Det. News Article predicting (number of) Michigan wins this season

Submitted by StephenRKass on

There is an article at the Detroit News predicting the Michigan win total in 2015.

LINK:  2015 Predicted MIchigan Football win total.

The tag at the end of the article with info on the author:

Ed Feng has a Ph.D. in chemical engineering from Stanford and runs the sports analytics site The Power Rank.

Feng predicts a total of 7.7 wins for Michigan. There are 7 games he has as almost certain or very likely wins, 4 toss up games, and one we are likely to lose:  Ohio State.

One major factor:  between regression to the mean, and Baxter as Special Teams coach, Feng predicts Michigan will be much better in terms of the number of turnovers. Peppers as a superstar could also make a huge difference. On the negative side of things, if Rudock is injured, and none of the other quarterbacks can put it all together, we could struggle.

It is encouraging to see a prediction of 8 wins. I personally think that Michigan does better, with the "Harbaugh factor," and ends up with a 9 - 3 record.

UMProud

August 17th, 2015 at 3:33 PM ^

His methodology is flawed.

"The preseason model is simple but accurate. However, it makes no adjustments for Harbaugh, a coach who engineered remarkable turnarounds with Stanford and the San Francisco 49ers." - quote from article

He uses a statistical calculation based on the previous 4 years which is invalid in this case for the very simple reason that there is a new head coach.  His assumptions are based on the program performance itself irrespective of who is coaching.

ak47

August 17th, 2015 at 4:08 PM ^

I mean it makes a lot of sense, change in coaching isn't a variable you can measure.  Sure Harbuagh is a great coach but he also went 4-8 his first year at Stanford because he didn't have enough talent to really do anything else. Talent is the most important predictor, expecially in year one of a coach.  What this guys model says is pretty much that even Hoke would get to 7 wins this year  (and considering he manged to get to that 3 out of 4 years it isn't that far fetched) and there are 4 toss up games that coach matters.  Because coaching was shitty last year the numbers are guess us losing 3 out of 4 toss ups and whether you bet the over or under on the hat is how many game you think Harbaugh is worth.  If he is worth to games over Hoke the numbers suggest we could get to 10 wins.  If you think he is worth 1 more game you get 9 wins.  If you think he is a better coach but a coaching change cancels out the benefit you get 8 wins and if you think a coaching change is a negative factor in your one you get 7 wins.  I'm hoping Harbuagh plus two toughest games at home gets to 10 wins but really it all depends on how ready the team is to go at the beginning of the year.  Beat Utah and BYU 10 wins is on the table, split and its probably 9 wins, lose both and 8 is the most likely.

dnak438

August 17th, 2015 at 4:14 PM ^

You can't come up with a methodology that accounts for Harbaugh. Ed's model is flawed but so are all models. If his model weren't flawed then we wouldn't need to play the games and it wouldn't be a prediction, just a statement of fact.

ottomatic

August 17th, 2015 at 6:45 PM ^

The model doesn't claim precision.

"Despite its simplicity, the model has predicted the game winner in over 70 percent of games since the 2005 season."

Feng clearly states that the model doesn't factor in the Harbaugh factor. Therefore your criticism is flawed. So assign a strenght to the H factor. Is he good +1, or +n wins in those toss-up games?

I'm going to go +1 on the H factor. +1 on the Peppers factor for 9.7 wins.

Above and Beyond

August 17th, 2015 at 3:37 PM ^

Well, comparing Jim's first year to Hoke's last season, I think having Harbaugh and his staff (along with some matured players, too, of course) are good for at least another two wins. Harbaugh would not have lost to Rutgers and Maryland. Seven sounds about right, but we have a fairly favorable schedule, so I will not be surprised if Michigan wins eight or more.

Rafi

August 17th, 2015 at 3:41 PM ^

Help me out at math, because I am stupid.

7 games we're all but sure of winning = 7 wins

1 game we're all but sure of losing = 1 loss

4 toss up games = 0.7 wins...

...wait... what?

ST3

August 17th, 2015 at 3:44 PM ^

Almost certain wins: UNLV (97.3 percent).

Very likely wins: Oregon State (82.2), BYU (67.2), at Maryland (75.2), Northwestern (77.7), Rutgers (83.3), at Indiana (74.0).

Toss-up games: at Utah (40.4), Michigan State (45.5), at Minnesota (53.9), at Penn State (46.1).

Unlikely wins: Ohio State (23.5).

lilpenny1316

August 17th, 2015 at 4:26 PM ^

Home/road has something to do with it, but more important is when the games are played.  I'm skeptical our offense is ready yet to go out there and put up 24-28 points against a quality defense based on having a new QB and new schemes finally in live game action.  The D will have to carry us in that Utah game.  We should have things figured out a little better on offense by time we play State.  

PeterKlima

August 17th, 2015 at 3:48 PM ^

So, Utah is the SECOND toughest game on the schedule?  It looks like we have an euqal chance of beating PSU on the road as MSU at home.  Both near 50/50.

ijohnb

August 17th, 2015 at 3:55 PM ^

Oregon State - Win 23-16

UNLV - Win 31-10

BYU - Win 21-20

Maryland - Win 34-21

Northwestern - Win 20-10

Michigan State - Win 20-17 (first top 25 ranking of the season)

Minnesota - Loss 27-17

Rutgers - Win 34-13

Indiana - Win 37-9 (second top 25 ranking of the season)

Penn State - Loss 24-17

Ohio State - Loss 38-24

Outback Bowl v. LSU/Ole Miss Win 24-16