Massey Ratings for Women's College Hockey

[Back2BackU-MnPride2002]

Not sure how many people are aware that the Massey Ratings also rates Men's and Women's Hockey.

Here are the updated standings for games through Saturday, Mar 16th


1. Minnesota ------ 2.62
2. North Dakota -- 1.75
3. Wisconsin ------ 1.68
4. Boston College - 1.64
5. Boston Univ ---- 1.63
6. Cornell Univ ---- 1.61
7. Harvard Univ --- 1.49
8. Mercyhurst ----- 1.47
9. Clarkson -------- 1.44
10 Northeastern --- 1.43
11 UM-Duluth ----- 1.36
12 Ohio St --------- 1.35

[skatefast]

Re: Massey Ratings for Women's College Hockey

What the basis, or the statistical methods, to come up with these ratings?

[NUWH DogHouse]

Re: Massey Ratings for Women's College Hockey

What the basis, or the statistical methods, to come up with these ratings?

The further west you are the more points you get? I'm kidding of course.

[Back2BackU-MnPride2002]

Re: Massey Ratings for Women's College Hockey

What the basis, or the statistical methods, to come up with these ratings?

http://masseyratings.com/index.htm


All of your questions should be answered there. He seems to be the person who has applied his rating system to the largest # of different sports out there.

[David De Remer]

Re: Massey Ratings for Women's College Hockey

From the perspective of marketing an alternative to the Ratings Percentage Index to the NCAA, the "Massey BCS ratings" are more appealing than the Massey Ratings, since the Massey Ratings use scoring outcomes. I hadn't realized Massey was doing women's NCAA hockey until recently, and I'm glad he is. Since there's at least one other college sport (football) that uses his ratings, that gives him some credibility relatively to the others as an RPI alternative (as unfair as that may be).

The Massey BCS ratings are very similar to the Rutter rankings, as far as I can tell. Both assume that the probability that team with rating A beats team with rating B depends on where A-B is in the normal distribution. The other differences between Rutter and Massey are fairly minor, at least at the end of the season.

KRACH, on the other hand, uses the logistic distribution, which has the simple mathematical property that the team rating A beats team with rating B is A/(A+B). Practically, the big difference is you're punished/rewarded more for extreme outcomes in Massey or Rutter -- a really bad loss will hurt you more in the two rankings that use the normal distribution.

Statisticians would say the logistic distribution has "fatter tails" than the normal distribution, meaning that you're more likely to have an unusual outcome by pure chance in a model using the logistic distribution. Since the normal distribution has thinner tails, unusual results end up being attributed more to your ranking in the systems using the normal distribution (Massey, Rutter).

[LakersFan]

Re: Massey Ratings for Women's College Hockey

To expand on Dave's comments, if you look at Massey's web site, his ranking system for college hockey may include home ice advantage and margin of victory. I do not include those two pieces of information in my ratings in order to have a better comparison with RPI. Massey's web site is football heavy, so exact details of his methods for hockey are unclear.

[IceIsNice]

Re: Massey Ratings for Women's College Hockey

What the basis, or the statistical methods, to come up with these ratings?

The further west you are the more points you get? I'm kidding of course.

From the perspective of marketing an alternative to the Ratings Percentage Index to the NCAA, the "Massey BCS ratings" are more appealing than the Massey Ratings, since the Massey Ratings use scoring outcomes.

I think you spelled this wrong....this looks like the Massey GDS ratings!

[Back2BackU-MnPride2002]

Re: Massey Ratings for Women's College Hockey

To expand on Dave's comments, if you look at Massey's web site, his ranking system for college hockey may include home ice advantage and margin of victory. I do not include those two pieces of information in my ratings in order to have a better comparison with RPI. Massey's web site is football heavy, so exact details of his methods for hockey are unclear.

Why do you need to have a better comparison with RPI? 75% of the RPI is based on what other team's do. It's a ridiculous system to use, honestly.

[Eeyore]

Re: Massey Ratings for Women's College Hockey

Why do you need to have a better comparison with RPI? 75% of the RPI is based on what other team's do. It's a ridiculous system to use, honestly.

A large chunk of all of the ratings systems is based upon what other teams do. Not to go very far in defending RPI, but the main reason the coefficients get larger for Opponents' Opponents' Winning Percentage is that this path will converge towards .500 for all teams and so you need to give it a heavier weighting in order for it to affect the rating much at all.

[David De Remer]

Re: Massey Ratings for Women's College Hockey

To expand on Dave's comments, if you look at Massey's web site, his ranking system for college hockey may include home ice advantage and margin of victory. I do not include those two pieces of information in my ratings in order to have a better comparison with RPI. Massey's web site is football heavy, so exact details of his methods for hockey are unclear.

He clearly uses scoring for the Massey ratings, and clearly doesn't for the Massey BCS ratings.

As for "home-field advantage," he again clearly uses it for the Massey ratings. For the Massey BCS, I'm guessing the answer is no, because the HFA effect he estimates is an effect on scoring, and if he did estimate something for HFA, he'd probably list it on the Massey BCS page. And since the Massey BCS ratings are very highly correlated with yours, I'm guessing he doesn't do HFA.

[myStuff]

Re: Massey Ratings for Women's College Hockey

How far back did Massey go in order to include Niagara in the ratings to Mar 16/13?

[LakersFan]

Re: Massey Ratings for Women's College Hockey

Why do you need to have a better comparison with RPI? 75% of the RPI is based on what other team's do. It's a ridiculous system to use, honestly.

If you want to show a model/statistical method is different than RPI, than you can't include margin of victory or home ice advantage. If you did, any difference in ranking/rating could be attributed to either of those factors as well as the different approach. By not including them, any discussion about Rutter/KRACH/RPI is strictly about methodology, not about what should or should not be included as data.

[Back2BackU-MnPride2002]

Re: Massey Ratings for Women's College Hockey

If you want to show a model/statistical method is different than RPI, than you can't include margin of victory or home ice advantage. If you did, any difference in ranking/rating could be attributed to either of those factors as well as the different approach. By not including them, any discussion about Rutter/KRACH/RPI is strictly about methodology, not about what should or should not be included as data.

OK, I get it, that makes sense. Thank you for explaining that.