somehow we're only 124th
Air Force's ability to consistently get the edge on Michigan's defense was the most frustrating thing about Saturday's game, and many theories have been proffered as to what was supposed to be happening, why it wasn't, and why we will or will not die on the rest of the schedule.
I'm of the opinion that Michigan's scheme was predictable and that as soon as Air Force started blocking Kovacs they were out of ideas.
Here's Air Force's first play of the second half. Denard Robinson's just gone 58 yards to put M up 21-10 and a poor decision by a Falcon player to fall on a squibbed kickoff sets the Falcons up on their own 12 yard line. Michigan has just sat in the locker room for 20 minutes getting coached up; Air Force comes out and runs the same triple option they've been running all game.
It does not go well.
Okay: I called this "near half-flex" for Air Force. Michigan is in their 4-4 under, which I know is actually shifted towards the nominal strength of the formation and so is technically an over. Michigan aligns to field, not strength—so they would flip their formation if it was on the other hash.
Against Air Force, Michigan brought Gordon down into the box and made their formation basically symmetrical. Mattison:
Jake and Thomas were the exact same position in our scheme. A lot of people play the same scheme.
Kovacs is playing center field. Earlier in the game, he was not getting blocked and doing Kovacs things. Like this:
Air Force was all like Eff that to the A and started blocking him. That took out Michigan's edge defender and opened up the corner. Michigan didn't really adjust.
Air Force's "triple" option" wasn't really that. They occasionally ran the dive to keep the defense honest but when they did that the QB just turned around and give it, no read. Here they're running the option with the token dive fake. Already in the above frame, bad things are happening.
Will Campbell(1) is tackling an Air Force lineman who's trying to get out on Bolden. He'll succeed at this, allowing Bolden to flow freely for the rest of the play, but he'll pick up a second defensive holding call doing so. On the edge, Gordon(2) is the optioned guy. Michigan is playing him to pitch like they have been all game. Kovacs(3), is the destination of the flexback.
In a second or two, Michigan is going to eat cut blocks:
Thanks for participating, Clark and Morgan, but you've been elimidated. Try again next play. Meanwhile, downfield…
…the ref is ANGAR at Campbell and Jordan Kovacs is decidedly not coming up to stop the pitch.
Why is Kovacs taking that angle? Why is he not attacking the run? That's an eligible receiver he is in man coverage on. He's got no one behind him, and there are two other receivers going vertical. He has to respect this guy as a receiver, or he could give up an 88-yard touchdown.
At this point it's pretty obvious, but Kovacs doesn't have good options.
Gordon forces the pitch. Michigan has Bolden ready to hit the QB if necessary, but he doesn't know that, and that's not the scheme.
The scheme is getting cut to the ground 13 yards downfield.
Presenting yet another ten yard run on a pitch. WSG Will Campbell holding flag.
[After THE JUMP: Air Force twists its mustache!]
The option has always been about making a defender wrong about who has the ball, thus effectively blocking him. Since you don't have to actually block him this means you can take out a slavering rage-beast with even the daintiest of skill position players.
Rich Rodriguez's innovation was taking the hazard-laden option and turning it into a simple yes-or-no handoff. The read option makes a guy wrong without requiring a pitch, and without getting your quarterback lit up time and again. Pairing that with plays that stretch the defense across the field horizontally opens up the box, forces safeties down, and creates the kind of environments that see his teams run for nearly six yards a clip.
Borges and Hoke have a different outlook on football. Last year when the inverted veer was running riot over Ohio State, they were consistently blocking the guy a Rodriguez-style offense would consider optioned off.
This worked, but I wondered if it was working because Ryan Shazier was an injured freshman who was pretty horrible in that game. It's hard not to look at what's going on with Michael Schofield in this clip and not pine for the guy to move past the OSU DE and take on Ohrian Johnson, thus likely springing Denard for another huge gain.
Last year both myself and fellow guy who does the picture paging Chris Gaerig thought that this was an execution issue that would be hammered out given enough time, but Tyler Sellhorn, a high school OL coach who frequently emails me tips and corrections, thought this was a philosophical thing:
I think Schofield and Omameh were coached to block the DE. Hoke/Borges do not like leaving unblocked defensive linemen out there. A famous unattributed coaching axiom that I am sure that Hoke/Borges believe in is: "First level defenders cause fumbles, second level defenders make tackles." To me, this is the "MANBALL" component of M's "option" game. True power running game people think like that. I think that is the reason there have been fewer really long runs (the second level has been blocked less consistently this season).
This is one philosophical difference: RR's first thought always was, "How can we mess with the safeties to get big yards when we break through the line", Hoke/Borges first thought is "How can we mess with the DL so they are less aggro (in run and pass situations) and we don't ever have a negative play." Both work well as we have seen.
The consistency with which Michigan guys were blocking the supposedly option DL was a point in his favor. At first I thought the Alabama game was the point at which this was undeniable, but now I think Alabama was blocking Michigan, not the other way around.
Optioning Nobody #1
It's Michigan's first drive. They've picked up a first down with a (horribly spotted) flare to Smith and a third down conversion from same. They come out in a two-back, three-wide set. Alabama responds with its base 3-4 set, half-rolling a safety into the box.
Michigan will run the veer. They pull Barnum (1), use Hopkins(2) as a lead blocker, and block down on the front side. This leaves the Alabama defender (3) there unblocked… for now, anyway.
Hopkins. You are not flaring out, my man. You are doing something that isn't that.
At the mesh point, Hopkins (1) has contacted the "unblocked" Alabama defensive end. This means he is now blocked. (Science!) Hopkins is also blocked. They are mutually blocking each other. Neither can go forward very easily.
This happens really fast. The DE is doing this on purpose. His goal here is two-fold: one, to force the handoff, and two to pick off one of the lead blockers.
Barnum(2) is still pulling for the front side; since the guys blocking down have actually done a pretty good job of getting push he's got a lane. Denard(3) sees the DE underneath Hopkins and gives.
And now it's over. Hopkins has indeed eliminated the Alabama DE, and Barnum reaches the hole as Smith sprints outside. Also sprinting outside: the totally unblocked Alabama LB.
Michigan's got some other problems, too, as the playside DE came through the double on the playside when Kwiatkowski released—you can see Schofield hunched over in an "oops" way right at the LOS behind Barnum. Given Smith's angle and Barnum's this is only a further indicator that Schofield got hammered on Saturday, not an actual reason the play doesn't work.
And that's all she wrote.
Who did Michigan block with Denard's legs on this play? Nobody.
[After THE JUMP: oops they did it again :( ]
Early in the 2011 football season I noticed an odd, seemingly-impossible task handed to one of the inside linebackers: going from one hash to the other while attempting to get enough depth to cover a receiver who's starting the play on the opposite side of the field from the linebacker.
Here's Brandon Herron trying this admittedly hard task:
This would end up a Western Michigan first down as the receiver would sit down at the sticks; you can tell that Carder has already identified the open man and is throwing before the WR makes his break. He'd do better on a subsequent attempt to cover this but still give up another completion. He got there, basically, but because of the angle he had to take to do so he wasn't in a position to do anything about the ball when it was in the air:
This was odd behavior to me. Most of the time a Picture Pages is trying to explain something; this one was just "I noticed this weird pattern… isn't it weird?" It seemed bizarre to ask a not-very-good coverage guy to make a very hard drop, especially when the quarterback is getting blitzed from the same side of the field and will naturally look for a hole in the zone from the direction of the blitz.
One of the ancillary benefits of checking out those coaching clinics, however, has been an increased understanding of what's going on when this happens. A hash to hash zone drop requires a particular set of circumstances:
- The defense is sending a zone blitz with a three-deep coverage behind it and three underneath defenders, one of whom is an inside linebacker*.
- The three underneath defenders are instructed to "drop off of" a particular receiver.
- The offense aligns or motions itself into a situation with three wide receivers to one side of the line.
In this situation… well, here's some Xs and Os that should help:
This comes from Coach Hoover via Smart Football and is a fire zone similar to one Greg Mattison explained at his Glazier Clinic talk. Michigan's running something similar above, with the WLB tasked with a "hot" coverage on a receiver. It is far from uncommon—the Coach Hoover post calls it "America's Fire Zone."
Our linebacker chugging across the field in a futile attempt to wave at a ball he's not looking at is "hot 3."
Hot 2? Hot 3? What? The goal of this defense is to get pressure without giving up big plays and often devolves into man to man coverage. Defenses number the opposing WRs from the outside. Above the two receivers are the #1 receivers and largely dealt with by the corners. The tight end is the field side #2 and will be the responsibility of the SS; the dropping "F" (in Michigan's scheme this would be the WDE, Craig Roh last year) is going to pick up any back coming out of the backfield to his side of the field; the WLB has whoever's left. Hot X defenders are supposed to get their man until he breaks to safety depth at 15 yards—again, pseudo man-to-man.
If you're running a three-under combo like this and you are facing an I-Form, no problem. The WLB is going to have to make up a yard or two of distance if he even gets a guy to play pseudo man-to-man on. He may watch a back stay in to block, in which case he's just an extra guy or becomes a delayed blitzer. If he does get a second guy leaking out of the backfield, he's probably a fullback. Crisis: not present.
Unfortunately for weakside linebackers everywhere, a million billion plays these days are run with three or four wide receivers on the field. This means the WLB is going to have to deal with a player who is a lot faster than a fullback and much farther away from his starting point, with results often like what you see above.
Mattison's video jockeys did find an example of the coverage working against a slot receiver, but where they had to go to get it was telling. It featured Brandin Hawthorne against Purdue running over the top of a seam route. It's not in the UFR because it was in garbage time.
Hawthorne took off for his drop the moment the ball was snapped without even thinking about the possibility of a run, which caused one of the coaches in the audience to ask after that odd behavior. Mattison hadn't selected the clips and this one did not jump out from his memory for obvious reasons, so he attributed it to Michigan's scouting and whatever the potential down and distance was.
He was right, but it doesn't take much scouting to predict a pass when the second team is in up 36-7 in the fourth. The one example Michigan had of this drop working against a spread formation was better evidence that it didn't work than it did.
His Rock, Your Scissors
Getting rained on like this is a frequent problem in the current college football metagame. Offensive coaches are always searching for ways to get bad matchups; defensive coaches are looking for free rushers and no holes. The hash-to-hash thing leapt off the page in the opener because it was strange and seemed really hard.
Unsurprisingly, it faded as the year went on. Like Mattison flipping his line every time an opponent changed its strength, it was a makeshift band-aid made necessary by a lack of experience with Mattison's defense. If Michigan's running a fire zone and gets a bunch of receivers to one side of the field, this year you'll probably get something like this:
That is from a post at Coach Hoover detailing a half-dozen coverage adjustments this blitz can undergo to combat bad matchups like you see above. Here the D sees a bunch of guys to the wide side of the field and switches the blitz, sending the WLB and giving the closer MLB slot duty. WLB high-fives himself, MLB grumbles, defense probably gets a better result.
That adaptation is well underway at Michigan. Linebackers will look at each other, pat their helmets or cross their forearms or give each other finger gunz, and check into something less ostentatiously weird. Not every time, but enough to relegate those hash to hash drops into the realm of oddity. We'll see them from time to time as Mattison tries to bait opponents into big wrong decisions and not much more.
[Hoover HT: Smart Football.]
*[Nomenclature NOTE: the middle and weakside linebackers are almost but not quite interchangeable and I use this term to distinguish them from the SAM, who is almost but not quite a DE.
Also while you're down here: these Purdue plays were actually cover four but all of Mattison's clinic stuff assumed cover three so I'll stick with that for the explanations. I assume Michigan was intent on preventing big plays in game one or didn't think WMU could run the ball at all.]
There's been some question over the no-catch ruling on Virginia Tech's 3rd down overtime prayer to receiver/punter Danny Coale. The play was ruled a touchdown live but overturned on review.
The setup: On 3rd down and 5 from the Michigan 20 in the first possession of overtime, Virginia Tech QB Logan Thomas attempted to hit Coale on a corner route in the end zone. Coale had Michigan's coverage (by safety Woolfolk and cornerback Avery) beat to the outside but the ball was slightly overthrown. Coale dove for what would be a spectacular one-handed catch, bringing the ball in just as he, and it, hit the ground just inbounds. The side judge ruled it a touchdown, but on review it was overturned and ruled an incomplete pass because the receiver did not have control of the ball when the ball hit the ground.
The rule: You can find it on Pages 72-73 of the NCAA rulebook (emphasis mine):
ARTICLE 7. a. Any forward pass is incomplete if the ball is out of bounds by rule or if it touches the ground when not firmly controlled by a player. It also is incomplete when a player leaves his feet and receives the pass but first lands on or outside a boundary line, unless his progress has been stopped in the field of play or end zone (Rule 4-1-3-p) (A.R. 2-4-3-III and A.R. 7-3-7-I).
The argument: The debate centers on whether or not Coale had "firm control" of the ball when it touched the ground. If the ball never touches the ground it's a clear reception, but since in this case nobody is arguing that the ball didn't touch the ground, the standard we're debating is whether or not Coale had established this firm control before the ball touched turf. For that we will consult the video.
With the grit of 40 Ecksteins, two Welkers, and half a Dileo, Coale reaches out and gets a hand underneath the ball. This is not "firm" control.
Coale now brings the ball between his forearms. This too would not be confused for "firm possession." However at this point he has a chance, if he can bring the ball to his chest and get it crooked in his arm, to prevent the ball from hitting the hard thing that is rapidly rushing toward him very fast.
So big and flat and round, it needs a big wide sounding name like 'Ow', 'Ownge', 'Round', 'Ground'! That's it! Ground! Ha! I wonder if it'll be friends with me?
Here is the money shot (clickening embiggens). Coale's elbow has hit the turf (just inbounds) but the ball is still between his forearms, not in his hands. It is hard to tell but the ball has now hit the ground as well, a nanosecond after the elbow.
The announcers were focusing on the elbow but the question is still one of whether the ball was firmly controlled by the player before it hit the ground.
A good test of this "firm control" (this is a sanity check not the final arbiter) is whether the ball moved when it hit the ground. It stands to reason that if the receiver had firm control of the ball when it came in contact with the ground it won't move that much.
If you recall, this is what doomed Junior Hemingway's TD catch attempt while down 8 in the closing seconds of the Iowa game. Hemingway actually managed to get two hands under the ball and secure it against his chest and arm a moment before coming down atop it. The "firm control" test in that case seemed to have been passed, but the catch was ruled incomplete because the reviewers saw the ball move in his possession after it hit the ground.
As you can see in the screenshots below, at the zero moment (when a part of the body made the player down) Coale's level of possession is way less than Hemingway's.
What controversially damned Hemingway was that after this the ball rotated about 90 degrees after the nose of the ball hit the ground, or so it was supposed since Hemingway's body blocked most of that. With Coale however the movement after contact with ground was pretty clear.
The ball is between Coale's forearms and possession only becomes firm long after the ball has impacted the ground. This I believe is what the review officials saw.
Incomplete! Time to bring in the 3rd string VT kicker who has been 4/4 today for an easy field g…oh snap!
But it's too close to call/not enough evidence to overturn! If someone is saying this to you they are confusing a Law & Order episode for reality. They have conceded that "incomplete" is the correct call, and are essentially complaining that it should have been ruled incorrectly because of a technicality in the literal meaning of the review rule. You cannot complain about calls the refs get right; that's not how complaining works. If you think the video is "inconclusive" you are conceding the call could have gone equally either way and saying it should be one or the other makes as much sense as whining that a flip of the coin should have been heads.
First: a confession. I really wanted to have the UFR's raring to go early this week. You must believe me. I wanted to play Fallout 3, which I'd saved as an end of the year treat, slightly more. So… yeah. I am through the main bit of that and am now plowing through the OSU game at all speed. I apologize for lackadaisical behavior and certainly hope Michigan is taking their film breakdowns more seriously than I am.
Second: a confusion. If you're like me, Michigan's inexplicable lack of a free safety was a surprising and disconcerting feature of the Akron State Golden Bobcats game. Before, Michigan had safeties. During… not so much. The reason for this was twofold. One:
Michigan spent a lot of time in formations like this with Kovacs rolled up to the line of scrimmage. This leaves just one deep safety.
The guy on the far right in this still coming over the top of a tight end that's pretty dang covered is Troy Woolfolk. Result:
Not quite an eighty yard touchdown. This was the very next drive after Woolfolk did the same thing on incredibly easy Braxton Miller touchdown one.
And then… I mean… WTF. It's third and twenty seven for OSU on their own three yard line, and despite having a mistake-prone Braxton Miller and third and a billion from the three, Ohio State throws.
OSU's in an I, Michigan is in its usual under, albeit a nickel. You can see Kovacs rolled up to the line at the top of the screen; Floyd and Woolfolk are your two deep safeties.
OSU goes straight dropback. Michigan rushes three. You can see Ryan dropping off in the frame below; He'll set up to contain scrambles. Martin and RVB are doubled and get nowhere, but Roh got a speed rush on Mike Adams:
Adams tackles Roh and picks up the holding call that will give Michigan a safety. Huge play from Roh against a first round pick. But that's another Picture Pages. (It's not, actually.)
Given time, Miller sets up and chucks it. Where is Woolfolk?
On the 25, covering the slot receiver. Oh, balls.
NICE. KNEW THAT WAS GOING TO HAPPEN THE WHOLE TIME. WOO COUNTESS.
54-yard TD on which Woolfolk and Countess jump the underneath route:
Near 80 yard touchdown:
Third and twenty-seven:
Items of Interest
WTF was this? The consistency with which Woolfolk was jumping the underneath route suggests it was part of Mattison's gameplan. Watch Woolfolk on the video just above: he is sitting on the slot receiver. But if Woolfolk is supposed to come up in a robber, why the hell is Countess playing outside of Posey on both of the latter two throws?
On the first one Spielman starts chattering about how Countess can't give up in the inside, and my immediate thought was "dude that is not on him, that is on the free safety losing his mind." Then the second one happens and… if it's third and twenty seven and your free safety vacates the deep middle for the third time in seven minutes(!) can it really be Woolfolk blowing an assignment? Probably not. I have not yet run across the sideline reporter screaming "OH MY GOD GREG MATTISON IS LITERALLY EATING TROY WOOLFOLK'S INTERNAL ORGANS FOR BUSTING SPECTACULARLY THREE TIMES IN SEVEN MINUTES AAAAAAAAAH THEY TOLD ME THIS WOULDN'T BE LIKE COVERING BRIAN KELLY NOT AGAIN NOT AGAIN NOT AGAIN." If Woolfolk had not been doing what he was supposed to, this would have happened.
So. We think Michigan is playing a three deep coverage on which the middle safety is intentionally sucking up on intermediate routes and the corner is playing outside. That does not make sense. That throw in the middle of the field is easy relative to deep fly routes down the sideline—that's why there's always a deep safety—and Michigan is giving it to OSU all day. Even on third and twenty-seven.
I don't get it, man. Even if you assume Countess is a freshman and thus screwing up, you're still putting him one on one with Posey all game with no help at all over the top. That doesn't seem like a good strategy.
On the other hand. Maybe you can't blame Woolfolk for these plays because he was executing his assignment. I find it hard to believe he is not at fault on the 54-yard WTF on OSU's first drive and the half-ending corner route on which he reacted very late. If Michigan lost that game that was going to be the kid's legacy, sorry to say. OSU's gameplan was based around attacking 1) Morgan and 2) Woolfolk with a side of Floyd and Countess.
Braxton Miller problems. Putting Kovacs in the box on every play restricts what you can run in coverage and exposes the middle of the field; that spot Woolfolk keeps running into is the same one Stoneburner will exploit for a huge gain on OSU's disconcerting 82-second touchdown drive in the fourth quarter. On that play the safety (now Gordon) stayed deep and there was no one to tackle once Stoneburner found the soft spot in the zone.
[Update: you can see Gordon turn and run for the post, leaving no one behind Demens.]
So your choice is between opening that up and opening up the deep stuff; obviously neither of those is a great choice.
Hurray Roh. Roh has not developed into the devastating pass rusher Michigan fans were hoping for this summer. He's got four sacks, which is amongst the team lead but far short of the numbers an impact player would put up. Here, though, he beats a very good tackle and gets paid for his effort. Thumbs up.
For a few weeks now people have been asserting that Denard Robinson has not been right. I wasn't sure what to make of those complaints since we didn't really get to see him in the open field much. After Nebraska, the answer appears to be "speed just fine, thanks" even if Lavonte David did hack him down on a couple of potential big gainers.
But, yeah, I do think something isn't right with Denard. That is not necessarily injury. Against Nebraska I've had more "argh" moments in re: Denard running than I've had in a long time. There was this:
That's a scramble on Michigan's botched end-of-half drive on which Denard, presented with a massive hole straight upfield, tries to bounce outside and gets tackled for a three yard gain—in bounds. Michigan ended up getting nothing here when making the really obvious cut (Denard knows the corner will be keeping leverage as his top priority) sees them set up with a first and ten near field goal range or better.
There was this on a play where Martavious Odoms came in motion to be a pitch man:
Denard handed off there when Nebraska had one guy containing two on the edge; Hopkins thumped straight ahead for three yards. Robinson keeps showing up in the run minus category because he's not pulling when the corner opens up like whoah.
Even when Denard did pull he made some inexplicable decisions. Here's one.
It's first and ten on the Michigan something or other and Michigan is in some variety of formation. So is Nebraska. I'm not entirely sure what they are because the director in this game comes from the Michigan Stadium Replay Guy school of framing where everything is a super tight closeup. I assume Michigan is in a standard three-wide set. You can see the slot to the bottom of the screen; M will have a WR on the line outside of him. The other guy could be in a trips formation below or to the top of the screen away outside.
Nebraska appears to be in a straight nickel with a safety as support.
Michigan runs a basic inside zone read, leaving the backside guy unblocked and flaring Koger out.
The above is the mesh point. Robinson is reading the end. The end is square but he is in shuffle mode.
What's shuffle mode?
Remember that period in 2009 when Carlos Brown or Brandon Minor would slam up the backside of the line and run untouched into the endzone? That was Michigan's response to various games defenses were playing with the backside end. The shuffle is the DL's response to that response. He is protecting the soft spot behind the backside tackle that the zone read often causes. He's not really containing the QB, though he's a lot more useful on a keeper than someone screaming down the line. He's more of an RB defender.
Michigan's started seeing this because they brought back the RR H-back inside zone where there is a guy cracking back on this DE.
On the next frame Robinson has just pulled it:
Denard Robinson is Denard Robinson. Koger has flared out to seal the playside linebacker. There is no slot to the top of the screen. Robinson is about to run to the corner for a billio—
I SAID, ROBINSON IS ABOUT TO RUN TO THE COR—
Robinson does run past Meredith and David but Huyge can't extend his block on that defensive end, and that defensive end…
…eats Denard after a two yard gain.
Items of interest
Yeah, this is kind of the same thing Scheelhaase did last week. You know, this:
The difference between the plays is that Scheelhaase screwed up his read and pulled when the defensive end was upfield in a QB contain mode. It was a plan B, one that worked. Here Michigan's plan A is "Denard in a race with a defensive end going the other way." That's a good plan. Let's try that.
It does almost work, but Huyge isn't great and he loses his guy. If Michigan sustains that block they get a decent gain.
Note the difference in the defensive ends. Clark is not shuffling down the line. He's a couple yards further outside and a yard in the backfield. Meredith is at the LOS and tucked in behind a Michigan OL. That should be enough to get Denard the corner but…
Denard seems hesitant. I don't know what the deal is. Unless there's something outside the frame that's relevant—not likely—it's really weird that Robinson wouldn't just run outside. Feel the panic in this linebacker:
That is a man going "oh shiiiiiiii" in slo-mo. He's done and Denard pops outside the DE and is dealing with that safety. Maybe he gets five yards. Maybe he gets ALL OF THE YARDS.
In recent weeks it seems like he's been less of a north-south guy. In this game he is way less of a read threat than he needs to be. In just the first half of last week's game you've got the above three plays, a speed option on which Denard cuts all the way to the backside of the line to little effect, and three or four seemingly obvious pull reads he's missed.
It is possible the reads aren't actually reads, but given what Borges has said it seems like they are. I'm not sure we've seen him pitch on the option yet, though. Is it really an option, or is it just a decoy?
The picture painted is of a guy who's thinking, not reacting.
Too much cram cram. This is a downside of not having a true base offense, I think. Lacking reps on all these things, Denard makes mistakes. If the speed option is just another way to run a QB stretch that's because they don't rep the option enough to be comfortable with the pitch. If they miss a bunch of keep reads it's because they're not repping it enough to make that clear to the quarterback.
Remember that last year Michigan largely dumped the read option in favor of just running Denard. This isn't a regression, it's Denard trying to do something he might not have been very good at last year. That plus an entirely new passing offense means there's a ton on his plate.
We've seen progress in the passing game. They may be emphasizing that since it turns out handing the ball to Toussaint isn't that bad of an idea even when it's a bad idea. Hopefully Michigan can get some of this corrected over the next week because Denard left a lot of yards on the field even when Lavonte David wasn't tackling him by his ankles.