ESPN matchup predictions

Submitted by TK on

Im not sure how they use these computer formulations and they will obviously change week to week, but the numbers are interesting at this point. 

ND we have 32 pct win probability 

Wisconsin 59 pct win probability 

MSU 39 pct win probability 

Penn St 55 pct win probability 

OSU 29 pct win probability 

 

So essentially solid underdogs on the 3 rival road games, and slight favorites on the big 2 home games.  Very surprised on the ND number.

 

Gameboy

August 30th, 2018 at 11:30 AM ^

I am not sure what you mean by "if computers decided these games". All the numbers are just percentages. Computer isn't "deciding" who is going to win, it is just telling you how likely it is. Even with the percentage above, it means Michigan should win 1 out of 3 games. That is a pretty significant chance of winning.

I am not sure how accurate the ESPN's model is, but if you polled all college football "experts" and averaged them, it will be pretty similar to the numbers above. We just think these numbers are too low because we are bunch of homers.

Joined: 11/28/2010

MGoPoints: 9656

The Maizer

August 30th, 2018 at 10:37 AM ^

Those predictions are based in large part on our terrible offense from last year. If we again have a terrible offense, those numbers are correct. If not, who knows?

Joined: 08/07/2015

MGoPoints: 10228

Arb lover

August 30th, 2018 at 10:47 AM ^

ESPN's computer prediction is basically based on current rankings, with a heavy weighting in favor of the home team. That's why even though odds have it as a close game, ESPN predicts  a much less likely Michigan victory. 

Joined: 11/26/2017

MGoPoints: 6915

Lumpy_wolverine

August 30th, 2018 at 10:48 AM ^

The percentages are based on ESPN's "Football Power Index" which has Notre Dame ranked #6 (!!) and Michigan ranked #11.

The FPI is described as:

The Football Power Index (FPI) is a measure of team strength that is meant to be the best predictor of a team's performance going forward for the rest of the season. FPI represents how many points above or below average a team is. Projected results are based on 10,000 simulations of the rest of the season using FPI, results to date, and the remaining schedule. Ratings and projections update daily.

So, some of the ranking would seem to be based on last season's performance, and on the difficulty of the schedule this season.  It's not surprising that Michigan is underrated in that metric.

Joined: 02/01/2013

MGoPoints: 1554

MGoStrength

August 30th, 2018 at 10:51 AM ^

Interesting.  I like our chances against ND more than Wiscy.  Clearly they think we have a significant advantage at home (Wiscy/PSU) and/or struggle on the road (ND, MSU, OSU).

Joined: 07/29/2010

MGoPoints: 20002

The Mad Hatter

August 30th, 2018 at 11:12 AM ^

Does playing in South Bend really give ND that much of an advantage?  It's a toy sized replica of our stadium and it will probably be at least 30% maize and blue in the stands.

Joined: 02/17/2015

MGoPoints: 112109

Ali G Bomaye

August 30th, 2018 at 11:14 AM ^

Two reasons why nobody should pay attention:

  1. ESPN odds are generally garbage. I don't know why, but their system spits out some of the most random crap I've ever seen.
  2. Empirical rating systems will (and should) underrate us this year, because they don't factor in transfers. I don't know if y'all have heard, but we have a pretty significant transfer player starting for us at the position where we most needed it.

Joined: 07/08/2008

MGoPoints: 17668

LeCheezus

August 30th, 2018 at 11:20 AM ^

It's based on ESPN's FPI metric.  I'm sure some contrarian will disagree with me but FPI is terrible.

If you're bored you can read about it here.

https://en.wikipedia.org/wiki/Football_Power_Index

Highlights:

- Includes data from previous 4 seasons with emphasis on previous season

- Includes returning starters on offense and defense with QB counting as more

- Strength of the team's recruiting class (with input for transfers)

My main beefs:

- The recruiting metric is a bit of a wildcard.  Does it adjust for transfers in and out properly? (Maybe- Probably?) Does it adjust for obvious busts or breakouts? (probably not)

- Roster and coaching turnover on a CFB team over 4 years is huge, there is a good chance data very irrelevant to the current team is being put into the model (example, this year's preseason FPI still includes Hoke's last year)

I know we're all info starved but common sense will tell you that any predictive model for this year that has no data from this year is going to be pretty shaky.

Joined: 06/11/2017

MGoPoints: 9855

danross

August 30th, 2018 at 12:28 PM ^

Vegas (-1) and Power Rank (-3.4) both like ND although Ed has said he thinks his own 3.4 point margin might be a bit high. Sagarin has the game as essentially pick'em.

Joined: 07/23/2009

MGoPoints: 322

The Fan in Fargo

August 30th, 2018 at 1:04 PM ^

39% chance at ND my ass. They’re going to get the beat down they’ve had coming for awhile. All I’ve been reading is whoever wins will have won the turnover battle. Really? Who gives a flying fuck if Shae airs one to Nico and it gets picked? How is it any different than a punt? Momentum change you’re thinking? That concept is for the soft and weak. Play like men and sack up. Act like a bunch of swinging d_cks and that it doesn’t phase you. Bring the wood and pain. No big f ups and Just Win Baby!!

Joined: 09/04/2016

MGoPoints: 1877

plamonge

August 30th, 2018 at 1:45 PM ^

I've thought about this a lot and for me it comes down to this: we are about to witness what an experienced, dynamic, athletic, and exciting mobile quarterback does for an offense. Michigan already had lots of talented offensive weapons; now we've added that key component to make it electric. 

Michigan 31 ND 17.

 

 

Joined: 09/29/2014

MGoPoints: 1025