I GET IT
SITE NOTE: as is traditional during a bye week, the UFRs will be delayed a day, allowing us to ruminate in some more detail on a couple more plays that seem to be representative of larger trends.
Today in Michigan's running issues: an example of how all it takes is one breakdown for an otherwise promising play to end in the backfield. Offense and defense are opposite creatures in this regard. On defense, if you make a mistake it may or may not be punished, because someone can beat a guy and clean up for you, or the offense may not see the open receiver or cutback lane. On offense, an error is going to leave someone free and he will end your play more often than not.
A lot of Michigan's struggles so far have been one-guy breakdowns. This sounds like a promising, easy fix, but it's an unfriendly math problem. When you've got seven guys trying to execute, even if everyone has a 90% hit rate 0.9^7 is a 52% shot at someone not executing. At 95%—each guy doing their job 19 of 20 times—you still have a 30% failure rate.
That's obviously oversimplified; there are different mistakes that can make the difference between an unsuccessful run of three yards and an unsuccessful one of negative two yards. But I've been saying things like "it's just one block away from a big play" for a while now without actually seeing a lot of improvement in that category, and the previous paragraph is one of the reasons why.
Our exemplar is a zone stretch midway through the third quarter. It's first and ten after Drew Dileo extended an out route into the short seam and Gardner hit him. Michigan's in the I; UConn responds with a 3-4 look that has an extra guy hanging off the tight end side on the hash.
UConn did a lot of blitzing from the outside in this game, and this will be no exception. They'll shoot the guy on the hash upfield to be the force player and slant the other two inside, way inside in the OLB's case:
On the snap, nothing much is revealed as no one's made contact yet except Miller, who's underneath the nose tackle already:
That nose tackle is going to end up a long way downfield. I know we're all trying to take Miller's job, but he made a couple of nice blocks on these plays in the second half.
In the above shot, you can see the first steps of the defenders highlighted in the arrow picture coming inside. a half step later Michigan has both adjusted excellently and not adjusted excellently:
Both Glasgow and Lewan have adjusted their flight paths to intersect with the slanting defenders and have successfully made contact that will allow them to shove them past the play and open up a huge hole on the outside of the line, but Joe Kerridge is now trying to hit a gap that is not open.
When he does, he gets whacked.
Now off balance and a gap away from the actual hole, he's unable to block anyone. That's the one guy. When Michigan does this on defense I mention they got a two-for-one and usually good things happen afterwards.
Here bad things happen because Yawin Smallwood is now hanging out in the hole unblocked.
Fitz compounds matters by seeing this, considering a bounce, and then deciding against it, which gives up a couple yards.
Second and twelve blues.
Items Of Interest
Well, poop. Michigan blocks this really well on the line, getting both slanters sealed inside and driving the nose tackle back into a linebacker. But once Kerridge gets picked off, this play has a maximum reasonable expectation of about one yard. It only takes one error.
This would take some pretty fast recognition to fix. A lot of zone teams either eschew lead backs or place them in positions such that they, too, have a long path to the hole (think about "superbacks" in spread offenses that start lined up next to the QB). Kerridge is lined up to the playside about three yards in front of Toussaint and his first steps are upfield as he tries to build momentum for a bone-rattling LB block. Toussaint, in contrast, kind of waits and runs lateral to the LOS for a bit, so he has time to see the slant develop and find the hole that is unfortunately filled with one Yawin Smallwood.
Kerridge doesn't have that time. If he's going to make that read presnap he's probably guessing that the OLB is going to attack the gap outside of Lewan, and when that turns out to be wrong he's already committed. I'm not sure he can be any other way when he's lined up so close to the point of attack.
If you're going to do this it almost seems like you'd have to consider Kerridge another lineman and that Lewan should release downfield into Smallwood once the OLB crosses his face, but holy hell is that complicated. Michigan should be trying to make everything as dead simple as possible so they can have uninspiring runs that do pick up some yards.*
So this is a rock paper scissors minus. I don't think Kerridge has time to change his gap, and that gap gets filled by a slant. Even if Kerridge 1) has the option to pick his hole, 2) made a presnap read of the blitzer, and 3) assumed the OLB would slant inside, the OLB is outside of Williams so a one-gap slant takes him outside of Lewan. This puts Kerridge in the right hole. When the OLB goes two gaps over that's when the problems happen.
Toussaint bounce attempt again. Like that Nix play discussed earlier, here Toussaint has grim prospects that he makes a little grimmer by trying to escape. Despite all your rage, Fitz, you're still just a rat in a cage. Here it seems like he has been told that he needs to go N/S and remembers that after his natural inclination. Or he just thinks he can't get outside Williams. Whichever.
It is not an easy decision to bench Jack Miller. The entire world has already benched the guy for the Minnesota game; I'm 50-50 on that. I'm frustrated with him sometimes as well, but here's another loss on which the offensive line appears to be working just fine. He got dumped into the backfield once earlier in this game and struggled in a couple of pass protections (a couple of other pressures that came up the middle were not on him), but I wouldn't be surprised if Michigan soldiers on with their current five guys. Even if they don't, how long is Chris Bryant going to be able to stay on the field?
Also, folks speculating that Michigan might move Michael Schofield back to guard and insert Braden or Magnuson should stop. Miller is not bad enough that switching three spots on the OL and sending a good right tackle back to guard so you can insert a freshman is anywhere near an upgrade. That's a midseason switch worthy of a Rodriguez defense.
*[This is iso's role in the world. It is the DURRR SMASH of run plays, requiring almost nothing other than brute strength and rarely picking up more than three yards, but rarely losing any.]
Estimates are approximate. Michigan's spent maybe half of their snaps in the shotgun/pistol on running downs this year, running about five things: jet sweeps to Norfleet, QB draws, speed option, the inverted veer, and a kind of alternate to the inside zone called "belly" that Rich Rodriguez was fond of during his brief spell in Ann Arbor.
Oddly, Michigan hardly runs anything like a base play from the shotgun. They don't run the stretch, they don't run any iso or power type plays. There is a faint smattering of inside zone, but that's it, and that's not anywhere near established. In their first three games of the year I've got them down for three inside zone runs from pistol or shotgun; they went for a total of three yards. Nobody's cheating to a base run play against Michigan.
This allows opponents to tee off on the things Michigan is kind of good at. More importantly, it often seems like they're going up against opponents who are better drilled at defending modern offensive concepts than Michigan is at running them. Here's an example:
Michigan's in the pistol with Kerridge as a fullback, Williams the tight end, and both WRs to the field. It's first and ten. UConn responds by shifting their line to the strength (an "over" front) and aligning their linebackers about evenly with a safety rolled up over Williams.
Michigan wants to read the end to the bottom of the screen. That will allow Michigan to blast the playside end off the ball with a sustained double; Williams will head for the safety as Kerridge deals with the playside linebacker. If the end crashes, Gardner pulls. If he contains, Gardner keeps.
Snap. You can see Williams release, Lewan and Glasgow begin to bash the playside end off the line, and the frontside UConn LBs react to gaps that may need to be filled.
Gardner is now considering the end, who does what ends are supposed to do these days: try to split the difference so that they can be useful on a handoff and still contain the QB. Gardner's trying to figure out what to do about this:
(Note that Lewan and Glasgow are battering their guy inside effectively.)
Now, I think that's a pull. I gave Gardner a minus for that, because I want Gardner to test the edge against a defensive end who's standing at the LOS. But it's a gray area for the quarterback. The end is neither flat-out containing or crashing down; this is a situation in which errors are common.
At the decision point, Gardner gives. Kerridge is staring down two defenders, doesn't know which one to deal with, doesn't really deal with either but it doesn't matter because whoever he does in fact block is just going to funnel to his buddy.
Poor Damn Toussaint, 2013 edition.
That's a loss of two yards.
Items Of Interest
Remember the wheel route from the Notre Dame game? That's the opposite of this. Borges saw the wheel open, gave it a try once, and then pulled it out in a similar situation later for a big gain. Here Michigan just abandons these runs. How is this a similar situation? Like ND, UConn is playing this play in a certain way. If they play it in the same way again, you can alter what you're doing to bust it open. But Michigan hasn't done this, and so rarely does things that are misdirection that twitter blows up about it when they get five yards on it.
Arc, arc, arc, arc. Nebraska demonstrated the tweak against Michigan a couple years back on an almost identical play. Michigan shuffled Jibreel Black down, planning to contain with Kovacs on the outside. The fullback approached the end, and then…
Black could not recover in time to get out on Martinez, Kovacs got a guy in his face, and Nebraska ripped off a 23-yard gain.
Here it's a little different because the end does have contain on Gardner, but if Michigan pokes at that belly play again they can do something similar. Instead of having a true read it's a designated Gardner keeper on which Kerridge's job is to get outside and block whoever that contain guy happens to be, Michigan can burn the shuffle.
This is a paragraph of disclaimers and explanations. That's my thought process when I see things like that on the zone read, because that was Rodriguez's thought process. He probably forced defenses to create the shuffle a few years back when he started blocking backside ends trying to crash down and shooting Carlos Brown or Brandon Minor through the gaping hole scraping linebackers would leave. That burned scrape exchanges hard for a while, and then the cat and mouse game moved on.
Michigan is deficient at cat and mouse in the run game. I'm not trying to suggest that Michigan has to be a spread option team for their offense to work better; I am pointing this out because it remains my wheelhouse and it's a good example of the things Michigan doesn't do because they are a jack-of-all-trades offense that doesn't see how a defense is responding and do something to break it. Because to do that Nebraska thing above your fullback has to rep it and sell it, etc. It takes practice time.
Michigan's not thinking the zone game well at either the field level or the box level because they're not committed to it, and that extends to everything from stretch to power to iso.
Also maybe chalk that up as a missed read for Gardner. Because Michigan doesn't rep it consistently enough? I don't know. Has to be a consideration.
In other sad runs Michigan got out-schemed on. UConn was sending guys off the corner with frequency, but Michigan did not recognize it despite UConn tipping it hard. This inverted veer featured the dead giveaway of a safety moving down to line up directly over a wide receiver:
And on this one, how would you describe the playside corner's presnap technique? Is "right angle to wide receiver" a thing?
Michigan just gets lined up with 14 or so seconds on the clock and thus doesn't have much time to recognize what the defense is doing and adjust, like you saw Notre Dame and Akron do to Michigan's detriment several times. They're just eating bad playcalls. That's a natural consequence of spending 25 seconds in a huddle and not recognizing that one of the most common responses to spread stuff is to send extra guys off the edge.
None of this has anything to do with the offensive line. These are two TFLs and one miraculous Gardner escape wiped out by a Funchess holding call (which, BTW, ugh) on which the offensive line plays no part. The problems go deeper than their issues, which we'll get to later. This is Borges and to some extent Gardner—I don't know if he's got checks here—getting beat by the defensive coordinator. They got some back with the speed option, FWIW.
Who's up for a tedious 150 comment thread questioning whether it's worthwhile to read this? I certainly am! I hope there are content-free arguments. Let's make sure to ignore Ka'Deem Carey's 2000 yards last year when we're incensed at the idea Rich Rodriguez might be able to coach a run game.
I'd planned on posting another Picture Pages this week from the Notre Dame game on the assumption that there wouldn't be much from the Akron game to discuss. Surprise! The good news—ish—is that this continues our discussion of where Michigan's line is.
This is another Toussaint lost yardage play that marks the last time Michigan's run their as-yet-unsuccessful counter to their zone game. ABC provided a slick closeup of events (the difference between doing this for an ABC broadcast and BTN one is enormous—viva ABC), so we'll get a zoomed-in look at goings-on.
ND's in an even front; Michigan has two tight ends. They'll pull Schofield as the rest of the line tries to sell another zone.
Michigan immediately runs into the problems that is Louis Nix, who either isn't buying or is just assigned to slant outside of Glasgow.
That's bad, that'll happen sometimes when you play Nix. As Nix surges upfield of Glasgow, Schofield sees him and knows he's got to deal with that lest Toussaint get swallowed in the backfield.
Glasgow violates the fake rule I made up by turning upfield. Schofield's coming, but he doesn't comprehend that he isn't totally screwed until…
Both guys go to Nix, leaving one of ND's ILBs unblocked. Toussaint makes things worse by trying to bounce around a rampant Nix, and gets chopped down.
That's a two yard loss.
Slow unnecessary for this one.
[After THE JUMP: Notre Dame faces the same problem, finds different results.]
Michigan's running game wasn't quite as bad as it looks in the stats, as they had some good gains wiped out by phantom holding penalties, but it warn't good. One of Michigan's main issues was not getting the Akron nose tackle blocked, but when this happened the guy in the backfield seemed only partially at fault, because Michigan was asking him to do something very tough.
Here's a play at the end of the first quarter. Michigan has just shifted various players around and gotten Akron to do this in response:
That is a massive gap between their nose tackle, who's "shaded"* over Miller, and their end, who's outside of Jake Butt. Michigan decides to run at this, which seemed like a good idea at the time.
*[ie, lined up between the center and guard, closer to the center.]
Michigan's going to run a stretch to the wide side of the field, and get a loss out of this. I know. So. Akron's got about seven guys in the box. Michigan has two uncovered OL to the playside plus Kerridge. This should be easy.
[After the jump, it's not going to be]
NOTE: UFR will be later than usual today, because of Dos A Cero IV. Lo siento.
BONUS: Soon to be number one in Google for highly competitive search term "mesh goofball punish." Dolla dolla bill y'all. See you on my yacht.
Michigan had seen enough of Notre Dame's maniac linebackers by the second quarter to expose their aggression. On their drive after ND had kicked a field goal to make it 17-13, M opened with four straight runs.
The one tailback touch in there was a one-yard loss on power. A Notre Dame linebacker shot the gap, meeting Kyle Kalis two yards in the backfield, and Toussaint bounced it outside without having any hope of doing something out there. The other three plays all worked because they used Devin Gardner's legs to punish the overcommitting Notre Dame defense. Each of these plays could have been a 35-yarder.
- Michigan fakes inside zone, has Gardner run for 7.
- A false zone read keeper breaks outside for 35.
- Toussaint stuffed for loss of one on power, holding on Houma.
- Inverted veer keeper for nine, phantom holding on Miller.
One of them did go for 35. We discussed the veer keeper a bit in the last picture pages post: it sucked a Notre Dame linebacker well into the backfield and may have been a touchdown if Funchess didn't spin around.
The opener is an interesting play; it follows.
Michigan has it first and ten from their 37 and comes out in a pistol 3-wide formation. Notre Dame goes with their 3-4. (I swear ND was mostly a 4-3 team in this game.) Shembo is tight to the LOS over the tight end with Day next to him. ND has two deep safeties and Jaylon Smith in the grey area over the slot.
Michigan will show inside zone, with quickly aborted doubles on Nix and Tuitt forced by the usual hard reaction by the ND linebackers.
I wish I'd remembered that Bob Diaco linebackers play like they're hopped up on goofballs before the game. Here's the mesh point on Michigan's first offensive snap:
One ILB is almost to the line of scrimmage and the other is a yard back. This is way closer than almost any other team will be, and it is absolutely consistent. ND linebackers fire hard on any run action.
For the most part it's worked for them. Michigan won the Denard after Dentist game despite getting ten yards on eight tailback carries. ND's defense last year was lights out. Notre Dame's hyper-aggression at that spot has been a problem for Michigan's run game for the last couple years, as they haven't had effective counters. Their main one is the waggle, and we all know how that worked out last year.
Not much changed in this one early. Michigan's tailback running game was drips and drabs because of a lack of an effective counter trey. (You know, that play they showed against Central where Taylor Lewan pulled to the backside… ineffectively.) The longer runs they did acquire were almost entirely Fitzgerald Toussaint forcing errors out of ND safeties. For example, the UFR chart on Toussaint's early 14-yard sideline run has four minuses for bad blocks and no positives. Yikes.
Let's get a baseline in this one and see how Michigan responded later. This is a second-quarter zone stretch in which Michigan puts two tight ends to the top of the screen; ND responds with a rare three-man front (they were a 4-3 in this game that occasionally lined up in a 3-4 as a curveball) with a safety walked down:
This looks like a called blitz but in practice it's difficult to tell the difference between an actual blitz and the playside linebacker hauling ass at the first gap he sees. It's just alignment. Notre Dame got some TFLs out of this gap-shooting, and even when they didn't those linebackers forced Michigan to disengage from double-teams on Tuitt, Nix, and Schwenke early, with predictable results.
Meanwhile, the backside linebacker would ignore any cutback possibilities and flow parallel to the line of scrimmage at approximately the same rate the tailback did:
The overall effect is six guys at the line with one hovering behind for cleanup and that overhanging safety able to provide quick support. Even when Notre Dame screwed up this was mostly effective.
A moment post snap, Michigan opens up a gap as they go to double the two defensive tackles:
Note that the backside players on the ND DL are stepping away from the play, which they can do because the MLB is jetting into the gap they vacate. This also allows the backside LB to flow as he does.
Some of this is tough to see, but in this frame:
- Miller has disengaged from Nix in an attempt to cut the charging LB, which he does not do. He does knock him off balance somewhat, possibly contributing to his overrun of the play.
- Meanwhile, Glasgow and Lewan try to handle two guys who have disappeared from the frame: the playside end and charging safety.
- Both tight ends have locked on the playside LB, who is the force player.
- Schofield chases the MLB, who he has no angle on, but could still block if Toussaint cuts back.
A moment later the LB flashes into the backfield wide of Toussaint and runs by; playside end got his legs caught up in traffic and ends up falling, pancaked. Both tight ends are still on the force guy:
This is one of them gap things?
Except it's got a linebacker in it.
BONUS! Here is a super slo-mo version.
(Does this help? If this helps let me know.)
Items Of Interest
Notre Dame got away with at least a couple errors here. The playside end ends up underneath Glasgow on the ground and they spent a linebacker blowing past Toussaint to little effect. (They did get Miller down but offenses will take one for one trades.) If that can happen and Michigan picks up three yards you can tell that it's tough sledding.
Tough sledding. The goofballs approach makes life tough on offensive linemen, who have to make split second decisions to leave guy and then try to block a rampant guy with tons of momentum before they are ready. This is tough, and Michigan didn't do a good job of it.
Toussaint could put his foot in the ground here and make a cut. Schofield is chasing that linebacker and you occasionally see the blocking develop such that the tailback can make a hard cut upfield behind that OL and suddenly make him relevant. Right about here…
…if Toussaint goes hard north and south aiming for the hash he may shoot past that linebacker and into open space. That's why Schofield keeps following that guy despite not having an angle. It may not work, but you're at least giving yourself a shot. Toussaint had a good day overall; here I think he missed a cut.
The offensive line… I punt. They had a very tough first half against this line, and these linebacker gap-shots don't help. Miller just barely throws off that linebacker if he does anything, but then again that linebacker zips past the play he's moving so fast. If that guy can't make a play, can the OL make a play?
Meanwhile Glasgow gets a pancake that is probably aided by the ND lineman tripping on the blitzer's feet; Lewan ends up putting a safety on the ground. Points for them. This one was a lot better blocked than some.
Funchess is very frustrating. On this play, the linebacker to the top of the screen is obviously the force player*. Butt obviously has him kicked out. Funchess continues to block the guy the whole damn play instead of releasing downfield and getting a hat on the safety. There is no way this is right.
Meanwhile, on the single inverted veer Michigan ran, Notre Dame hyperaggression bit them as one of their linebackers roared up a gap and pursued Toussaint, as did Tuitt. Gardner pulled and got a nice gain. It could have been a lot nicer, but Funchess turned around again:
Also not right, as with Kalis headed to the outside the linebacker is the optioned guy. I know Michigan's blocked guys who are supposed to be optioned before; even if that is the nominal plan, nothing good ever comes of turning 180 degrees when you're a blocker.
That left no one to take the only guy standing between Gardner and a touchdown:
I guess it's better that the play wiped out by a nonexistent holding call was nine yards instead of a thirty-one yard touchdown?
We just saw this happen against Central Michigan; it's closing in on a pattern. Funchess remains a tight end in name only. The mental stuff is more bothersome than any lack of technique. All he has to do on some of these plays is vaguely bother a guy and Michigan can break a long one. Hopefully he makes some progress here in the next few weeks, but the relative prominence of Jake Butt in this game is not a coincidence.
*[IE, the guy who sits on the end of the line and accepts a kickout block. He positions himself such that if the back tries to bounce it outside he either gets tackled for has to take such a circuitous route that by the time he gets the corner for guys are waiting for him. Since things usually go badly—very badly—for the defense if the force player is not doing his job, he is limited in how dynamic he can be what with throwing blockers away and getting TFLs, so doubling him is useless.]
Michigan did exploit this, eventually. You may notice that I'm not complaining about how Michigan didn't adjust to this. This is a tease.