Previously: The Story. Podcast 12.4A, 12.4B, 12.4C. Quarterback. Running Back. Wide Receiver. Tight End. Interior OL. Offensive Tackle. Defensive End. Defensive Tackle. Linebacker. Cornerback. Safety. Special Teams.

1. Can we have some speed in space this time?

give it to the atom [Patrick Barron]

Probably. Last year's edition of this post went to a lot of trouble projecting what Josh Gattis's offense would look like based on the scraps Michigan provided in spring:

Michigan did mix in a number of plays that hinted at an evolution of last year's arc read system and a new frontier in misdirection and wide plays. … [this looks like] what Michigan's going to try to do on the ground:

Reads. Two of these plays are genuine post-snap reads, and many more were sprinkled throughout.

Edges. Michigan threatens the edge on all these plays. The split zone that kicks it off freezes the playside DE because McCaffrey is a threat to keep and the arc is a threat to have that keep go a long way. A stretch naturally tests your edge. Then a speed option and the capper: threatening one edge and attacking the other one.

Tight ends. Gattis has the offense but he's not sticking his fingers in his ears and shouting "na na na." Warinner and Harbaugh have their influence and you can see it in this package. Michigan's 2TE packages gave them an absolute ton of stuff they could do with arc games last year—more even than you may remember, because there were a number of plays that were there tactically but weren't executed because of a lack of familiarity. It looks like Michigan is building on last year's arc game.

Very little of this actually happened in the first half of the season. The quarterback refused to keep the ball on zone reads. Tight ends were extant but their involvement was muted compared to the 2018 offense, in which Sean McKeon may have been Michigan's most critical run blocker. Michigan threatened some edges in the opener by adding a pitchman to their arc package, but then that went away with nary an actual pitch.

[After THE JUMP: Gattis as Jalen Mayfield]

In the air it was similar. RPOs were extant but still mostly the basic oops-this-is-a-slant stuff that works against MTSU but not against better defenses. And the edges were un-explored there as well. Here's that graphic of Shea Patterson's pass distribution again:

red: more targets than average; blue: fewer targets than average

Speed in space did not happen. So what gives? A whole bunch of different things.

Transition costs played a role. People don't often think of screens as a skill a team or player has (including me: UFR dumps them out of our success metric for QBs). They are. Precision timing is required. Michigan didn't have it. Michigan did try to incorporate screens early, but messed them up:

[Patterson] also had multiple screen issues. Once he pulled the trigger too fast and let a DE recover; once he airballed entirely on one that was set up to go a long way; once he flung a ball too high for a covered Eubanks in the middle of the field.

Personnel was another issue. Patterson was relatively short, which can be a problem on screens that have to get over defensive linemen with outstretched arms. Michigan did not get comfortable using Giles Jackson until late in the year, when he was a much better player. Chris Evans was unavailable. Michigan's two main backs were not the sort of guys you dink it to and watch the YAC counter spin. This quote from the running back preview also applies here:

"Comparing to last year, I would say lot of times a guy would get out on checkdowns where they weren’t really a factor in the quarterback’s thought process or progression. A guy like Chris who can beat the majority of linebackers or safeties he runs a route on and can make the play and catch the ball, he adds a little bit different element where you can start including the running back in a progression for the quarterback of having the ball go to him first or second."

Young quarterbacks love checkdown guys.

This year the coordinator returns, the quarterback is 6'5", Giles Jackson is fully operational, and Chris Evans is back. Also that quarterback is a giant wildcard who will likely have training wheels for a while. Speed in space is going to happen this time around.

2. What are we expecting from Gattis?

I project Gattis's binder has pages that are not blank [Bryan Fuller]

Josh Gattis was a roll of the dice for Jim Harbaugh, a guy who had never been a solo offensive coordinator before Michigan stepped in with an offer of total control. That involved an unprecedented step back from Harbaugh, who's always been his own offensive coordinator in fact if not in name.

Gattis was the coaching equivalent of a new starter, and there were hiccups. The Army game still stands out as an inexplicable approach:

So what are you supposed to do?

There's a lot of stuff designed to attack shuffle ends since they're extremely common. Michigan even uses one as a staple: split zone. The super baffling thing is they barely ran split zone in this game. It was used on two short-yardage touchdowns and one five yard run on which Collins ate a –2 for not blocking the guy who ended up tackling.

Other than that Michigan ignored the other half of the arc package. The third and three before the second disastrous fourth down is a scrape exchange, which split zone is great on because the crashing DE gets kicked and the LB runs himself out of the play. Bredeson puts a DT on his ass, but Eubanks runs past the DE on the arc:

Split zone wins here; split zone is great against exchanges; split zone is literally the play you are faking when you run arc. Split zone was a success all three times it was run in this game. Where the hell were the other 10 instances of it?

Michigan ran inside zone over and over. The blocking was good. Zach Charbonnet did well. He got 3.3 yards a carry. Army defensive backs played ten yards off Michigan receivers with impunity. It was depressing.

Things got uglier as the season went along. Michigan almost did not run the ball against Wisconsin and their 2-4-5. Michigan scored 10 points against Iowa; UFR's rock-paper-scissors grading came in at –5. Michigan had lost that trademark Harbaugh wrong gap trickery:

Michigan was pretty good at getting guys going the wrong way under previous editions of Harbaugh and is now pretty bad at it. The main source of wrong direction was the arc read—yes—which featured more heavily than you'd think because Michigan had one pull on it before their attempted four-minute drill.

By the fourth quarter when a couple of Michigan gambits had petered out the results were too often like this:

Everyone gets their block, unblocked LB at POA, two yards.

You can see this in the stats from this game in Bill Connelly's advanced box score. Against a tough defensive front, Michigan's 2.69 line yards beat the national average of 2.45. Their stuff rate of 13% was a lot better than the 20% national average. Their opportunity rate—which has been redefined to the percentage of carries getting four yards—was a whopping 58%, 12 points higher than the national average. But Michigan cannot escape the fact that most of the time their run game is playing down a man in the box. Highlight yards per opportunity*: 2.9. National average: 5.1.

Things started to turn around. Michigan got up to even in RPS against Penn State because they hoovered up free yards on bubble screens, which in turn opened up holes on the ground when Penn State started overplaying them.

PSU LB #6 to bottom

They didn't get a tactical win in that game because they had their snaps timed up on a half-dozen plays, which is a good way to lose RPS points as long as you fix it—and Michigan did.

The next week was Notre Dame, and I probably don't have to tell you what happened there. But it's fun! Despite a first-half monsoon that all but prevented passing Michigan's previously moribund ground game had Notre Dame chasing ghosts. Michigan ripped off a giant chunk on a trap, a Harbaugh staple; they ripped off a giant chunk by threatening a speed-in-space flare screen:

That felt like the offense we were imagined preseason: a fusion of the Harbaugh/Warinner ground game from 2018 with an edge-testing modern spread approach.

After that Gattis was locked in. RPS against ND: +6. Maryland: +1. MSU: +17(!). Indiana: +5. OSU: +6. The first halves against OSU and Alabama were both outstanding but submarined by execution errors. By the time the dust cleared:

All systems go. The most optimistic thing about the way the year finished: Josh Gattis dialing up several gameplans that were, as the kids five years ago say, straight fire. Gattis straight up dragged Shea Patterson to a finishing stretch that disguised his season-long regression with a ton of RPS+ plays on which he was able to turn short throws into big gains. Michigan nuked MSU's defense in their Super Bowl and had first halves against OSU and Alabama in which Michigan put up nearly 300 yards but did not convert those yards into points efficiently because of execution errors.

Second halves in those games did not go nearly as well; even so  Michigan put up an average of 400 yards against the SP+ #2 and #7 defenses. Failing to score the way you should is an issue but Michigan's players spurned tons of opportunities that were there. You can't argue with the approach, just the execution.

Michigan will benefit greatly from Gattis's maturation as a playcaller and what looks like a midseason hybridization of the approach. If a player had this kind of debut season I'd say he's on a stardom track. Gattis should be the first Michigan offensive coordinator who people actually like since… uh… Fritz Crisler?   

3. Can the offensive line hang together?

Filiaga and Vastardis: certainly big enough [Patrick Barron]

We're now done with major offensive questions that can be backed with any data. This one and the next one are pure handwaving. Here's my handwaving about the offensive line: it'll be all right. The situation:

• LT: Third-year player who was a standout pass protector in two starts last year and allowed Michigan to leave their star at RT.
• LG: Enormous fourth-year consensus four-star who got stuck behind NFL OL.
• C: Fifth-year senior walk-on who beat out a freshman and maybe another potential fifth-year guy.
• RG: Fourth-year player who was a solid run blocker as a redshirt freshman against three elite Ds. No longer at tackle so pass protection issues minimized. Almost won the job away from the star of the line until he got hurt.
• RT: Future first round pick who went toe-to-toe with Chase Young and gave up no pressures.

There will be a step back; I don't think it'll be particularly noticeable from 1,000 feet because Michigan will be winning "blocks" they didn't win last year with reads and RPS positives.

The shakiest spot is indeed center, the spot most likely to implode the entire enterprise. Vastardis had less competition than anyone else; Carpenter dropped off the radar in a way that makes me think he was held out of practice for a while either because he had COVID or had to isolate because of a close contact.

Still, there is circumstantial evidence this will be okay. Vastardis has had the backup C job since midseason last year and from all appearances everyone is comfortable with him stepping into the lineup. There have been no sudden position switches, and Michigan has a couple of guys (Barnhart, Rumler) who certainly could have been jammed into a battle at center if one was warranted. I also place some weight on the press conference takes where Vastardis is the first name out of a coach's mouth.

It'll be okay. Probably.

4. Uh… Milton?

somewhere in this tangled mass of possibilities is Milton [Patrick Barron]

I TOLD YOU I DON'T KNOW

come on

Well fine: Milton has a few things going for him other than being a 6'5" cyborg. One: it looks like he's going to have a wide array of little bastards who are impossible to catch that he can use as a crutch. Two: the first time he has a game where he keeps it five times and opens up the offense on the ground he's going to make life so much easier on himself as a passer. Three: he's got the operational Gattis described above.

I think he'll be okay most of the time and wildly over-confident 17% of the time. There are going to be moments he fits a ball in tight window that most quarterbacks would never attempt. There are going to be moments the window intercepts the ball and runs it back for a touchdown. His pocket presence is unlikely to be good, but that won't be a change from last year's offense.

He's not likely to be a guy who changes the state of the rivalry. Hopefully he looks like he might be able to do that next year.

5. Well?

A ramp-up season beckons. Michigan is incredibly young. Of the 22 guys on the two deep only Chris Evans, NIck Eubanks, Ben Mason and Andrew Vastardis are seniors. Ronnie Bell, Andrew Stueber, and Chuck Filiaga are juniors. Everyone else is an underclassman.

This is going to make for a season of exploding in all directions. Michigan should be far more explosive than they were a year ago. The QB has a cannon, the gameplans are fixed, and a wave of incredibly fast guys is coming online. They will also be prone to head-shaking gaffes. Milton is going to whistle some hideous interceptions at defenders. We will remember what a missed blitz pickup looks like.

It should still feel better than last year because those errors can be written off as youthful indiscretion instead of wasting Michigan's best offensive line in at least a decade. It'll be easier to focus on the whiz-bang stuff, and if the Milton-Bell mind-meld takes the overall results should be better. A turnover spike looms, though.

BETTER

• Any semblance of a QB run game >> Patterson 2019
• Massive stable of running backs >> Freshman or hurt versions of those guys minus Evans
• Sophomore Jalen Mayfield > Freshman Mayfield
• Sophomore slots > freshman slots
• Ronnie Bell post Gallon mind-meld attempt > Sophomore Bell
• Erick All, Nate Schoonmaker > junior Eubanks
• BEN MASON, FB > ben mason, dt

PUSH

• Nick Eubanks == Sean McKeon
• Joe Milton throwing == Shea Patterson throwing

WORSE

• Sophomore Ryan Hayes < NFL-guard level Jon Runyan Jr
• Three new interior OL starters <<< three NFL players dancing blindfolded
• No Nico <<< Nico
• One backup QB in the era of COVID < Two with no COVID
• ~zero backup outside WRs << DPJ/Black as #3/4 WRs

LAST YEAR'S STUPID PREDICTIONS

This isn't going to be pretty.

Patterson sets a Michigan record for touchdown throws. He finishes 7th in Heisman voting.

Patterson was the ~7th best Big Ten QB. Bzzzzzzzzt.

Tru Wilson gets the most snaps of any Michigan back and barely pips Charbonnet in carries. Charbonnet has the most yardage with 650.

Half point. Charbonnet did lead Michigan in yardage with 726, but Wilson got hurt early and Charbonnet Wally Pipped him in pass pro. Hassan Haskins was the #2 back, just a few yards and carries back. Wilson was #3 but a long way back.

Peoples-Jones cracks 1,000 yards and has one punt return exclamation point.

Quarter-point, as DPJ failed to come anywhere near that prediction with 438 yards but did have a couple of exclamation point punt returns. Why wasn't this about the guy averaging 12 yards a catch as a sophomore? I don't know.

Nick Eubanks gets more snaps than Mike Sainristil, but fewer touches.

Speed in space failed to materialize so Eubanks crushed Sainristil in touches 25-8. Bzzt.

Sean McKeon has the breakout year I predicted for him last year, becoming an arc blocking assassin and utilizing that to bust into the open field a bunch. He more than doubles his catches.

McKeon was hurt for chunks of the season; this was still a bad prediction since his blocking was decent but he was close to a nonfactor as a receiver. Bzzt.

Mayfield is a step back from Taylor Lewan as a redshirt freshman but clearly on a stardom track.

Hey, something unambiguously correct. One point.

Michigan finishes 8th in offensive SP+.

This ended up being not completely terrible since Michigan was able to recover from an abominable start to finish 21st. A 13 point gap is still a bzzzzzzt.

THIS YEAR'S STUPID PREDICTIONS

• Milton has a better YPA and completion percentage than Patterson (8.0 and 56%, respectively) last year but throws 50% more INTs per attempt.
• Ronnie Bell is the clear leader in targets and has 85 yards per game there; the competition for the #2 spot is a complete free-for-all that sees four or five different guys in a relatively tight band.
• Between Giles Jackson and Chris Evans, spread H is a real thing. Combined they average 5 catches a game.
• The offensive line is a B+ unit with a couple guys PFF loves.
• Erick All is the leading receiver amongst TEs.
• Ben Mason emerges as a short-yardage snowplow once again, this time as the lead blocker. There's a two-back package that features him and gives the business to the first team that sees it.
• No SP+ prediction this year because the lack of nonconference games is likely to throw that system for a loop; I will assert that the offense will be better overall than last year's, which finished 21st.