What do the computers say about Michigan versus Washington?

Submitted by goblue_in_colorado on January 2nd, 2024 at 4:25 PM

Disclaimer: I'm a stats nerd not a football expert or even a writer so apologies if you're coming here hoping for witty football commentary. My goal is just to provide the consensus opinion across the computer rankings using two helpful websites (Massey's composite ranking and the Prediction Tracker) to understand what various algorithms think about our chances next Monday.

In order to evaluate the consensus computer opinion, I chose to pull the computer predictions from Prediction Tracker, which has the predicted score differential from 41 computer rankings (plus the Vegas line) as well as the Massey Composite rankings, which uses the average computer ranking from 95 different computer systems.

TLDR: 71%-93% of computer algorithms predict a UM victory, with only a handful of computer algorithms clearly favoring UW.

Prediction Tracker

The Vegas line opened at +4.0 for Michigan and is currently sitting at +4.5, and 38 of the 41 computer algorithms (93%!!) have Michigan winning by an average of 6.1 points. The most common predicted outcome is a relatively comfortable 2-score Michigan victory:

Massey Composite Rankings

The composite rankings are a little more favorable to Washington, with the Massey Composite Index listing Michigan at #1 and Washington at #2. 28% of the computer ranking systems actually favor Washington, but the vast majority of those systems have Washington ranked #1 and Michigan ranked #2, which I would call a toss-up. From a ranking system perspective, the most common outcome is a tight game (where the ranking has Michigan and Washington within 1-2 ranks of each other):

Conclusion

It's interesting to me that the predictive algorithms have Michigan a much heavier favorite than the ranking systems, but I suppose some of that is the nature of those two types of computer systems. Predictive algorithms aim to predict the outcome of a specific matchup, whereas ranking algorithms just try to say which team is better without trying to predict a future outcome. Given that the predictive algorithms are more tailor-made for evaluating a specific matchup I'd tend to favor their expectations over the consensus ranking, but either way you slice it Michigan should feel good about their chances in this game.

Go Blue!

Comments

goblue_in_colorado

January 2nd, 2024 at 4:36 PM ^

Quick note: a lot of these rankings haven't been updated yet to include the results of last night's games. Given how both of those games played out I can't imagine the consensus here changes much, but it's possible a few of the toss-ups flip from slightly favoring one team to slightly favoring the other team.

Yeoman

January 2nd, 2024 at 5:46 PM ^

And some of those that have updated don't seem particularly serious. Here's an example that's included in Massey's composite page:

NCAA Football Rankings (FBS)

Anybody here think Penn State's better than Alabama? Better, does anybody here think Liberty is better than both, better than Ohio State, better than Texas and Washington...?

I appreciate that Massey includes everything he can find and leaves it to us to decide how to take it. But it's worth remembering that that's what his consensus is.