Beating The Shotgun Horse Comment Count

Brian

MGoBlog: where no sleeping dog is left to lie, and no dead horse is to remain unbeaten.  -Blue in South Bend

beating-a-dead-horsedenard-shotgun

left via

In re: shotgun + Denard + site obsession with Denard in shotgunFootball Study Hall put up a post with interception rates that highlights one of the many problems Michigan had turning yards into points last year: Denard's interception rate. Amongst a sample of 100 D-I quarterbacks* he finishes 84th. The only BCS quarterbacks to do worse were Garrett Gilbert, Stephen Garcia, Jeremiah Masoli, Steven Threet, BJ Daniels, and Jacory Harris. This is not good company. Harris and Garcia are 1-2 on this list

THE ZESTY INTERCEPTION WATCH.

1. Jacory Harris. The nation's leader in zesty interceptions won't let being benched stop him. If it gets too bad with new boss Al Golden, he'll just go throw 'em in the street if he has to, because swag like Jacory's never sleeps, and when it does it lands wherever it wants.

2. Stephen Garcia. With confidence. With verve. With GARCIA.

…and the omission of BJ Daniels, who either throws an 87 yard touchdown or three interceptions every play, must have been an oversight thanks to South Florida's ability to fade into the background.

Denard's interceptions weren't zesty. They were like—and I say this in all seriousness—watching the cutest puppy in the world fly headlong into another puppy's head, killing both. The defense was like watching the puppy blood run into the gutter. This is the most precise analogy ever made. Also the field goal kicking was like watching the deceased puppies reanimate just so they could poop all over everything. The Rich Rodriguez era: defined.

SO, ANYWAY

Right. So forwards into the endless and admittedly pretty pointless discussion about the best thing to do for the team the next couple years when they have a 5'11" dreadlocked bolt of lightning at quarterback. My position is blindingly clear: Shotgun Today, Shotgun Tomorrow, Shotgun Forever. For the next two years, at least.

Objections raised from the comments largely revolve around the idea that last year's turnover and redzone performances were flukes that should be expected to magically repair themselves. An example:

I think its a pretty big reach to say there's any "evidence" to suggest that the offense will revert to the mean.  College Football  red zone offenses are not random occurrences within a normal population.  Oregon and Auburn weren't so good in the red zone because they got randomly lucky.  Michigan wasn't terrible because we weren't randomly unlucky.

The offense was terrible in the red zone because:

1) Nobody could make a FG longer than 25 yards (this isn't something that will revert until someone can kick the ball)

2) Our offense simply didn't work as well in the red zone (I don't know why---playcalling, B1G defenses, nerves, but it isn't something that happened because of random chance)

There is no guaranteed regression to the mean in nonrandom circumstances, like football.  Michigan was terrible in the red zone because being terrible in the red zone WAS the mean for Michigan in 2010.

You hope #1 will be solved by the addition of Matt Wile. We are all gunshy about this but highly rated kickers—which Wile was by the end of the year—usually do well. That actually turns out to be irrelevant, about which see this long footnote**. The redzone issues come down to two things: turnovers, about which see above, and giving the ball back on downs.

Michigan did the latter four times last year, all of them late in already-decided games (one against Wisconsin and OSU, two against Mississippi State). They missed one field goal. They failed to score eleven more times because they straight-up turned the ball over.

As far as #2, the whole reason people do these study things and use stats is to have something to argue against people who use the word "simply" as their conversational gambit. Oh, it's simple to you, is it? Well, fine then. I guess you and your galaxy-spanning intellect win. It is possible that NFL football is so different than college football that studies do not cross over, but it is extremely unlikely, and that FO study showed really good redzone teams one year are almost precisely average the next.

In Michigan's case they should expect more than randomness to work in their favor. The common thread of Rich Rodriguez's tenure at Michigan was young or terrible quarterbacks. Three years of Threet/Sheridan, Forcier/Denard, Denard/Forcier should see you give away turnovers like they're candy. There are no upperclassmen on that list except the walk-on; there's only a few confused snaps from a hopelessly raw Denard preventing that list from having any sophomore starters.

The spread 'n' shred in general and Rodriguez in particular haven't shown they are turnover-prone. On the contrary, being able to run 70% of the time and have a good offense should cut down on turnovers since passes are inherently more risky.

From Maize 'n' Brew:

And that is what this comes down to. Common sense. Your eyes. If your eyes are telling you that you're watching a turd of a football game, well... you are. If your reaction to the Wisconsin Michigan game was that Michigan just got completely curb stomped by Wisconsin in the first half, mounted a minor comeback when Wisconsin took a third quarter nap, and then still got blown out by 20 points at home, well... that's what you saw. Perhaps the stats tell a different story. Maybe. But while the stats say that Michigan ran up an astounding 442 yards against Wisconsin they don't relate what actually happened at the game.

I try to back up my opinions with statistical evidence because the use of tools is the thing that separates bloggers and chimpanzees from other primates like newspaper columnists and sports talk radio hosts not on WTKA.

If you want to go on your gut, I can do that too: Michigan has a 5'11"-ish quarterback who ran for 1700 yards last year and an offensive line that's now 100% recruited to zone block all day. They don't really have a promising running back. I feel, like, not good, man, about Michigan in the I-form.

Or I could say that "common sense" suggests that Wisconsin was not trying to let Michigan score in the third quarter and that the overall results should be taken in appropriate context, but then we're back to feelings, man.

What Is The Core?

I just don't see how the spread offense is responsible for turnovers except insofar as it puts an erratic Denard Robinson on the field instead of a finely-polished artillery piece, and who wants to fix Michigan's issues by replacing Denard Robinson?

/crickets

/Munn Ice Arena

/people stapling each other's hands to their sides just in case they have a hand-raising seizure

Not having Denard drop back from center does not make his throwing mechanics worse. If anything it allows him to ignore a complicated facet of football—NFL coaches are constantly bitching that college quarterbacks no longer know how to execute a five-step drop—and focus on throwing it to the guy who's really open because you're not running the ball.

Meanwhile, the run game was kind of good last year despite having the worst set of tailbacks at Michigan since at least that year BJ Askew got half the carries. This is directly attributable to putting Robinson in a position to run, something an I-form doesn't.

There are quarterback draws and waggle plays, yes. Opponents will be all over them because those are constraint plays—not your base. Smart Football on how you build an offense:

The idea is that you have certain plays that always work on the whiteboard against the defense you hope to see — the pass play that always works against Cover 3, the run play that works against the 4-3 under without the linebackers cheating inside. Yes, it is what works on paper. But we don’t live in a perfect world: the “constraint” plays are designed to make sure you live in one that is as close as possible to the world you want, the world on the whiteboard.

Constraint plays thus work on defenders who cheat. For example, the safety might get tired of watching you break big runs up the middle, so he begins to cheat up. Now you call play-action and make him pay for his impatience. The outside linebackers cheat in for the same reason; to stop the run. Now you throw the bubble screen, run the bootleg passes to the flat, and make them pay for their impatience. Now the defensive ends begin rushing hard upfield; you trap, draw, and screen them to make them pay for getting out of position. If that defensive end played honest your tackle could block him; if he flies upfield he cannot. Constraint plays make them get back to basics. Once they get back to playing honest football, you go back to the whiteboard and beat them with your bread and butter.

The argument here is about the core of the offense: in the I-form that's Denard dropping back to pass or handing off to someone else. In the shotgun it's the zone running game. As the core of the offense you can't remove Denard from the game. You cheat and then there's a guy wide open. While Denard's legs are a terrifying constraint, Michigan has to force the opponent to cheat to use them.

I'll believe these tailbacks and this offensive line and this almost total lack of fullback and tight end can do that running power up the middle when I see it. If they can't you've just taken the most dangerous weapon in college football*** out of the game. You shouldn't do that. It's common sense.

------------------------------------------------------------------

*[I'm not sure why there were 100 quarterbacks instead of approximately 120 + a few injury replacements, so keep that in mind.]

**[Long aside on Michigan's historically awful field goal kicking goes here. Nonnair posted a diary asserting that the lack of field goal kicking was not a factor in red zone efficiency because Michigan actually scored more points than they could have if they kicked it:

The other seven fourth-down attempts I am dividing into two groups: (1) FG is the likeliest option and only a riverboat gambling coach or a team without a FG kicker would go for it, and (2) FG is only a possible option, either because it'd be very long, or because there was only 1 yard to gain for a first down so going for it is a viable option.

Bottom line?  If we had tried FGs on all seven of those drives last year, even if we had Adam Vinatieri circa 2002 and he went 7-for-7, the most UM could have scored was 21 points.

As it was? UM got 27 points out of those drives. Six more points.

This is only one half of the equation, though, because Michigan did attempt a bunch of field goals and they went like this:

awful-kickin

All that red in the Michigan zone is value earned by the offense that was lost by the kicker on obvious kicking opportunities. So on the field goals Michigan tried last year, we threw away 16 points, versus the six this study shows M getting back by being forced to do a statistically correct thing that teams don't usually do because their fans don't trust statistics.

Misopogon threw this behind a jump on Sunday.

Nonnair turns out to be right: the field goal kicking did not have much of an impact on the red zone efficiency because Michigan's misses are all clustered just outside. However, the statistically correct behavior Michigan engaged in also had no effect. Six of the seven attempts were outside the red zone and the one that was inside it, a fourth and one from the Penn State 13, was converted and led to a field goal anyway.

So we're down to just the massive turnovers. I hope this section has highlighted how goofy red zone efficiency is.]

***[Other than Charles Robinson.]

Comments

jmblue

July 7th, 2011 at 6:00 PM ^

That's why you can't fall behind by 21 points in the first place.  You don't fall behind by that many if your offense is getting the job done.  In all those games, we went through long stretches in the 2nd and 3rd quarters in which we were dominated on both sides of the ball.  Only after our opponent built a seemingly-insurmountable lead did our offense finally show up.  We needed it to show up much sooner. 

The problem with your line of reasoning ("30 points is enough") is that it implies that we would have scored exactly that many if the game had been close.  There is no particular reason to believe that.  Again, in all these games we were shut down offensively for long stretches, and only came to life in the fourth quarter, when the game was basically over.  It happened too many times last year for it to be a fluke. 

gbdub

July 7th, 2011 at 7:16 PM ^

The thing is, "getting down by 21 points" doesn't take very long when your defense gives up a TD on every possession, which is effectively what it did for long stretches against good opponents. This also resulted in a much smaller number of possessions for the offense, since good opponents could (and did) chew clock at will to keep the ball away from Denard.

Going 2-3 possessions without a TD is not particularly uncommon when a good offense faces a good defense, and is hardly fatal when your defense has a pulse. Even the best offenses don't score on every possession against average D, so it's absurd to expect the Michigan offense to do so against great D. And yet whenever they didn't in 2010, they found themselves down 2-3 scores very quickly.

So keep in mind when you say "they shouldn't have gotten down 21 points" what you're really saying is "they should have scored a TD (on an 80 yd. drive) on >50% of their possessions" which is a pretty high standard. One they actually almost met against Wisconsin, despite the lopsided score.

MGoNukeE

July 7th, 2011 at 6:54 PM ^

What's the root cause? If Chunkums is right and the two top-ranked teams in the country did not score a ton of points initially, then the primary reason that Oregon and Auburn did not fall behind by 21 points six times this past season is because they had sound defense and special teams play to mitigate the offense's inability to score touchdowns. Michigan did not have sound defense and special teams play to mitigate the offense's inability to score touchdowns (in fact, it's been repeated over and over again that sound defense and special teams play actually helps an offense score touchdowns). Therefore, either Chunkums is wrong about how Auburn and Oregon's offenses played at the start of games (I'd rather not do the research myself since I trust Chunkums) or the offense is not the primary culprit to blame when discussing why Michigan fell behind by 21 points six times this past season.

The past couple debates on this subject have been over pins and needles in any case, with one side saying Michigan's offense is great but not super great because Michigan didn't score enough points and the other side saying Michigan's offense is super great but only looked great because of the poor defense and special teams play. I'm probably responding to a straw man anyway, since you may not even think the primary culprit to blame is the offense. Since "primary", "significant", "kinda", etc are all arbitrary abstractions that could mean more or less than one another, I'll sit this one out until someone says "The offense only has an overall rating of 8.76/10," because I will pounce all over that with "The offense is definitely a 9.456/10 with a repeating decimal. BOOYAH!"

chunkums

July 7th, 2011 at 7:34 PM ^

RIght, but they both fell behind several times.  We're talking about offense right?  It's not our offense's fault that our defense could do nothing.  Also, Auburn did fall back 28-0 at one point.  The difference is that their defense actually started getting stops whereas ours was.  Looking at other option teams (Oregon, Auburn, Georgia Tech) they tend to pick it up drastically as the game develops.  Unfortunately for us, we didn't have a defense to stop anyone while the offense was adjusting.

el segundo

July 8th, 2011 at 10:58 AM ^

I agree wholeheartedly with much of what you've written, especially your point that, in football, certain statistical measurements are problematic because not all of the events they encompass are truly comparable.

One of the big problems with the statistical analysis of football and other sports is that statistical analysis is premised on the idea that the events being counted and compared are all fundamentally similar -- so similar that they are commensurable, that is, that they can be measured by the same standard.  In football, differences among plays with respect to down, distance, and field position (not to mention game score and time in the game) make it much more difficult to maintain the assumption of commensurability, at least with certain measurements like total yards gained on offense.

For example, there is a big difference between a play run on 3rd and 2 at the opponent's 38 and a play run on 1st and 10 from your own 20 because the risks and benefits associated with each play are different.  In the first situtation, gaining three yards would be a great success, and it would be rational for a team to choose a play designed to reliably gain short yardage with an absolutely minimal risk of a turnover.  In the second situation, a three-yard gain would be, at best, a mediocre result.  The objectives informing the play call and the play execution are substantially different and cannot be measured by the same metric of success. In other words, three yards in the first situation has a higher value than three yards in the second situation.  It's misleading to count each of those different three-yard gains as equally valuable.

This is why, in the passage Brian quoted, Maize-n-Brew was right to devalue the yardage gained by Michigan's offense against Wisconsin, once the game was out of hand.  More importantly, it is also why Brian is wrong to suggest that Maize-n-Brew (or anyone else) lacks objectivity when making a statement like this.  

Too often on this blog, Brian and many of his supporters equate objectivity with quantifiability.  Their view seems to be that statements are objective only when they translate perfectly to quantifiable terms.  More precisely, their view seems to be that statements are objective only when they can be translated into existing modes of statistical analysis.  In the context of football, the problem with this view is that the existing modes of statistical analysis are often too crude to be truly informative.  Moreover, there are perfectly objective judgments that can't be readily expressed in the current terms of statistical analysis.  It is objective to say that it's more valuable to gain three yards on a 3rd and 2 on the fringe of field-goal range and it is to gain three yards on 1st and 10 at your own 20.  But, as far as I know, there is not yet a statistical metric that permits the quantification of this difference in value.

I hope more people on this board can recognize that a healthy skepticism about statistics or an awarness of the limits of statistics is not the same thing as placing feelings above objective facts.

Farnn

July 7th, 2011 at 2:48 PM ^

I got the feeling that a lot of our redzone turnovers were due to a young QB trying too hard because: A, he felt the offense had to carry the team since the defense wasn't holding up its end of the deal, and B, he knew that kicking a field goal really wasn't an option.  When players try to force things bad stuff happens, Forcier in 09 is another example of this, as is Brett Farve and his most interceptions for a QB in NFL history.

CRex

July 7th, 2011 at 2:49 PM ^

What I fail to get is this irrational hatred of the I-Form and the "OMG MORE SHOTGUN" demands.  The shotgun offense and liberal usage of designed runs is without a doubt where Denard is the strongest.  I don't think anyone is dumb to argue against that.  However people like Rawls and the other manball backs are strongest in the I when they get a nice downhill run going.  WRs excel when the QB plants his feet and throws a nice fade route or something (see: Edwards, Manningham, Arrington, etc).  Same with the TEs.

Last year we went full on shotgun and ran Denard right into the ground.  Tate had to finish games and of course we burned Devin's redshirt as well.  Plus we shit the bed against ranked teams.  At the end of the year we still lacked a clear #1 WR, a clear dominate RB and of course the TEs weren't a huge part of the RR offense.  So we had Denard, a good offensive line and not much else.  

When some says MANBALL, I-Form or things like that it doesn't mean we're screaming for the return of 3 yards and a cloud of dust (although that did work out pretty well for us), it just means we want to have offensive weapons beside Denard.

colin

July 7th, 2011 at 3:32 PM ^

Is there really much evidence that the team HAS weapons anywhere near as good as Denard?  Certainly not at running back.  If we're running the ball 30 times a game, it better be Denard getting 15-20 of those carries.  Brian's not kidding when he points to the relative mediocrity of the backs.  And Magnus would very likely agree (while making a loves-the-Cox joke).

I would on the other hand make the argument that we do have the receivers for a solid if not spectacular Borgesian version of the Moeller-Lloyd offense that would seem to be down the road.  Problem is, Denard really doesn't have the skillset to make that work (again, ask Magnus and check out those INT numbers, recall how many offers he got at QB, etc.). So my question is if you're going to devote your energy to changing offenses, can we at least give it to the guy who, of our QBs, is incredibly well suited for it?  Whatever percentage of the time Denard can't be a run threat, for either down and distance or health purposes, can't Devin take those snaps?  

JeepinBen

July 7th, 2011 at 3:45 PM ^

Rather than a Junior who almost won the heisman after throwing for more yards as a sophomore than Michigan's leading career passer threw for as a sophomore?



Sophomore Denard:

2570 yards: http://espn.go.com/college-football/player/_/id/480237/denard-robinson

Sophomore Henne:

2526 yards: http://en.wikipedia.org/wiki/Chad_Henne#Statistics

colin

July 7th, 2011 at 3:54 PM ^

Like I said above, it's fairly clear that Denard has serious issues with passing and has no pedigree.  When Junior Henne turned out to be RoboHenne, there wasn't a whole lot of surprise.  That's what he was expected to be by those who were recruiting him and those who rated him as a recruit.

Denard has the opposite pedigree.  D1 coaches saw a WR or cornerback because they noted his lack of suitability in the passing game.  He is not a dropback/complex read QB.  He's a QB in an offense that ran the ball 65% of the time.  He was a triple option QB in HS.  All that said means that his passing deficiencies were a function of his physical attributes, not his lack of experience (though that surely was a contributing factor).  He is a game changer as a runner specifically.  But his passing skills are very probably matched or exceeded by Gardner, who ran a passing spread in HS and was well regarded for his traditional QB skills.

JeepinBen

July 7th, 2011 at 4:03 PM ^

OTOH, We saw Denard "struggle" in his first year getting meaningful snaps as a QB (I don't really count his 40 odd passes as a freshman). 

What's to say Devin won't struggle as well in his first real meaningful stints? 

I'd much rather see the returning B1G POY on the field as much as possible and not voluntarily take him out. Yes, he was injured last year often, but that doesn't mean he will be again.

/also, for the record, I'm not collapsing/neg voting your posts. I'm all for honest debate

colin

July 7th, 2011 at 4:19 PM ^

TY.  I'm definitely not "trolling" or whatever.

And I agree with you outside of the fact that it seems clear to me that Denard will either be forced off the field or forced into formations/plays that negate the things that make him the incredible game changer that he is. To me that says there are some (pulling figure from ass) 20% of the plays where Denard will be basically as good or worse an option as Devin anyway.  If that's the case, I think we should make it a multi-package thing. 



The only way the under-center stuff turns out to be excellent is if the line emerges as a dominant force (or a dominant RB emerges, but that seems far less likely) so Denard doesn't really have to do much 5-step/7-step stuff except off play action.  Denard as a pure passer is only ever going to be just okay.  And I think Devin can probably do that as well or better and he's also the one best suited to transition this offense from Spread and Shred to Multiple Pro Sets.

If you don't agree Devin has caught up in pure passing, you're not going to be in favor of my argument.  And if that isn't true, I wouldn't want the offense run in the way I'm suggesting either.  But that seems to be the consensus of the people who know football best.  And it also seems like as fans we'll tend to overrate Denard's skills as a passer.  To me it's all just following the logic string.  If you dispute the assumptions, then sure the conclusion won't follow.

JeepinBen

July 7th, 2011 at 4:21 PM ^

I don't think Devin >= Denard as a passer, so I'm not going to agree with you there. 

I think that the extra year of actually playing, and completing passes for 2500+ yards is going to give Denard the advantage over Devin in most things, if not everything. 

I definately agree with you on paper... if Devin is 5 star awesome Multi formation QB, he's going to push for playing time and see the field. But We don't play the games on paper, so that's where I disagree

UMaD

July 7th, 2011 at 5:02 PM ^

I think assuming a major difference between DG & DR in terms of pro-style effectiveness is a serious(ly flawed) assumption.  You acknowledge the point, but saying "consensus of people who know football best" is a stretch.

The player DG resembles most is Vince Young and Texas changed their offense to a shotgun-spread-based attack to fit his skills.  DG also has no experience running that offense - or taking snaps under center.  His presumed superiority is based on height and general reports out of practice that he's somewhere between Denard and Tate in terms of accuracy - and both of those guys were INT prone for one reason or another.

We can all project and assume, but if we're playing that game, I'd rather just assume Denard is going to make a leap in throwing accurately.

colin

July 7th, 2011 at 6:39 PM ^

I don't think VY/Pryor is a good model for Devin.  He's definitely a passer first. He's more of a Daryll Clark (that's the comparable Rivals cites). The spread he ran in Inkster was a pass-first spread.  As for projection, it's necessary.  We're talking about some kind of new offense and guessing at what will work and what won't.

Also, it's more than simply Devin's height that got him evaluated as a better passer coming out of HS than Denard.  Devin was the best QB at his Elite 11 camp per Rivals.  In a down year for QBs, sure, but Denard wasn't even rated as a QB by Rivals.  It really shouldn't be that controversial a step to say that Devin very well could be a better passer given that Denard's reps as a passer are so few.  In both HS and college, his passing game was largely dictated by the threat of the run.  RR had him running 3-step and 4 verts, both of which are appealing as complements to the run game in part because they don't take much practice time to get down, at least compared to what Borges eventually wants to do.

Borges-type passing is exactly what everyone figured Denard would be ill-suited for if he played QB at all, which (for instance) Urban Meyer didn't even see him as.  He still has basically zero experience which is not true of Devin.  The essence of what I want is just to keep Denard in his comfort zone and make the transition to the future offense through the offensive line and Devin.  To me, that's a good compromise between investment in the future, showing consideration for Denard's health and trying to win now.  

UMaD

July 7th, 2011 at 9:04 PM ^

Clark seemed more compact and powerfully built.  More of a RB than a WR, if you will.  To me, DG is a VY clone, right down to the awkward throwing motion and the long-striding (deceptive) speed.

"It really shouldn't be that controversial a step to say that Devin very well could be a better passer given that Denard's reps as a passer are so few." 

Neither of these guys has any sort of high school offense that would prepare them for a Borges style college system.  The Inkster highlights from Tim's FNL videos showed a lot of QB scrambles.  They may not have handed off much, but there was a good bit of running going on.  FWIW - A quick google of high school stats (via Max Preps) indicates that Denard threw more and ran less than Devin.  That seems to indicate the opposite of what you said about passer reps.

"What everyone figured..."  isn't a convincing argument here.  Most people figured Denard was going to be moved to DB.

I think if you're going to make the case that Devin has more experience (as a RS freshman) than Denard has as a Junior and returing conference POY based on high school, the burden of proof lies on you.  It comes off as rather ridiculous to me, but I know that most people seem to agree with you based on preconceptions.

If Devin proves he's significantly better at a certain package of plays then I have no problem rotating QBs for that purpose - it just seems that's extremely premature speculation.  Maybe

 

colin

July 8th, 2011 at 2:10 AM ^

hmmm.  to your point:

http://www.youtube.com/watch?v=LoZiQtlTauc

it's at least in the same neighborhood as Devin, though who knows what got left out.  i specifically remember articles about how they ran a lot of triple option.  not sure how to resolve these two bits.  

also i have to say that i was really discouraged by the spring game, though i suppose there's no reason to really like Devin's performance either.

 

jmblue

July 7th, 2011 at 5:22 PM ^

If we're running the ball 30 times a game, it better be Denard getting 15-20 of those carries.

I think it's been established that his frame probably can't handle that heavy a workload.

As for the relative mediocrity of the backs, how can we say that for certain? Shaw looked good when healthy (which was admittedly not often). Hopkins showed flashes in his brief outings. Cox has never really gotten a chance. Smith wasn't anything special, but he was also in his first year back from ACL surgery. I don't think it's out of the question that we can get more out of this group than we did last year.

redhousewolverine

July 7th, 2011 at 4:33 PM ^

I don't think Brian is hating on the I-Form, but just wants us to keep going to what works until we have all the necessary pieces or the coaches feel the team is ready for a change (to I-Form). As you said we have a good offensive line and Denard, who are all more suited/recruited to play more of a shotgun set, so Brian and others think we should play to our offensive strengths. Yes, we all want offensive weapons beside Denard (I think you should had inserted Denard's running ability), but our WR are goodish, just need more consistency and more consistency from Denard passing the ball. We all want a RB to emerge, it just seems the talent isn't there or hasn't fully developed. RR utilized Denard so much because our RBs sucked. Smith would hit huge holes but be to slow to do anything with them. Shaw and Toussaint were hurt. Cox never bothered to learn the playbook. And Hopkins really wasn't a dynamic force and fumbled more than our defense could bear. When you scream MANBALL, I go sure, we need the talent first. Just because we got Rawls doesn't mean we are going to be able to run MANBALL. We don't know how good he is. Each year Jackson sells us (well some of us) on the RBs with his hyperboles like an selling an eskimo snow.

I think some people's frustration with the I-Form goes to subpar Offensive coordinators for Michigan in the past and maybe spread offenses winning national championships etc. Three yards and a cloud of dust worked well. I think the hope was that the spread would take us from being a good national power to a great or dominant national power.

Also, WR don't need to have Prostyle QBs to succeed, which I assume is your point about planting feet and throwing fades. Spread QBs can do it too. See: Michael Crabtree, Jeremy Maclin, Percy Harvin, Ryan Broles, Gilyard (guy from Cincinnatti with Kelly), etc. They have been just as good, nearly as good, or better than your list. Here is a nice article from Tim at Varsity Blue on it: http://www.umvarsityblue.com/2009/05/the-spread-offense-wide-receivers-and-the-nfl/  Also, some RBs who are pretty good out of the spread: Darren Mccfadden, Lamichael James, Steve Slaton, Noel Devine, Ian Johnson (I think he is the guy from Boise) etc. Chris Johnson, the best RB in NFL, was in a spread like offense and he seems to be doing okay.

Yes, RR's spread didn't like TEs, but Koger also liked dropping the ball and Webb wasn't exactly a dynamic playmaker from the TE position. Shitting the bed against ranked teams dealt more with defense than offense. Players can develop from spread also; see above. Moral is better recruiting and development.

 

imafreak1

July 7th, 2011 at 2:53 PM ^

I understand the point that Cook and Fremeau (in HTTV) are making regarding the Wisconsin game. What I don't understand is why both Brians picked that game to focus on when making their points regarding these new statistical methods. That game was 24-0 before Michigan started scoring. Michigan went on to be stomped. These factors make the argument unnecessarily more difficult.

As to the idea that Wisconsin wanted Michigan to score in the 3rd quarter (and Cook's contention earlier that Michigan's last TD wasn't a junk TD)--this is an offensive over simplification. When one team is up 24-0 or clearly going to win in the closing moments, it is not so hard to believe that maybe they lose some intensity. I'm not saying that happened, I'm just saying no one thinks Wisconsin wanted Michigan to score so suggesting such is pointless.

I very much enjoyed the Fremeau article in HTTV and read it within minutes of recieving my copy. However, I was intruiged by his contention that with certain caveats Michigans offensive day against Indiana was better than Wisconsins against Indiana. He makes the point that Michigan had much worse average field position making this possible. I contend this does not pass the eyeball test (something every good scientist should subject his conclusions to.) Wisconsin had their way with Indiana. Their offense greatly outperformed Michigans with regards to score and points. If Michigans offense looks better it is a function of starting field position which is offense independent. These stats are designed to factor out special teams so Wisconsin should not be punished for them either.

In this case, the advanced statistics have lied.

Not coincidentally, I think bad starting field position was a factor that allowed Michigan to perform so well, maybe undeservedly, in terms of FEI ranking for various reasons.

AAB

July 7th, 2011 at 3:01 PM ^

is a good distillation of what is making Brian so mad (even if he's responding to a strawman, which I think he is).  

The "eyeball test" needs to be forever abolished as a retrospective tool of analysis.  Our eyes are terrible, terrible, terrible judges of what happened. 

And you're getting the causation entirely backwards on the field position issue.  It's not that field position makes Michigan's FEI numbers look really good.  It's that field position made Michigan's offense look far worse by traditional numbers (especially points scored, which is as bad a metric of offense as ERA is of pitchers) than it actually was.  As you note in your own post, an offense has no control over field position.  As a result, if two offenses drive 60 yards, it's insane to reward one offense because its drive started on its own 40 and punish another offense because its drive started on its own 20.  

AAB

July 7th, 2011 at 3:10 PM ^

prospectively, that's not even close to true.  Again, Brian is not arguing that Michigan "really" scored 40 points against Wisconsin. He's arguing that Michigan's offensive performance last season according to advanced metrics means that its' offense will be awesome this year if Michigan does not change it.

He is using past data as a way to determine what to do in the future. The scoreboard usually doesn't tell you a damn thing about any of that. Worse, it often gives you a completely misleading impression. If a team wins a game 28-14 on 2 interception returns for touchdowns but gets outgained 550-260, you'd sure as shit want to be the team that gained the 550 yards going forward for the rest of the season.

The entire point of advanced statistics is to isolate factors that units have control over, so we can see what they do well and what they don't do well, and whether they're good at what they do or whether they're bad at what they do, which is exactly what we care about moving forward.

Belisarius

July 7th, 2011 at 4:26 PM ^

What the use of raw data and statistics fails to take into account here is the price Denard payed for carrying the entire team on his speedy little legs. If you rely on Denard for your entire, meaningful running game, which provides the bulk of the offense (as passing worked off fear of Denard's run) you're A.) Not sympathetic to the guy's pain B.) Assuming that he will not miss any more time, especially whole games based on more serious injuries and C.) Assuming that whatever time he misses can be negated by the use of a backup QB as good or better than Tate.

What RR was doing was not a sustainable offense. One of the measures of what happened was not merely playing better teams, but teams that understood how to beat us based off of our arsenal (1: Denard 2: Passing based off of threat of Denard 3: Desperation).

A theory often discussed is that the threat of Denard running might become more potent if you attempt it less and cultivate other methods of attack. If you combine a passing game with a viable running game not based on Denard, you open up great opportunities for No. 16 to race through while exposing him to less constant punishment.

I want to give Borges every opportunity to experiment, to save our QB as much as anything else.

El Jeffe

July 7th, 2011 at 3:21 PM ^

Is that why you have a scoreboard as your avatar? Because all that matters is the scoreboard? So what does the scoreboard tell you in terms of whether we should run more shotgun or more I-form in the coming year?

Exactly. You'll be able to get back to us at the end of this season and tell us. That is the definition of post hoc reasoning, and it is baaaaaaaaad.

JBE

July 7th, 2011 at 3:33 PM ^

Statistics, depending on how they're used, can rationalize any perspective, while the eyeball can be a nice tool to challenge that rationalization.  This can work the other way around as well.  

For instance, last year the FEI states Michigan was the #2 offense, but my eyeballs saw serious and various issues on that side of the ball in a number of games, and after watching a ton of other offenses last year I would put Michigan top ten instead of top five.  Or, in the reverse, my eyes saw a offense that struggled significantly, but the FEI claims that they were pretty efficient, so my eyes are, to some degree, untrustworthy, and Michigan, according to the numbers, performed better than my eyes led me to believe.  

Naturally, the "truth" is probably somewhere in between.

But to discount the eyeball test and replace that experience wholly with stats, or vice versa, is shortsighted.  Eyes lie, and numbers lie, and to trust wholeheartedly in one or the other is what Fascism is made of. 

   

gbdub

July 7th, 2011 at 4:18 PM ^

I think the point is that not that a high offensive FEI says "Michigan was a great offense". It says "Michigan was great offensively at the things the offense should be expected to be independently good at". One of the "lies" your eyes tell you is that the ability to get lots of yards without points must be a flaw of the offensive system. FEI indicates that field position and field goal kicking were primary contributors. Also supporting the "offense was good at offense things" is that our red zone TD efficiency was quite high. In other words, I think the primary lie your eyes tell is that inability to score is always the fault of the offense, when in reality defense and special teams can play a large role in the offenses ability to score.

I think the main problem last year that can actually be blamed on the offense was Denard's interception rate. Red zone turnovers completely changed a few games last year.

JBE

July 7th, 2011 at 4:38 PM ^

[The FEI] says "Michigan was great offensively at the things the offense should be expected to be independently good at".

Then my eyes say, "Wait, they were not great offensively at the things the offense should be expected to be independently good at.  They had long periods where they struggled to move the ball and score points."

Then the FEI says, "Yes they were."

And I listen to both because I'm one open minded mutha.

gbdub

July 7th, 2011 at 7:27 PM ^

Then the streakiness is what's lying to you. "They had long periods where they struggled to move the ball and score points." They certainly did. They also had long periods where they moved the ball and scored points at a blistering pace. If the former outweighed the latter, Michigan would have ended up much lower in the offensive FEI ranking.

Your eyeballs are hardwired to recognize patterns. Unfortunately we also tend to overemphasize the significance of observed patterns. The struggling stretches stick in your mind because they resulted in bad losses. The good stretches are less glaring since all of our wins were close.

Mitch Cumstein

July 7th, 2011 at 3:07 PM ^

I think its important to realize that the era of advanced stats came about b/c analysts would watch and study results with current stats and realized conclusions from those stats didn't pass the "eyeball test".  In other words, just b/c something observable can't be quantified by our current system of quantification doesn't mean it isn't important.  As a scientist I think it is important to review results critically that don't pass said "eyeball test" (results that go against scientific intuition).  They might be correct, but definitely warrant a second, closer look.

AAB

July 7th, 2011 at 3:14 PM ^

my eyes tell me that Justin Verlander causes hitters to make weak contact more than Brad Penny does.  Advanced stats tell me my eyes are lying, and Justin Verlander and Brad Penny will have a virtually idential batting average against on balls in play.  My eyes also tell me that the guy who makes a ton of diving catches is a great, but oftentimes he had to make a diving catch because he was too slow to get to the ball in the first place.  And finally, my eyes tell me that David Ortiz and Derek Jeter are clutch, but advanced stats say they perform exactly as well in "clutch" situations as they do in any other situation.  

A lot of advanced stats are useful precisely because they go against what our eyes tell us is true.  

Mitch Cumstein

July 7th, 2011 at 3:38 PM ^

if you are interested in who causes more weak contact Penny or Verlander, you would watch for those specific instances, or in this case look at the stat the measures that specific thing.  If you use this stat and this stat alone to determine who the better pitcher is, that wouldn't pass the "eyeball test" b/c you see Penny get rocked every 5th day and JV dominate, yet your stat told you they were the same.  Thats essentially my point.     I'm not agreeing with the poster that you originally replied to, but I am saying that you should make conclusions off of stats that adequetely reflect the item in question, and assuming that stats are available to conclude complex sports questions is a bold assumption. 

MichFan1997

July 7th, 2011 at 3:38 PM ^

balls hit in play will, over the course of a ton of sample size, fall in for hits at a rate of probably .290 to .310 ish. Jhonny Peralta has a higher career average on balls in play than does Albert Pujols when I last checked a week or two ago. This is because batted balls land randomly.

imafreak1

July 7th, 2011 at 3:36 PM ^

Abolishing commen sense (the eyeball test) as a tool of analysis is completely ridiculous.

If you design an artificial system to analyze or measure something and it spits out results that are completely contradictory with common sense then you've got a problem. Stats are merely a tool. Used incorrectly, they can be just as misleading as anything else.

I don't actually disagree with Cook that much. Sometimes, I am perplexed by his arguments but rarely with his conclusions. I very much doubt that I am the distillation of what he is mad about.

 

colin

July 7th, 2011 at 3:59 PM ^

There are significant questions the eyeball test can answer and some of those questions can't be answered w/ statistical analysis thanks to low sample size.  Tom Tango's Fan Scouting Report on defensive performance in baseball, for instance, is a great tool.

Monk

July 7th, 2011 at 6:00 PM ^

in your scenario.  A lot of teams have bend but don't break (OSU comes to mind as did UM under Carr) and they're more than happy to give some yards in the hope of making the offense execute more plays and maybe turn the ball over or come up short on 3'rd down. 

The team that gets the ball in the endzone is better than the one that get stalled on 4'th down at the opponent's 30.  And here's why, the team that scored converted their 3'rd downs and found a way to score when the defense tightened up.

I think the spread does break down in the red zone (see Oregon last year in the NCG) as things like TE crosses and fade routes are better used to get the 7 pts than the read zone.

chitownblue2

July 7th, 2011 at 2:56 PM ^

On this very site, Tim transcribed the following.

 

Shotgun - used quite a bit. "We're gonna gun more than we've ever gunned." [ed: w00t.] They've run plenty of shotgun in the past, but this will be more than that. Two-tights is part of the offense, but you'll see as much shotgun as you'd see from any NFL team, and a lot of the tight end in motion.

Haven't done a whole lot of stuff this year that he's never done. "We have more QB runs, but that aside, our offense is our offense, and we're going to gear what we do to what he's capable of doing."

I mean - it looks like you even got excited that Borges...said he would run more shotgun than he normally did which was, per him, "quite a bit".

El Jeffe

July 7th, 2011 at 3:25 PM ^

FWIW, I don't think Brian is arguing with Borges. I think he is arguing with fictitious MANBALL advocates and non-fictitious people like jg2112.

I think we're seeing a bit of Brian's OCD tendencies acting up here. In fairness, though, he did title the post something something dead horse something something.