Michigan Museday: Revisiting the Decimated Defense

Submitted by Seth on August 31st, 2011 at 10:00 AM


Note: During the season I plan to post this kind of stuff on Tuesdays but since Aug. 30 set a new MGoRecord for total words on the front page, I waited.

On Monday Brian posted the secondary preview for 2011. It had its share of woe, and the "Never Forget" poster again. But among the now-usual fits of "this can't go well" at FS, tiptoeing around Kovacs's Ecksteininess, and general radioactive fallout from when having a pulse put you on the cornerback 2-deep, there were things that are not so familiar: a capable senior cornerback rescued from a bunker, returning starters, a few guys here and there playing positions that suited them.

Attrition was hell (see: Google doc) but the unit has begun the slow climb back:

Team Recruited Diff v '09 Retention Diff v '09
Penn State 44 +4 86.36% +9.44
Notre Dame 50 +7 80.00% +3.72
Michigan State 54 -10 77.36% -7.05
Ohio State 56 -3 73.31% +4.66
Michigan 58 +10 63.79% +5.46
Alabama 67 -16 62.69% +0.04

The retention % is that of total defensive recruits '07-'11 still on the roster. Under two thirds is still bad but it's not worse than a team which signs five guys for every four available scholarships, like it used to be. Also when I first did this Michigan had lost 2 of 5 guys from a starting position more like Penn State. We needed dudes. As the secondary depth chart says, we now have dudes.

That's not to say everything's hunky dory in Hiroshima '46. Most of the casualties were upperclassmen and Michigan's more highly rated players.

Team Best 22 Best 22 left Diff.
Alabama 5.91 5.84 0.08
Ohio State 5.89 5.82 0.07
Notre Dame 5.84 5.83 0.01
Michigan 5.80 5.67 0.13
Penn State 5.75 5.72 0.03
Michigan State 5.65 5.61 0.04

I'll explain what these are in a second but you're meant to read it thus: imagine the two-deep of each team is made up of all upperclassmen with the above rivals ratings. So attrition has taken Alabama from a team full of Top 100 guys to a team of Top 250 players.Recruitin is Legit Yo

This is my attempt at taking the noise out and looking just at how much performance loss is caused just by attrition itself. It's a hack combination of Rivals Rating with the value of underclassmen adjusted down so that a 6.1 (5-star) is worth 5.7 (= a high three star) as a freshman, 5.9 as a sophomore, 6.0 as a junior, his full ranking thereafter, and on down. Then I just took the highest 22 scores for each team before and after attrition. It's a cheap formula that rates a player 80% by his rivals rating and the rest on account of his age but the image at right shows the concept works: the higher up the depth chart, the higher the ratings.

The thwack that Michigan took is more visible when you look at it from a depth chart perspective:

Michigan Depth

Click that to get it readable if you can't see it all from a glance. Also note the scales are a bit off; Bama goes up to 6.1 but OSU and ND stop at 6.0, PSU and MSU at 5.9. The visual here is you want your secondary color (attrition) to the left (see the devastation wrought on Bama 3-stars) and your starters to the right.

Bama took some big hits to its 4-stars but retains the highest value among starters. Michigan meanwhile seems like we were facing that guy with the unlimited airstrikes hell bent on killing any worm named "Safety*".

The result is a two-deep that doesn't really have options until the 3-star level whereas the other guys only have a few odd dudes who weren't heavily recruited on the field:


(Again, click = big). That's not…terrible. Given the players who've made it far enough to likely see the field, even with all the attrition Michigan could be expected to field a defense not so different from that of MSU, who should be…okay. Okay is better than we're predicting now. Then again, don't confuse this year's young MSU defense with last year's Greg Jones-inclusive defense:


Yes that's Michigan (demonstrating perfect pad level) with more 5th year seniors than anybody save Penn State, whose starting 11 have been around long enough to remember when their coach built the pyramids. And that's MSU starting a lot of sophomores and true freshmen. Bama, OSU and ND are mostly upperclassmen (ND is weird about redshirting still). Michigan leans much more sophomoric. This is a huge improvement from last year when the seniors were few and the freshmen were legion. Young means high variance—some days may go very well, others very not well.

Any difference from here would be in player development and coaching. Also, positional (showing expectation values of starters and two-deep):Expectations-Depth chart

  Starters Two-Deep
School DB DL LB Avg DB DL LB Avg
Alabama 5.93 5.70 5.93 5.86 5.73 5.69 5.76 5.73
Notre Dame 5.83 5.90 5.80 5.84 5.66 5.81 5.76 5.73
Ohio State 5.75 5.71 5.83 5.76 5.77 5.66 5.65 5.70
Penn State 5.63 5.63 5.77 5.67 5.53 5.53 5.77 5.59
Michigan State 5.48 5.64 5.58 5.56 5.43 5.56 5.51 5.50
Michigan 5.28 5.58 5.70 5.50 5.58 5.58 5.65 5.60

..and there's the cheap flight. You can't blame Kovacs; I took out the walk-ons for this specifically because his walk-on-iness isn't the glaring problem. Sophomore 3-stars are. Now guess where Michigan was hit the hardest by attrition?




That shows the sum total of the projected value of middle-3 stars (5.6 to rivals) and higher recruited for each position. The primary color bits are the guys who are still around; the maize those who are gone. The Ohio Bobcats' colors are green and peach I think.

Never Forget

5143016631_5e2cd49ef6 stagesofgrief

Unlike some other peoples' versions, my N.F. poster has Cullen and Vinopal on it!

Did we learn anything here, other than that you can print a chart to pdf in excel? Well yeah: attrition was a great big nuclear blast that will take years to recover from. Even if the talent on hand doesn't regress or get hurt, thus exposing further weaknesses, the starting point for this Michigan defense is that of a Same old Spartans unit.

Blame is a bit less easy to assign, though some of the flameouts and underclassman transfers in '09-'10 were either directly or indirectly pinnable on the old staff.** Two years of recruiting after bad years (2008 didn't seem to have the same effect) and some resulting recruiting holes at MLB and WDE make any climb back among elite D's a long-term project—probably not until 2015.courtney-avery-coverage

Yet there is hope all over the place, thanks to there being dudes. To some degree players retain most of the traits that went into their rating (Jordan Kovacs can't develop into Marcus Ray), but there will always be 3-stars who grow into defenders more than capable of playing, as Hoke/Mattison call it, "Michigan defense." The chances that 2/3 of Christian/Talbott/ Avery would be capable last year of said 'Michigan defense' was near nil. The chances that one or two decent players emerge opposite T-Woolf among five 3- and 4-star-ish freshmen, two 3-star sophomores and a 5th year walk-on is much higher. Eight shots in the dark (several after practice shots) are better than three.

And next year more dudes arrive. Hugely hyped, turned-down-offers-from-all-of-the-above dudes at defensive end and middle linebacker, and a smattering of the same at the other spots so that 2014 gets a nice selection (MOAR DTs please, kthx).

And though the odds be against us, you never know: Michigan could not have its best two defensive players knocked out by injury this year. Maybe (a planetary version of) Heininger or Brink will be the next Kovacs. Maybe Greg Mattison is a wonderful teacher who, like Pelini in '08 and Dantonio in '07, gets a few lights to go on from previously unheard of sources.


* Pro Tip: Don't name your worms for Michigan Heisman winners unless you want your MSU alum opponents to gang up on you (as if Desmond hasn't had enough harassment from that ilk!)

** Things you can blame on RR & Co.: 1) Not getting enough interior linemen after the two soft guys flipped in '09, 2) Recruiting four guys (Witty, Dorsey, D.Rogers, and Kinard) who couldn't get past the NCAA Clearinghouse or M's higher standards, 3) sucking so hard that by the Class of '11 the top regional recruits were looking elsewhere, and 4) A string of the worst LB coaches in M history and a manic concept of positional switching so that the guys on hand were hardly given opportunity to improve at any one position in one defense.



August 31st, 2011 at 10:05 AM ^

Those charts you prepared are outstanding.  Extremely informative and, at the same time, very pleasing to the eye.  All the pretty colors made me feel happy to learn this somewhat complicated information.



August 31st, 2011 at 11:05 AM ^

i'm not quite as enamored of the graphs. you can't see all the information in them (e.g., in a few, you can't see what's going on with alabama when there are troughs blocked by all the opaque stuff in front). you also have to put some work into deciphering them if you aren't an excel graph-builder yourself.  so it seems to me the information in some of them could be made much clearer using a different graphing program, or just different options within excel.


August 31st, 2011 at 11:20 AM ^

From the graph labeled "Recruiting is legit, yo" down to the first one under the "Assume the positions" heading, information is lost because the representation is opaque. I assume you did it like this instead of just making a line graph (like Michigan-Depth.jpg but without the opaque filling) because line graphs would be ugly, right? This is probably just me being pedantic, but you would also be able to see everything that way. 

I think the other complaint, about them requiring some work to decipher, is just me being lazy.


August 31st, 2011 at 1:09 PM ^

Since you appear to have asked, I will offer some constructive criticism. Please do not be offended by my direct style of writing. It is just the simplest way to communicate.

The labelling on the graphs is too small to read unless you click on the graphs and that makes it hard to directly compare them. There is very little labelling on the bar graphs. I'm forced to guess what the y axis represents and what yellow and blue mean. I'm not going to guess.

Due to my career, I tend to skip past the words and straight to the data. Then I see if my conclusions are the same as yours without being lead there by you. Alternatively, I don't waste my time reading the text if the data or data presentation is not compelling.

These graphs are too hard for me to understand. They certainly look pretty and I know from reading things you've written previously that is important to you. Personally, I think you could communicate the very interesting data that you have using plainer methods. Perhaps a summary table--Michigans average is X and OSU's is Y so of course they kick our ass.

I guess there are some summary tables but they are very busy and confusing. For instance, there is no label for what those numbers mean. Plus, all the numbers are five point something. Nothing jumps out at me because all the numbers are so close. Maybe a figure legend, like you would have in a scientific paper, would help. That way, I am not forced to scan the text to answer my questions.

I think that you have so much interesting data that you're overloading your audience. You might be better off sticking to a few important high points rather than drowning your readers with reams of data. Maybe you can summarize other points near the end in the text.


August 31st, 2011 at 10:35 AM ^

sheer volume of charts I am now by default an engineer.  Thank you Brian I now look forward to recieving my Engineering Degree in the mail shortly.


Thanks I couldn't have done it without you. 


August 31st, 2011 at 10:59 AM ^

Naw, Michigan Museday it is.

I had so much hell picking a name you wouldn't believe. I put this in the hopper at 11 or so last night; as of 10:00 the name was going to be "Uncle Lloyd's Tremendous Watch."

Other top names considered:

  • Tremendous Tuesday (HT: my buddy James)
  • Miso Soup (HT: StephenRKass)
  • T_______ Tuesday (where each week the 'T' changes)
  • tl;dr (HT: Dark Blue)
  • tl;dr for Beavis
  • Somebody's Wrong on the Internet (HT: jdfogel)
  • Three Days Later (HT: Six Zero)
  • Tuesday Tink Thank (HT: BlueDragon except for the part about "Think" pronounced like Tweety would say it
  • Summer Glaum (I'm the only person who liked that)
  • The Apostate, Three Days Late (HT: Dave, because Julian the Apostate and whatnot but I still accidentally read that word as "App State" from time to time and unlike certain nameless athletic directors of major universities, I don't like being reminded about that every week.)
  • Miss _____ (play on prefix. Each week it's Miss something, like Miss Analysis, Miss Taken Identity, Miss Depointe, etc.), with cartoon ladies as leadoffs. Episode 1 would have been Miss De'Seniors

Not considered b/c I work with the guy it's parrotting, but I still haven't stopped laughing for 24 hours:

Also good:

  • Hoban "Misopogon" Washburne's Tuesday Dinosaur Playground (BlueVoix)



August 31st, 2011 at 10:35 AM ^

Those charts!  So many of them with such nice colors!  Just awesome.

As to the substance, while I will not use such things as math and stats to justify my points, I agree with your conclusion that (1) we are on the uptrend, and (2) we can expect continued improvement  until around 2015 at which time the rebuild will be complete and we can be an elite level DT. 


Blue in Seattle

August 31st, 2011 at 10:38 AM ^

But I think it's getting clearer on assigning blame.  I'm not a big tracker of the recruiting rankings, but on an article that compared Hoke's last partial class and the next potential class to the previoius years, it seems like Rodriguez did have good recruiting classes in there.  I'm not sure how you blame attrition from those classes to anyone else but Rodriguez.

The true test will be this season.  If this defense improves throughout the season, that to me will be an indication that coaching and player development were huge issues and are directly on the previous coaching staff.  Lloyd Carr did not leave a traditional (in our minds) level of talent in the cupboard, but the fact that what was there was not developed, and what was selected to replace that bare cupboard continued to not pan out is directly on Rodriguez.



August 31st, 2011 at 11:22 AM ^

Or not having freshmen and walk-ons everywhere throughout your defensive 2 deep. I remember in year 2 a point where we were starting 3 walk-ons on the defensive side of the ball. But you're right, RichRod should have recruited more upperclassmen out of high school. I swear people forget that RichRod only had 2 (or 3 if you count that first one of Lloyd's) classes of his own recruits that are not true freshmen on the team now. Whatevs...potato potatoe.


August 31st, 2011 at 11:20 AM ^

I hate math and stats...and since I have statistics, advanced algebra and financial accounting this semester I am staying away from any post by misopogan. I need pure football talk


August 31st, 2011 at 1:53 PM ^

One point is always lost when people use this decimated defense series to rationalize RR's failings with the defense.  This series compares Michigan to elite defenses plus MSU.  It does not compare us to the dregs of the Big Ten with whom we shared the bottom of the standings during the RR years.  This work might explain a situation where we finished in the middle of the Big Ten, below the top tier, but it does not explain us being at the bottom.  Show some graphs with us versus Purdue, Indiana, NW, etc.  I bet you will see that our defense had more talent than those teams.  Yet, those were the teams we finished with or below during the last 3 years.

This series shows why we were not an elite defense.  It does not explain why we were ranked below 100 out of 120 teams!  That cannot be blamed on talent.  There is no way we had worse talent on defense than 85% of teams in the FBS. Clearly, coaching is a major factor in the abject failure that was our defense of recent vintage.


August 31st, 2011 at 3:25 PM ^


I am always impressed by the time you spend on your writings here, and I enjoy them.  Your decimated defense with the data and graphs is a tour de force.  It just frustrates me that in the past whenever the failings of the previous regime on defense were brought up, people would just write "DD-decimated defense" and think that it explains everything.  It certainly is valuable and explains a lot, but it does not explain everything.  That is why I wrote what I did above.  It was not meant to criticize you.  Your work is part of what makes this site great. Keep it up.