Massey Ratings - what is the formula?

[Kearney Kingston]

There seems to be a big difference between RPI and Massey Rankings, yet, both are mathematically calculated by results of each team and success of opponents. I know the 25% / 50% / 25% formula of the RPI, but what is Massey’s mathematical make-up?. Quite honestly, over the years, Massey seems more accurate.

[mikegarrison]

Proprietary.

[BeachbytheBay]

Nov 15, 2024 13:27:09 GMT -5 Kearney Kingston said:

There seems to be a big difference between RPI and Massey Rankings, yet, both are mathematically calculated by results of each team and success of opponents. I know the 25% / 50% / 25% formula of the RPI, but what is Massey’s mathematical make-up?. Quite honestly, over the years, Massey seems more accurate.

I doubt Ken Massey would provide his 'secret sauce'

I have emailed him, primarily to correct results. He has his ratings covering tons of sports.

1) His predictive tool is NOT accurate, or believable, if one runs it best of 5 sets. I do not know why but it does not make sense. it gives too many lopsided predictions. as an example Riverside just beat Long Beach and Hawaii, his predictor said it was 100% before the match for Beach or Hawaii.

now if one ran the tool in best of 1, then the prediction was like 94%, which makes more sense given history, that if the teams played 100 times, Riverside would win a few. I don't know why that is, possibly that nothing resets after a 1st set, if a team wins 1st set, and is playing well .....who knows. Or match predictions are better based on points, vs. sets. & Massey uses only set scores, i.e. 3-1, 3-0, 3-2 (not points, like Pablo, I believe)

2) There are two ratings, the standard, and power, and power is what 'seems' IMO to be more accurate and I believe is based on factoring sets won vs. just match scores.

again, one would have to ask Massey. His rating is founded in using statisical techniques unlike RPI, not some 25%/50%/25% forced math function. His tool I believe factors in opponnents in ordering teams in ranking.


just because RPI is contrived, it's hard to take it too seriously. Others may have more insight. Anyone can contact him via email to ask.

[The Bofa on the Sofa]

Nov 15, 2024 14:04:34 GMT -5 mikegarrison said:

Proprietary.

Even so, it's not a "formula." It's a procedure. There is certainly an underlying w/l model that underlies it, and there are quite a few foundations out there, but in the end, calculating Massey rankings is not a "formula."

When people have asked me about the "formula" for Pablo in the past, I usually have just ignored them because they are so far off the rails to even start. Nothing personal, there, but it's just not that simple.

I spent time this summer/fall helping hand the Pablo reins over, and in the end I created something like a 30 page document. That's a long formula! And that was working with someone who knows their stuff and was really motivated to figure it out.

Massey isn't going to be all that much easier.

[Kearney Kingston]

Pablo, I love your formula. Which is the ideal ranking or rating for the NCAA to use - or should they combine several respected systems to use as their guideline?

[staticb]

Nov 15, 2024 13:27:09 GMT -5 Kearney Kingston said:

There seems to be a big difference between RPI and Massey Rankings, yet, both are mathematically calculated by results of each team and success of opponents. I know the 25% / 50% / 25% formula of the RPI, but what is Massey’s mathematical make-up?. Quite honestly, over the years, Massey seems more accurate.

Doesn't someone use the all the various systems to enter the pick'em contest over the years?  Which one comes on top most often?

[bluepenquin]

Nov 16, 2024 21:59:59 GMT -5 staticb said:

Nov 15, 2024 13:27:09 GMT -5 Kearney Kingston said:

There seems to be a big difference between RPI and Massey Rankings, yet, both are mathematically calculated by results of each team and success of opponents. I know the 25% / 50% / 25% formula of the RPI, but what is Massey’s mathematical make-up?. Quite honestly, over the years, Massey seems more accurate.

Doesn't someone use the all the various systems to enter the pick'em contest over the years?  Which one comes on top most often?

Yes.  But the problem with pick-em contest is the sample is the most 50-50 type matches and high profile teams.  Everyone considers RPI to be inferior - but for these kinds of selections RPI isn't really going to be that far off.  Start playing matches between mid level B1G teams against decent (good record) mid majors - and RPI will get destroyed.