THE POWER RANK

  • About
    • About The Power Rank
    • Start Here
    • Contact
  • Predictions
    • Games
    • March Madness
  • Content
    • Must Read
    • Blog
    • Podcast
    • March Madness Book
  • Rankings
    • NFL
    • College Football
    • MLB
    • Cluster Luck
    • College Basketball
    • NBA
    • CFB yards per play
    • World Soccer/Football
  • Members
    • My Account
    • Login
    • Become a member
    • COVID-19 Policy
  • Log in

Warning: Don’t think Michigan should be too big a favorite over Michigan State

By Dr. Ed Feng Leave a Comment

harbaugh_streetsMichigan travels to Michigan State this weekend in a game with two program headed in opposite directions.

Michigan has blossomed in Jim Harbargh’s second year as coach. The defense has excelled as the Wolverines have dominated every game except Colorado and Wisconsin.

In contrast, Michigan State started the year 12th in the AP Poll but have lost their last 5 games. It’s not a surprise the offense has declined with the departure of key talent from last year. But the drop off on defense led by Malik McDowell has been a shock.

So what will happen on Saturday in East Lansing? As of Tuesday afternoon, the markets favor Michigan by 23 point.

This point spread is too large. Let me explain.

The curse of small sample size

Some type of analytics give a spread even larger than 23.

For example, consider my rankings algorithm that takes margin of victory and adjusts for strength of schedule. Using only games from this season, this method makes Michigan 32 points better than Michigan State. With home field for the Spartans, this gives a predicted point spread of 29 points.

However, this prediction considers only 7 or 8 games for each team, and it’s dangerous to draw conclusions based on small sample size. In addition, randomness can affect the margin of victory through plays like turnovers, as Michigan State has a -4 turnover margin this season.

To get a better assessment of the right point spread for this game, let’s consult other sources of data.

Adjusted Yards Per Play

At The Power Rank, I also rank teams based on yards per play, a powerful efficiency metric that captures the ability of the offense to move the ball and the defense to prevent this movement. The same ranking algorithm takes data from each game and adjusts for strength of schedule.

By adjusted yards per play, Michigan State ranks 38th in the nation. While not stellar, the Spartans look much better by yards per play than adjusted margin of victory (90th).

To explain the discrepancy between yards per play and margin of victory, we consult Bill Connelly’s data on finishing drives. By points per trip inside the 40, Michigan State ranks 94th and 89th on offense and defense respectively.

The Spartans haven’t been clutch this season. However, this could change any week, as an offense ranked 34th in adjusted yards per play will almost certainly finish drives better than they have so far this season. And the defense overall should be better than their rank of 62nd by adjusted yards per play.

Michigan is 5th in team rankings based on adjusted yards per play, and this suggests they should be favored by about 10 points.

Market rankings

The betting markets provide another way to evaluate teams. This year, I started applying my ranking algorithm to closing point spreads in the markets to rank teams. Just like with the team rankings, each team gets a rating that gives an expected margin of victory against an average FBS team on a neutral site.

In these market rankings, I weight recent games much more than games earlier this season. For example, the spreads in week 1 get only 3% of the weight of this past week’s games.

This bias should favor Michigan, as they have improved from preseason expectations. In addition, they got an extra bump last week as a 40 point favorite against an Illinois team starting their 3rd quarterback.

The weight towards more recent games should also hurt Michigan State, as the markets have adjusted to their decline. The Spartans would have been more than a 3 point favorite at Maryland last week if the game had been earlier this season.

However, the market rankings predict a 9.5 point win for Michigan at Michigan State.

Ensemble Prediction

Members of The Power Rank have access to ensemble predictions that combine the factors discussed here with other calculations. These predictions have gone 53.7% (217-187 with 8 pushes) against the closing spread this year.

My member prediction has Michigan by 10.7 points over Michigan State. That seems low to me based on the play I’ve seen from both teams this year, and I could see Michigan covering more than 20 points.

However, if you think Michigan will one hundred percent dominate this game, don’t underestimate the randomness of college football.

Michigan State might play above their true talent in this game, as Mark Dantonio hasn’t suddenly become a terrible football coach the last 8 weeks. Throw in an ill timed turnover from Michigan, and it could be a close game on Saturday.

Filed Under: College Football, College football 2016, College Football Analytics, Michigan State Spartans, Michigan Wolverines

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Predictions from Ed Feng

I use my Stanford Ph.D. in applied math to make football and March Madness predictions.

To get a sample of my best American football predictions and March Madness cheat sheet, sign up for my free email newsletter.

Enter your email and click on "Sign up now!"

Popular Articles

  • How to predict interceptions in the NFL
  • The ultimate guide to predictive college basketball analytics
  • Accurate football predictions with linear regression
  • The surprising truth about passing and rushing in the NFL
  • Football analytics resource guide
  • The Reason You Can’t Avoid The Curse of Small Sample Size
  • The essential guide to predictive CFB rankings
  • How computer rankings make you smarter about sports
  • How to win your college football bowl pool
  • Do you make these 3 mistakes with college football statistics?
  • How to win your NCAA tournament pool
  • The Top 10 Things to Know About The Power Rank’s Methods
  • 5 insights from academic research on predicting world soccer/football matches

Recent Articles

  • Members: Super Bowl analysis for Kansas City vs Tampa Bay
  • Podcast: Drew Dinsick on Predicting NFL Conference Championship Games
  • Members: Football analysis for NFL Conference Championship, 2020 season
  • Podcast: Chris Andrews on Predicting the NFL Divisional Playoff
  • Members: Football analysis for NFL Divisional Playoff, January 16-17, 2021

© 2021 The Power Rank Inc., All rights reserved.

About, Terms of Use, Privacy Policy

Get a sample of my best football predictions

While I usually save my best predictions for paying members of the site, I offer a sample in my weekly email newsletter.


To get this service, sign up for my free email newsletter.


Enter your email and click on "Sign up now!"

No thanks, I'll make predictions without data and analytics.

{"cookieName":"wBounce","isAggressive":false,"isSitewide":true,"hesitation":"","openAnimation":false,"exitAnimation":false,"timer":"","sensitivity":"","cookieExpire":"","cookieDomain":"","autoFire":"","isAnalyticsEnabled":true}
  • About
    • About The Power Rank
    • Start Here
    • Contact
  • Predictions
    • Games
    • March Madness
  • Content
    • Must Read
    • Blog
    • Podcast
    • March Madness Book
  • Rankings
    • NFL
    • College Football
    • MLB
    • Cluster Luck
    • College Basketball
    • NBA
    • CFB yards per play
    • World Soccer/Football
  • Members
    • My Account
    • Login
    • Become a member
    • COVID-19 Policy