Michigan football fans are not conditioned in their modern state to feel anything good. The sport so many of them live and die with has burned them too often over the past 20 years for many Wolverines fans to feel truly excited as the calendar flips from month to month in the fall.
This September has put these fans in a tough spot. Michigan football is off to a 3-0 start, and if you’ve watched the games or looked at the numbers, you know it’s been a crisp 3-0. The Wolverines have beaten Western Michigan, Washington and Northern Illinois by a combined 107 points, the largest cumulative margin of victory among Football Bowl Subdivision teams. Michigan was supposed to win all of these games, but there is winning and then there is covering the spread by nearly 19 points per game. Save for losing star receiver Ronnie Bell to a right knee injury for the year, it would be tough to imagine Jim Harbaugh having a smoother first three weeks of a season.
Yet Michigan fans understand how this goes. Harbaugh was 18-3 in August and September since his takeover of the program in 2015 through last season (though there were no such games for the team in 2020). That was the seventh-best winning percentage in FBS in the season’s opening stretch during those years. The problem was October and November, months in which Harbaugh has gone 30-15 — a nice enough record for most teams but not nearly enough to get Michigan over the Big Ten East hump. Many are understandably circumspect about what lies in store for 2021.
Before we go further, it’s worth doing the necessary Michigan-related hedging. The Wolverines might fall apart and lose three to five games, as they’ve done several times in recent years. They might be great but fall short in their regular-season finale against Ohio State, which is now an annual bit of misery (save for 2020) no matter what’s happened up to that point. In short, Michigan could do what Michigan has done too many times before. These potential outcomes all point to fans shielding their hearts from vulnerability. Why get too serious with a team that has hurt you? But if Michigan fans can stomach it, they should let themselves live a little bit. Three games into 2021, there is no reason Michigan won’t have Harbaugh’s big breakthrough this year.
In Harbaugh’s first six years, his offenses were good, but not great, and “good” does not win the Wolverines’ division, much less anything beyond that. From 2015 to 2020, Michigan was 19th in FBS in expected points added per game on offense, adjusted for opponent quality. The results varied a bit by year, but its most productive offense by far was in 2016, when Michigan came a J.T. Barrett fourth-down spot away from making the Big Ten Championship. That team produced an adjusted offensive EPA of 15.16 per game. That was the best college offense Harbaugh has ever fielded, save for the 2010 Stanford outfit that had Andrew Luck and various NFL pass-catchers and produced a figure of 20.19.
The Wolverines have been unable to approach their 2016 level over the last four years. But this year, Michigan leads all of FBS in adjusted offensive EPA and is tracking, at this early date, to produce the most efficient offense by EPA of Harbaugh’s career. Adjusted EPA flattens out the results some, since it accounts for the competition level faced, which means the computers aren’t calling fraud on Michigan’s excellent start.
Along the same line, Michigan’s September success looks like a shift upward even when compared only with its previous nonconference schedules. In strictly regular-season, nonconference games in Harbaugh’s first six years, Michigan’s offense was a combined 44th in adjusted EPA per game. This year, Michigan is second. And again, that’s an opponent-adjusted stat, which suggests that Michigan’s leap can’t be chalked up totally to a schedule that lacks a team like Florida or Notre Dame, two recent season-opener foes for Harbaugh’s teams.
Harbaugh’s new starting quarterback, Cade McNamara, is outdoing his predecessors in a similar way. The junior, a former four-star recruit, threw for 10 yards per attempt in nonconference games and posted a 60.5 Total QBR. Both figures put him ahead of Harbaugh’s previous starting QBs in the same situations: Jake Rudock, Wilton Speight and Shea Patterson. And Michigan’s lead tailback, Blake Corum, has blown away Harbaugh’s previous primary running backs in both yards per attempt (8.5) and missed tackles generated per touch (0.3) against teams outside the Big Ten. In various offensive areas, Michigan is lapping its old self.
On defense, the Wolverines are under new management this fall. First-year coordinator Mike Macdonald arrived in the offseason from the Baltimore Ravens, where he had worked since 2014 for Harbaugh’s brother, John. (Macdonald was going to be co-defensive coordinator with fellow new hire Maurice Linguist, but Linguist left to take the top job at Buffalo before the season.) Macdonald replaced Don Brown, who’d been Michigan’s defensive coordinator since 2016 and had a reputation for his infatuation with two things: blitzes and man coverage. During Brown’s time in charge, the Michigan defense played man-to-man coverages on 48 percent of opposing dropbacks, the ninth-highest rate in the country. (The average was 33 percent.) In front of all coverages, Michigan blitzed 12.6 times a game, 10th-most in FBS. So far in 2021, the Wolverines have kept playing a lot of man coverage, though less than before — 40.7 percent of the time, which still ranks ninth. Their blitzing has also eased up a bit, to 10.33 times per game (tied for 46th-most).
It’s a little early to know what shape Macdonald’s defense will take the rest of the way, but it looks like his plan is working. In nonconference, regular-season games under Brown, Michigan was fifth in adjusted defensive EPA per game. Under Macdonald, Michigan is 19th in the same situations — a step back overall, but not compared to Brown’s most recent work. There’s no way to make an apples-to-apples comparison from last year to this year because there were no nonconference Big Ten games last fall. But the early returns say Michigan’s defense has cleaned up a 2020 mess where the Wolverines finished the season an ugly 109th in adjusted defensive EPA per game overall. This year, Michigan is 16th, and its straightforward yards allowed per play are down at this point from 5.6 to 4.5. Even if the defense regresses significantly in Big Ten play, it looks like the bleeding from 2020 has slowed.
Michigan has played what looks like a light schedule, but it’s not that hard to look at it in the right light and see something decent. WMU beat Pitt, NIU beat a Georgia Tech team that is probably bad but almost beat Clemson, and Washington looked dysfunctional but has one of the more talented rosters in college football.1
Whatever you think of this schedule, Michigan has pulverized it to an unusual extent. Forward-looking projection systems tend to believe, even if lots of humans aren’t there yet. Bill Connelly’s SP+ and ESPN’s FPI, two opponent-adjusted systems, each have Michigan No. 6 overall. In SP+, the Wolverines rank 12th on offense, eighth on defense and second on special teams. In the AP Top 25, Michigan is 19th, which is fair for now and may turn out to look low.
The Big Ten might be ripe for the picking, too, or at least more so than usual. The conference has a lot of interesting teams that weren’t at all interesting last year. Penn State looks like a serious contender. Iowa has a punishing defense, Michigan State seems to be getting back to some of its mid-2010s ways, and even Maryland and Rutgers are presently undefeated. But the league’s biggest hoss looks more vulnerable than usual. Ohio State is 37th in Defensive SP+ and has already taken play-calling responsibilities away from its defensive coordinator.
The Game is in Ann Arbor this year, and that combined with a slightly reduced OSU gives Michigan one of its best chances in a while. Predicting a Michigan win would be foolish, but for a rare change, so would be dismissing the possibility out of hand that the Wolverines give their much more successful rival a lot to handle.