eaten up by an offensive lineman, does not pass for an adequate WV-like 3-3-5 that could actually work here.


July 12th, 2011 at 1:15 PM ^

I could easily see an offense that finishes with far less gaudy overall numbers (which we largely attained by going for 600 yards on all our non-conference opponents) yet actually become a more consistently irritating force for opposing defenses.

Take OSU's offense last year. Their numbers were not as strong as ours but in the Big Ten season, I'd argue it was more effective. The constant threat of Pryor scrambling (something we didn't really have with DRob in our passing game but I hope we do this year) made any third-and-long do-able. They played to their defense and often worked slow but when they had to score, it wasn't an easy unit to get off the field.

If we are more methodical and diverse, we can be the same way this year.




July 12th, 2011 at 2:16 PM ^

I think the key to a more consistently potent offense will be a consistently decent running back. With Denard being are main/only threat last year it was easy for defenses to key in on him, thereby making the offense marginal when they were successful. There needs to be a RB not only to share the load, but to provide another potent option for when Denard is being overwhelmed. It'll be interesting to see who (if any) steps into that role.


July 12th, 2011 at 4:12 PM ^


I could easily see an offense that finishes with far less gaudy overall numbers (which we largely attained by going for 600 yards on all our non-conference opponents)

Even though I think it's possible for Michigan's offense to finish with far less gaudy overall numbers yet still be better (HOLD ON TO THE DAMN BALL), remember that the FEI rewards teams less for playing well against bad teams than good teams. The FEI will not have Michigan's offense at #2 if it did nothing against good teams in the Big Ten. Hopefully, the discussion is moot in the future as improved defense + ST more than offsets the offense's lack of numbers, but reading this makes me facepalm on the inside.


July 12th, 2011 at 4:41 PM ^

My knock on FEI is that it's too yardage-dependent.  Yards don't necessarily tell the story.  In our case, our average starting field position was so regularly bad that we could often rack up a lot of yards on a drive without scoring, which is not the offense's primary goal (especially in a year like 2010, when our defense could not be protected by any amount of field position).  And yards are easier to come by the farther you are from the opponent's endzone, when you have more room to maneuver and the defense is more willing to concede modest yardage.  It gets harder when you're in close, and in our case, our redzone offense was very disappointing. 

The most blatant case of yardage being deceptive was the Iowa game, in which we outgained them by about 150 yards but were beaten by 10 on the scoreboard.  We "held" Iowa under 400 yards almost solely because their scoring drives weren't that long, while we consistently started inside our own 20. 


July 12th, 2011 at 6:46 PM ^

Yardage is not a perfect method to judge an offense because bad defense/special teams play can give an offense more opportunities to gain yards than if the defense/special teams are actually doing their job. The primary reason I think yardage > points is because the short-term goal of every offense on every play is to gain enough yards to stay on the field (or score a TD, whichever comes first). Scoring points requires that the offense succeeds at its short-term goal until it reaches the endzone, which can require anywhere from 1-10 short-term successes (EDIT: for the offense to score points. A field goal (special teams points) requires the offense to fail but be close enough for their kicker, while defense/punts/kickoffs can score independently). Sadly, the only way to find a fatal flaw in FEI is to break down FEI and all of its inner workings to know just how Michigan ranks #2 offensively over every team in the country except Auburn, but I don't know their formula :\

On a side point, the only thing I disagree with is the quote "our redzone offense was very disappointing," because this is probably based on redzone efficiency, where Michigan only scored on 79% of its red zone possessions for 82nd in the country.  OTOH, Michigan was 24th in the country in points per redzone possession, and almost certainly ranks higher than that in red zone touchdown percentage, where Michigan scored 40 of 56 redzone possessions. I trust either of these metrics more than redzone efficiency when judging the offense, since redzone efficiency is a strong function of field goal kicking. Playing devil's advocate with myself, could these redzone stats be helped by great showings against bad teams? Of course. So the next step is to quantify how much more than other teams in FBS is Michigan helped against cupcakes than the next team, or include drives that end at just outside the current definition of red zone (which will hurt Michigan because of its lack of field goal kicker). Assuming some mean enhancement for every FBS team against cupcakes/bad teams, it's hard to believe Michigan truly gained all/most of its numbers against nonconference opponents when FEI has Michigan at #2 and I haven't seen the data that proves it wrong.


July 12th, 2011 at 2:20 PM ^

Since Brian's predictive powers are weak at best I won't lose any sleep over any of this. Let's wait and see how things pan out in the fall. Personally I'll be cheering for the team to exceed expectations - will you do the same or does it serve your purpose to see the defense ranked 99th in the country?

To use your own terminology "save us your hymnal cooler pooper".


July 12th, 2011 at 4:28 PM ^

So offensive regression is as likely as the sun rising?  Why?  32.8 ppg isn't that remarkable.  Toss out the three OT periods against Illinois and the average was 30.9 ppg in regulation.  Is that really impossible to improve upon?

We might gain fewer yards, but not necessarily because our offense will be worse, but because our defense will hopefully give the offense better starting position.  By the measure that actually counts - points - I think we can improve upon last year.


July 12th, 2011 at 12:26 PM ^

I might agree with that assessment if our OC continued to be GERG.  But we now have 9 returning starters and Mattison.  Will this defense be earth destroying?  Naw.  But I expect we might rise to the level of "average" or "slightly better than average."  I expect we are going to be pleasantly surprised this year by the improvement. 


July 12th, 2011 at 1:48 PM ^

I have it from good sources that when you run the same calculations for the 72 T.C. Williams High School team there is no chance they could have won with the amount of returning starters or players added because of forced busing.  Should not have won that state championship.


Coordinator is a crafty s.o.b. for Michigan.  He'll get more out of a more experienced pool of players than what Gergy would have.  That counts for somp'n in these perts.


July 12th, 2011 at 1:56 PM ^

Would be very interested in Brian/someone with far better math/statistical skills than mine taking a look at the pretty staggering year over year improvement the illini had, it seems like the only thing out to give us any/much hope about the D for next year.


July 12th, 2011 at 12:38 PM ^

Yeah, this post pretty much confirms my expectations for the defense this year, and will only sadden people with unrealistic expectations. If we get the average improvement for this defense based on returning starters and experience, completely ignoring the staffing changes, we'll be ranked 82nd. Once you consider the hiring of a well regarded defensive coordinator that actually understands the scheme he will be running, we should improve to average-ish, trending towards excellent in future years. I am happy with this considering what we saw on defense the past few seasons.


July 12th, 2011 at 12:48 PM ^

GERG is being scape-goated to some extent.  He's not good - that's pretty well established - but he's also not the main reason last year's D stunk.  That reason was talent and experience.

The upside for this defense is being an average unit in the conference.  One need only look at personnel and compare our starters to the rest of the league:

1 Above average starter (established):  Martin

2 Average starters (established):  Van Bergen, Woolfolk

2 Average starters (projecting):  Roh, Demens

3 Below average starters:  Kovacs, Floyd/Avery, Gordon

3 Cross your fingers:  DT, WLB, FS

You have a lot of below-average weighing down the above average.  Some of the above will be wrong (e.g. Gordon and Roh could be all-conference) but it goes both ways (e.g. Gordon could be benched and Woolfolk could be hurt).

However, its extremely optimistic to expect this defense to be above average.

Mattison may be brilliant, but it takes time.  The key for this team is this: maintaining a top 3 offense in the conference.  With that and an average defense we can contend for a conference title with just a smidgeon of fortune mixed in with a favorable schedule.

A coaching staff change, even an upgrade, comes with costs for installing a new system - indecisiveness, uncertainty, confusion, inefficient timeouts, blown playcalls, etc.



July 12th, 2011 at 12:55 PM ^

I don't disagree.  We will not be world destrying overnight.  But I expect we will be pleasantly surprised by the amount of improvement a top quality DC will make, even with below average talent by historical standards for Michigan.  Remember the power of this one truth: "I coached Ray Lewis."