Michigan Museday Charts Shoe's Arm

Submitted by Seth on December 20th, 2011 at 8:00 AM


For those tracking Denard's passing acumen the tale has been one of major progression before 2010, followed by regression in 2011 followed by re-progression as he a.) grew more comfortable in Borges's offense, b.) played more out of the shotgun, and c.) gave his staph infection time to heal.

If you were reading the weekly previews this season you would have noticed the space for Michigan's passing game was consistently fretting about Robinson's accuracy. This would be followed by a game with some flash of the laser precision he seemed to possess at times in 2010, followed by a bomb that overshot Hemingway/Roundtree by 20 yards. This was our concern. The more intelligent announcers talked about where his shoulders and toes were at their release, and Borges pressers reiterated the footwork theory.

Then sometime around Purdue-Iowa-Illinois, said all, 2010 Denard worked his way back. I'd like to use this space to test if that was really the case.

The Hennechart you know (screens and Snackycakes have been removed):

2009 2009, All Of It 1 7 4 2 4 4 - - ? 44%
2010 Connecticut 2 9 - - 3 2 - - 2 69%
  Notre Dame 3 17 2 4 1 - 3 2 - 71%
  Michigan State 4 11 1 6 1 - - 2 2 68%
  Iowa 1 8 2 2 2 - 1 - - 64%
  Penn State 3 9 1 4 2 - 1 - - 63%
  Illinois 4 8 1 4 1 3 - - - 60%
  Purdue 2 11 1 3 1 1 1 3 - 68%
  Wisconsin 3 12 1 2 2 2 - - - 71%
2011 W. Michigan - 5 4 3 1 - - - 1 56%
  Notre Dame 6 6 1 5 5 1 1 1 - 50%
  SD State - 8 - 4 2 1 - 1 - 53%
  Minnesota 1 10 1 3 1 - - - - 73%
  Northwestern 4 9 1 7 2 - - - 1 59%
  Michigan State 1 7 3 6 5 - 1 7 1 40%
  Purdue 1 6 - 1 2 1 - 2 - 64%
  Iowa 2 21 2 7 1 - 2 2 - 70%
  Illinois 1 3 1 2 - - - 1 1 67%
  Nebraska 1 9 - 2 2 1 - 1 3 67%
  Ohio State 3 7 2 1 - 2 - - 1 77%

That's lots of numbers. The easy metric to break these down metric is Brian's Downfield Success Rating at the far right. That's Dead-Ons and Catchables divided by all the rest (marginals are excised). But a few years ago, while trying to get a handle on what we had in Forcier, a few users thought to visualize this. I try that now with Denard's career:


I centered in the middle of the marginals to show how good the very goods were and how bad the very bads got. You kind of have to look hard to see it, but there is a regression apparent. Denard seemed to level off in the Big Ten season last year to a good chunk of accurate balls, one or two bad reads, and as many inaccurate as were dead on. For a good part of this year it was that one temptress of a perfectly thrown ball, one to five bad reads, and almost as many balls to Tacopants as the vicinity of his receivers. By Ohio State, on pure downfield success rating, it was just outside the UFR-era hall of fame (on many fewer attempts):

2007 HENNE Purdue W 3 20 N/A 1 1 1 1 2 29 85.20%
2007 HENNE EMU W 4 16 N/A 1 0 1 2 1 25 83.30%
2005 HENNE Ohio St L 6 27 N/A 1 1 4 1 N/A 40 82.50%
2008 SHERIDAN Minn W 2 20 4 3 2 1 0 0 32 78.60%
2005 HENNE MSU W 4 25 N/A 1 3 3 1 N/A 37 78.40%
2011 ROBINSON Ohio St W 3 7 2 1 0 2 0 0 15 76.92%

FTR by this metric, the Michigan State game this year is 3rd all time in the hall of shame, better only than Sheridan in the Badge of Fandom Endurance game vs. Northwestern, and Threet versus Purdue. Sheridan being on both lists was one (happy) fluke between games his coaches hardly let him throw more than a screen for fear of triggering an early duck season. 2011 Denard's is the opposite: one bad game amidst a bunch that range between mediocre and okay. His games aren't in the Junior Henne/Early Forcier range; they are about on par with Big Ten Forcier as a freshman, and he's better than freshman Mallett. This is without the legs.

There was also wide variance in number of throws, partly due to game-planning, but also having a lot to do with Borges leaning somewhat more on the running game when Michigan led. Look at the paucity of passes for Michigan against Purdue and Illinois, versus huge stacks for MSU (look at their pressure metric!) and Iowa. The percentages chart below can adjust for that a bit:


Click it to embiggen. I took out a few more bad defenses to make that one if you're wondering why fewer bars. Also those marks are the rankings by FEI of that opponent's pass defense—the worst pass defense would be at the very bottom, the best at the very top. Take with a huge grain of salt since FEI's weird this year. (No way Iowa and Purdue have the same secondary, nor do I believe either are 40 spots worse than Minnesota). Anyway it shows the metric is at least defense-independent.

This one has the story we've been telling: 2010 was fairly static, while 2011 was a dropoff followed by progression in the new offense (and a stinker in a trash tornado in the middle). Denard also maybe scrambled a bit more at the end of the season (the white bars). Overall you'd almost expect the two years to be flipped, with the hard learning and scrambling a sophomore campaign and the leveling off near the peak of the previous year the work of an upperclassman. If you consider time in the system, it's more like the work of a redshirt freshman followed by a true freshman.

The reads are another thing that fixed over time (Nebraska's weekly BR looks bigger in a small sample). The % of bad robinson-michiganreads this year all told took a rather scary dip from pushing Sr. Henne to Threet-ish:

Player BR/Att DSR
HENNE 2007 6.12% 71%
ROBINSON 2010 6.67% 69%
FORCIER 2009 7.73% 70%
THREET 9.09% 55%
ROBINSON 2011 9.17% 61%
SHERIDAN 10.00% 60%
MALLETT 10.69% 51%
ROBINSON 2009 14.29% 44%

I'm ready to believe this was related to the footwork thing. If the staph infection affected him, it couldn't be more than the beating he took last year blamed for the perceived reduction in Big Ten play. There is evidence of greater pressure—the 7 categorized "PR" in the MSU game is one fewer than Brian gave for all of 2010—and all that.

How much this regression "hurt" Michigan this season can be overstated. Using all plays charted in UFR, Denard averaged 6.93 yards per play, as opposed to the 7.25 yards per play in 2010. That's not about bad defenses; against real opponents Denard's 6.55 YPA is better than his 6.30 in 2010. This is a result of the long passes against Notre Dame (10.09 YPP – which is ridiculous), but if we normalize every play longer than that to a cap of 20 yards, this is what he looks like per passing attempt (2010 schedule futzed with to match comparable games):denard-robinson

2010   2011  
Connecticut 7.48 WMU 6.57
Notre Dame 6.00 Notre Dame 7.77
Bowling Green 10.75 EMU 6.55
Penn State 6.29 SD State 5.88
Massachusetts 7.56 Minnesota 8.84
Indiana 9.00 Northwestern 10.96
Michigan State 6.10 Michigan State 3.17
Purdue 5.91 Purdue 7.14
Iowa 5.79 Iowa 4.31
Illinois 7.95 Illinois 6.64
Wisconsin 6.75 Nebraska 6.73
Ohio State ??? Ohio State 7.35

Including only non-theoretical defenses (No FCS, EMU, BG, Indiana, WMU, NW), and again, counting everything over 20 yards as 20, Denard was getting 6.47 yards per attempt last year, and got 5.96 per passing attempt this year. That's still good. And it's a good bet, with a second year fusing with Mr. Borges, the performance level he got back to from Iowa through Nebraska is conceivable for the bowl game and beyond. If he can somehow sustain what he did against Ohio State he would be inconceivable.


Drew Sharp

December 20th, 2011 at 11:40 AM ^

This game or the 5-3 purdue game? (I think that was the score).  It was probably the NW game because we lost and it wasn't all that close.  Plus, the wind with the rain and snow changing back and forth.  I actually stayed until the end of the 3rd quarter, which is more than most I saw.


December 20th, 2011 at 9:07 AM ^

I don't think that word means what you think it means.

But seriously, great Museday, as always. It doesn't really shed much light on the cause(s) for the regression and subsequent improvement (that would be pretty much impossible to explain beyond speculation/educated guesses), but certainly does put his performance in context and makes me very hopeful for further Denard-Borges fusion cuisine. RIP Tacopants (hopefully).


December 20th, 2011 at 9:29 AM ^

you beat me to it.  But you forgot to include:

/said in my best Indigo Montoya voice...

Although I see what you're saying that the causes for regression and subesquent improvement aren't directly identified, I think we may be underestimating the power of Denard's comfort in the system.  When Hoke was hired, he (Denard) must have said to himself, "Self, going under center more now.  Gonna have to be a pro-set QB."  It seemed to me watching his progressing this year, he may have forced the ball more thinking that's what he needed to do because that's what Borges system was demanding of him.  As he grew more comfortable and as Borges playcalling did a better job of meeting Denard in the middle, his confidence went up and it shows in the numbers. 

I'm sorry to see Tacopants not at least get a farewell luncheon before going, but I hope he finds a good home somewhere else...maybe in ohio.


December 20th, 2011 at 9:18 AM ^

While I don't understand why the passing yards for any given attempt were capped at 20, that had a much greater effect on the 2011 numbers than it did on the 2010 numbers. (I believe that DVOA caps the yards on any given play but FEI does not. So, there is some precedence for this type of data management.)

This would suggest that the 2011 offense produced more big plays (per passing attempt) than the 2010 offense which is the opposite of what one might have predicted given the infatuation with QB OH NOES and the supposed RichRod-generated free TD passes.

In fact, I think Brain did predict exactly the opposite of this.

Given the fact that the supposed regression can only be seen in conference play when the stats are masaged just so and the way Denard played at the end of the season, I would say there was no regression either over the entire conference season or in his play at the end of the season. What regression that did exist was quickly over come in conference play by the strong games at the end. 


December 20th, 2011 at 11:33 AM ^

I tried to give both. For rushing I believe in capping at 20 because after that point if they haven't tackled you the amount of yards you get is basically up to what your field position was. For passing it's not so simple because long passes are part of offenses. Really if you don't count the yardage, Robinson's numbers for 2011 start getting close to Sheridan's in 2008. This is because Sheridan was hardly ever asked to throw the ball more than 5 yards, whereas Denard probably chucked 1 in 5 balls deep this year.

HOWEVA when you're getting into long downfield passes there's a lot of work done by the receiver. Perfectly placed balls on long passes are credited as "DOs" above so the YPA has to be taken in context of the downfield success charts. The yardage metrics can't tell the difference between that and chucking a ball 50 yards to a double-covered Hemingway, and then Hemingway outplaying both DBs who were in better position to make the catch. Or 65 yards to a wide open Jeremy Gallon (though Denard did hit Gallon in stride). There's a very good case to be made that Denard deserves credit for all 70 yards of that 3rd quarter perfectly placed ball to Hemingway while wearing a DT as an ankle bracelet. But then that pass went 25 yards and ND's coverage was responsible for the rest.

Normalizing to 20 yards still makes a long passing play look big in the stats, but de-dupes the performance so that the big plays don't dominate and a guy who chucks one 80-yard jump ball that connects and then three BRX's isn't given a 20.00 YPA.

Blue in Seattle

December 20th, 2011 at 4:14 PM ^

...is that even with a lot of massaging to take out outliers and provide a balanced comparison, the overall result is pretty close between 2010 and 2011.  I think this is also the issue that FEI has when it separates the offense and defensive metrics.  Also special teams can have an affect on the playcalling, which then alters the balance between calling passing plays and calling running plays.  Consider for example the averate field position in 2010 versus 2011, and the improvement in 2011 from beginning to end.  First effect in 2010 is that the defense allowed so many touchdowns that the majority of drives started from the 20. Also different between 2010 and 2011 is the number of turnovers spotting the Michigan offense in favorable position.  Both of these affect the kind of playcalling that is going to happen.

Overall football is a composite game that cannot easily separate offensive and defensive performance unless you have enough data to really compare each kind of play against itself.  This in fact is what it sounds like Borges and Mattison do as coordinators, compare performance of specific plays against the previous performance of that play.  And even for them it's difficult to boil it down to "just Denard" because there are 11 people who have a job to do, and failure of any one job has an effect on the play execution.  For example, is it a passing "success" that Denard underthrows Junior in double coverage when Junior ends up with the ball?  According to Borges and Denard it is for two reasons.  Borges stated that he coached the QB's and WR's to use an underthrow as a way to defeat close coverage.  And Denard knows that Junior can outjump almost anyone.  How do you judge that play against ND as a bad decision?  especially in a result based scoring system?

I've replayed the TD pass to Odoms over and over.  Watching the play from the camera angle on the sideline, it looks amazingly open and the defenders look dazed and confused.  Watching the replay where it's more from Denard's angle of view, I can't believe how he saw that opportunity and put the ball on the money like that.  It seems like Denard held on to the ball so long that both Junior and Odoms had completed their routes.  Somehow Junior sees Odoms start to cross under the safety and Junior cuts back outside dragging a defender with him.  Then Odoms crosses in and finds the spot that open because Junior left it.  The LB that was helping on Junior underneath is completely confused and crashes into the deep coverage safety, and that allows Odoms to sprint into the end zone untouched.

I'm sorry, there is no way that Borges drew that play up that way.  That play happened because Denard had been destroying OSU with his running, and because there is a high comfort level between the WR and Denard.  Also watch Denard's footwork during the delivery compared to before the delivery.  Lots of bouncing and head moving until it's time to throw, then he lasers it into Odoms.  That was a Pro-Style throw in my eyes, aided by the fact that the LB spy covering Denard is NOT blitzing because Denard is NOT Henne.

I believe that Denard improved on his passing skills, reading, footwork, and accuracy, because Al Borges says he has.  I believe the offense will continue improving because all the coaches are returning.  But watching the results of the offensive execution, especially that pass to Odoms, will continue to be inconceivable.

That's why they play the game, and that's why I watch the game. Inconceivable.


December 20th, 2011 at 11:46 AM ^

You cap the pass plays becaused (particulary for Denard the last couple years) most completions over 20 yards involve significant YAC and/or busted coverage (though 30 seems like a more fair cap level perhaps)

DSR% (particularly in the first half of the season) was significantly lower than last season, and Denard made more bad reads. He also threw more interceptions. None of these stats require "massaging".

YPA was lower in 2011 both before and after the capping, though as you note, it's interesting that the caps hurt 2011 more (I'd attribute this mostly to the success of heave'n'pray in the ND game though). Denard did seem to have many fewer opportunities to throw to a wide open guy with lots of YAC space this year though (as you'd expect with the system change).

Denard's passing definitely was worse this year until late in the season (at least in terms of effectiveness per attempt). There were real growing pains - let's not pretend there weren't. That said, the last couple games of the season hopefully suggest that Borges and Denard have found a comfort zone.


December 20th, 2011 at 12:33 PM ^

I never said there weren't growing pains because clearly there were. What I said was that over the course of the conference season Denard was as good or better than he was over the same time period in 2010. He certainly seems as good right now as he ever was in 2010 because while in 2011 he started bad and got good in 2010 he started good and got worse.

Regarding the capping; if Denard throws the ball 30 yards in the air it doesn't make sense to cap that at 20 yards. That pass is harder and has more variance, which is why DSR% is lower, but by not giving it full statistical credit your are artificially limiting the ceiling while still taking full accounting of the basement. In RichRod's offense Denard's long passes were catch and runs. In Borges' offense, Denard's long passes were long passes.

Regarding the selective viewing of stats; Denard completed a pass to a double covered Hemmingway one time. The amount of play that gets makes it seem like he was doing it every week. I don't see why you would strive to remove the WR's contribution to one set of plays while ignoring such a contribution on another set of plays. Or ignoring the contribution of the scheme. If we're going to harumph about Denard throwing to a wide open Gallon 30 yards down the field in the ND game then why aren't we also concerned about Denard hitting a WR on a five yard slant who then runs for 70 yards because the defense was fooled.

What happened happened. If you're going to ignore or purposefully devalue some plays (especially ones that are very important in one scheme  but not another) then you are not really giving an accounting of the stats. You are instead trying to make the stats show want you want them to.

 "YPA was lower in 2011 both before and after the capping,"

Against 'real defenses' the YPA from 2011 was larger.

" against real opponents Denard's 6.55 YPA is better than his 6.30 in 2010."


December 20th, 2011 at 12:55 PM ^

You make a great point. I actually agree with you. After I normalized to 20 yards and posted that part it was past 2 a.m. and I went to bed. When I got up a few hours later to post it I was wondering myself "why did I do that" and had to reconstruct my thought processes behind it. In retrospect 30 yards would have been a better place to cut it off. Unfortunately I'm now at my office on little sleep and can't recalculate. I did finally remember the thought that made me do it: if I don't normalize some, a pass to Tacopants or Nachoshorts to a wide open receiver within the Sheridan zone would count the same as a deep ball that Roundtree didn't adjust to correctly, counts the same as a ball he Braxton's 20 yards too deep. Those were my thoughts around 2:40 a.m. After 2:30 a.m. I probably shouldn't have had any more thoughts.

FTR it was really that ND game that was throwing the stats off. When I removed the bad defenses we were down to 7 or 8 games, and a big chunk of the passes from those games that weren't thrown in a trash tornado were during the 4th quarter vs. Iowa or during the comeback vs. Notre Dame.


December 20th, 2011 at 10:38 AM ^

I'm hopeful, but I think the truth with Denard lies almost exactly in line with these numbers.  His passing is going to be variable because his mechanics are inconsistent, especially when on the move.  There is no doubt to me that Borges got more comfortable with Denard's abilities as the season went on, which led to the better efficiency down the stretch.  But I'm not so sure we're going to see consistently efficient performances with Denard.  There are going to be some missed reads and throws to tacopants.  Regardless, I'll take it because of dilithium and all. 

One Inch Woody…

December 20th, 2011 at 12:18 PM ^

Like someone above said, I think it was less Denard being squeezed into Borges' system than Denard squeezing himself into the mindset that he HAS to stay in the pocket and HAS to take shots downfield. He has that attitude that he desparately needs to prove people wrong, so I guess he wanted to prove everyone that he could just stick in the pocket and be a straight pocket passer. The fact that he has found out how to scramble now makes him that much more lethal of a passer because now people have to account for him running out of the pocket.

The thing that frustrated me the most about passing this season, though, was the incompetence of the WR's throughout the season. Our friend Dough Karsch gave a little sideline report of what was going on during the Nebraska game and there was that deep bomb to Roundtree at first that Denard was trying to explain to Roundtree that he needs to turn and look at his outside shoulder because the DB was inside and to the left. I truly believe that there was a method to Denard's dragon-unleashing. A lot of those passes to Hemingway during the ND game - Hemingway was NOT bracketed. He was double covered over the top. If Denard had thrown a perfectly placed ball to hit Hemingway in stride, it would have been picked off.

Yes, you read that right, with 2 DB's running Hemingway's route for him, if the ball was thrown in front of Hemingway, it WOULD have been picked off. Denard knowing Hemingway's exceptional leaping ability, plus recognizing the lack of a bracket coverage led to his placement of the ball a little short so that he could actually, you know, catch it. 

Another instance of Denard's receivers not doing what they're supposed to was during the Illinois game, I believe, where Denard attempted to hit Odoms deep and just overthrew him. The problem with that route was that Odoms had beaten his man but then attempted to abort into a deep comeback route even though he was open over the top. I'm surprised Odoms even got near the ball despite aborting. Finally, we all know about Denard throwing the pick-six at the end of the MSU game because V. Smith didn't run the right route.

Little things like these might have put us at 12-0.. you never know.

Enjoy Life

December 20th, 2011 at 12:23 PM ^

What about the INTs?? In 2010 the Int% (ints per pass attempts) was 3.8% and it skyrocketed to 5.8% this year. On a game by game basis, there does not seem to be much of a trend as it bounced all over the place.


December 20th, 2011 at 1:36 PM ^

for hanging in there and not getting frustrated and distracted during a big dose of "on the job" training this year.  Change is not easy for many to handle and almost impossible for some people.  Add to that 100,000+ critics in the stands while you are working through everything and it would have been easy for Denard to crumble and collapse after the trash tornado game.  Instead, he proved he is mentally tough and worked it out...his mechanics, a new system/playbook, a new OC, and huge expectations based on 2010.  Oh ya, add a "boo boo" to the mix, too.  Impressive growth once he pushed through the problems with a peak right when we needed it the most...against ohio.  If he does maintain that peak, you're right...inconceivable.