I thought about calling this Diary "What the hell are you doing Josh Gattis?" but that seemed like it might be a little harsh. I actually started this out as a way to avoid BPONE and hopefully find a bright light in Gattis's play-calling, but the more I dug the more I fell into the BPONE. If you're looking for the optimistic take, it's that Michigan is actually quite good at running a spread offense. They're just not actually running a spread offense all that often.

Full disclosure: I'm a researcher and data analyst but have never coached football. So this is coming from a purely analytics stand-point.

My analysis here is based on the premise that Josh Gattis and Jim Harbaugh ultimately want this offense to be a spread offense. But given the lack of offensive success and discussion of transition costs, I wondered 1) How much has Michigan been running a spread offense? and 2) How well have they been running a spread offense?

To do this I scraped the play-by-play tables from Brian's UFR film analysis, in particular focusing on the personnel on the field, play-call, and outcome.

There are lots of ways that spread offenses try to put get their athletes in space, but for the first part of this analysis I decided to focus on two fundamental aspects of spread offenses (if you have suggestions for others I'm happy to include them):

1. Percentage of non-goal-line plays with 3+ WRs
2. Percentage of non-goal-line plays that utilize a read option, QB run, or run-pass option

The figure shown below shows the percentage of total plays for run plays (left panel) and pass plays (right panel) where Michigan used a spread concept with either less than 3 WRs on the field (maize bars) or 3+ WRs on the field (blue bars). Here I'm counting a play as using a spread concept if it includes at least one of the keywords "RPO", "read", "option", "QB", "keeper", "waggle", or "boot".

Michigan does actually use spread personnel quite frequently (67% of the time), but the interesting thing to focus on here is that they are only using spread concepts (rpo, read-option, waggle, etc.) 15% of the time! If we drill down a little further, they're only running RPOs on 8% of plays, which is especially crazy when you look at a team like Illinois, who Seth just charted in Illinois' FFFF as running RPOs 26% of the time.

So Michigan is putting personnel on the field in order to spread the defense out, but not using play concepts that are designed to exploit that fact.

If you look at the yards/play, success rate, and percentage of plays that gain at least 10 yards, you'll also see that Michigan is a much better offense when they have 3+ WRs on the field AND incorporate actual spread concepts into their play design.

Michigan is WAY better at passing the ball when they spread the field with 3 WRs and use RPOs, waggles, or boot legs. They're also better at running the ball with 3 WRs on the field, a stat that I'm sure makes Harbaugh shed a tear. Spread concepts don't really help the running game that much, but that is mostly because the QB hands the ball off on the vast majority of read-options, which negates any advantage spread concepts give the running game.

Don't believe me? For the next analysis, I broke down Michigan's running game into three types of plays: plays with a  designed QB run where the QB actually keeps the ball (plays with keywords "QB", "keeper", "read", or "option" where Patterson, McCaffrey, or Milton were the ball-carriers; maize bars below), plays where a RB ran the ball that came less than 10 plays after a designed QB run (blue bars below), and RB carries where there wasn't a recent QB run (gray bars below).

First, Michigan QBs have only run the ball 22 times all year in non goal-line situations, and only 14 of those plays have been designed QB runs (the other 8 were QB scrambles).

Those 14 designed QB runs (maize bars below) have actually been pretty effective on their own (5.5 yards/play, 39% success rate, 21% gained at least 10 yards), but what is more important is what they've done for the rest of the running game. Michigan has averaged 6 yards/play on the 39 RB carries that happened less than 10 plays after a designed QB run (blue bars below), with a 43% success rate and 17% gaining at least 10 yards, which is way better than RB carries that didn't follow a QB run (gray bars below), which only got 4.4 yards/play with a 36% success rate and 15% gaining at least 10 yards. 

What does this mean for the rest of the season?

So what can we take away from this? For starters, Michigan hasn't done a good job actually running a spread offense. But, they're actually quite good at running spread plays, and doing things like running the ball with your QB make the rest of the offense better too. Michigan only has to have the QBs keep 2-3 more times per game than they currently are to keep the defense honest. So the optimistic take is there are a number of relatively small changes that Michigan could do that would dramatically improve the offense. But, it's already week 6 of the regular season, so why haven't they done these things? Hopefully it's true that they weren't running the ball with Patterson because he was hurt. And hopefully as the season progresses they can implement more spread concepts into their offense. How much you believe that will happen depends on how deep in the BPONE you are.