The Game. Ohio State and Michigan are undefeated. Again. This episode previews the game based on what types of models are better at predicting college football outcomes.

This episode of The Football Analytics Show is also available on:

• YouTube
• Spotify
• Apple Podcasts

Here is the script used to create the podcast:

With one second left in the game against Notre Dame, Ohio State scored a touchdown as Chip Trayanum plunged into the endzone from the one yard line. It capped a 15 play 65 yard drive and earned Ohio State a critical early season win.

After the game, Ohio State coach Ryan Day couldn’t hold back his feelings. He said:

I’d like to know where Lou Holtz is right now. What he said about our team, I cannot believe. This is a tough team right here. We’re proud to be from Ohio. It’s always been Ohio against the world, and it’ll continue to be Ohio against the world.

In the aftermath of a big win, you can excuse Coach Day for not knowing some of the facts about Ohio State’s toughness.

First, Notre Dame only had 10 players on the field on the last play of the game. They were also missing a player on the previous play from the one yard line. That play was an incompletion to Marvin Harrison Jr.

Ohio State had the odds tipped in their favor as they won the game.

Second, Ohio State was not that successful in running the ball against Notre Dame. It seemed like RB TreVeyon Henderson had a great game with 104 yards on 7.4 yards per carry. This is much better than the college football average 5.1.

However, 61 of those 104 yards came on a touchdown run in the 3rd quarter. Otherwise, Henderson had 43 rushing yards on 3.3 yards per carry.

One explosive can skew a metric like yards per carry, which is why I prefer to look at success rate. For an offense, a play is a success based on the necessary yards towards the next first down.

• 1st down, 50% of these necessary yards.
• 2nd down, 70% of the necessary yards.
• 3rd, 4th down, obviously all of the necessary yards.

Success rate measures consistency on a play by play basis, and one outlier play can not skew this metric.

Against Notre Dame, Ohio State had a 26.9% rushing success rate. This is much worse than the college football average of 42.5% this season.

The run game was an issue for Ohio State earlier this season. TreVeyon Henderson came into the season as a Heisman contender, but didn’t show up against Notre Dame except for a lone play. He missed the next 3 games with an injury.

A few weeks later, Ohio State earned a tough 20-12 win over Penn State. However, the Buckeyes had a 32.4% rushing success rate in that game, albeit against a tough defense. Ohio State won the game because Marvin Harrison Jr. played up to his potential as the best WR in college football. He had 162 yards on 11 receptions and a critical TD.

However, things changed the next week. Henderson came back from injury and had 162 yards on 24 carries against Wisconsin.

Over the past 4 games, Ohio State has had a 48.1% success rate on run plays. This recent performance matters in making a prediction against Michigan.

Break

Michigan has had a remarkable season. After somewhat sleep walking through a weak out of conference schedule, they started dominating Big Ten teams.

• 52-10 at Minnesota
• 52-7 vs Indiana
• 49-0 vs Michigan State

No one is going to write letters of recommendation to the college football playoff committee about beating any of these three teams. However, Michigan has had a better margin of victory against these three teams than Ohio State. They were also more dominant in SR against Minn and Michigan State.

Based on data from the current season, I take metrics like margin of victory and success rate and adjust for opponent with my proprietary algorithms. I put these together to get a sense for team performance for the 2023 season.

Based on this current season data, Michigan is about 3 points better than Ohio State.

Michigan has answered two key questions since the preseason.

First, the receiving corp was considered the weakest part of this team. However, WR Roman Wilson has emerged as a top threat, as he has gained 3.14 yards per route run. Any statistic over 3 is considered excellent in college football. TE Colston Loveland continues to fulfill his potential with 32 receptions, second behind only Wilson.

Second, the defense has exceeded my expectations. Michigan ranks 12th in my adjusted success rate. 

One thing is clear. Michigan is in a better position to beat Ohio State this season than either of the last two. As of Wednesday morning, Michigan is favored by 3.5 points at home.

Contrast this with the past two years. Last year in 2022, Ohio State closed as a 9 point favorite at home. There was likely late market movement towards Ohio State when it became clear that Michigan RB Blake Corum wasn’t at full strength. In 2021, Ohio State closed as a 6.5 point favorite on the road.

For those past 2 seasons, Ohio State had CJ Stroud at QB. With his success as an NFL rookie, Stroud looks like the best QB prospect since Patrick Mahomes.

Kyle McCord has taken over at QB for Ohio State, and he is good. With probably the best set of receivers in college football, Ohio State is 8th in my adjusted passing success rate.

However, McCord is not CJ Stroud and never will be. With this change and Michigan’s strong performance this year, the market has shifted on the results of this game. 

Michigan is favored by 3.5 points. My adjusted numbers based on this season would make this spread even larger. However, season long averages are not the best way to predict the outcome of college football games. Let me explain.

Break

For a long time, I ran my member college football predictions based on these season long results. This supported The Power Rank as a business for almost a decade.

Then in 2020, I couldn’t run those season long numbers like I usually did. With COVID, all one hundred and some teams were no longer connected by out of conference games. This renders the math I use for schedule adjustments meaningless.

I had to adapt. I developed a way to adjust a team’s rating each week based on the past game results. This is a more aggressive system that moves quickly on a team. This system tends to lean more towards recent performance like Ohio State in their last 4 games as opposed to metrics from the entire season.

I’ve found this more aggressive system to give a smaller error than my old model based. I trust this model even more when there is a reason for the recent surge in performance. Situations like when a Heisman caliber running back returns from injury.

Remember, Ohio State struggled running the ball before TreVeyon Henderson came back from injury. After the Penn State game, Ohio State ranked 112th in my rushing success rate adjusted for opponent.

Henderson came back the next week, and Ohio State has moved up to 64th in my adjusted rushing success rate. This rank based on season long numbers doesn’t reflect how good the run game is right now.

For most of the past month, my aggressive model had Michigan ahead of Ohio State. This past week, Ohio State moved ahead of Michigan if only ever so slightly.

My member numbers based on this aggressive model favor Michigan by 1.5 points at home. I’d lean Ohio State +3.5.

Break

I live in Ann Arbor, and the Harbaughs live in my neighborhood. While I’ve never seen Jim, I always see his dad Jack walking in the neighborhood.

Earlier this season, I ran into Jack and exchanged some pleasantries. I asked him how his son was doing, as Jim was serving a three game suspension and didn’t coach those non-conference cupcake games.

Jack said he was doing fine. I asked him whether as a dad he was going to hang out with his son during those games. Jack replied:

Hell no. I’m going to the game.

Jim Harbaugh will not coach against Ohio State, and it seems safe to assume that his dad will not join him in his Ann Arbor home. But does Jim’s absence matter?

Maybe. I can’t make any quantitative arguments about it, but it seems helpful to have every coach available in the most important regular season game of the year.

In addition, Michigan LB coach Chris Patridge was fired last week for his involvement in the sign stealing activities. Michigan is down two coaches. We will see if this has an impact on the game.