Does anyone know why Calpreps and Massey computer rankings differ so much?

May 18, 2015
1,666
845
0
I have looked at these two for years and still don't understand how a team can be slotted so low in one compared to the other.

Does anyone know a good shorthand way to explain the difference?
 

jwarigaku

All-Conference
Jan 30, 2006
4,199
1,557
73
Cap'n

If you are looking for a serious answer I can refer you to both their explanations. Calpreps uses Freeman to do the math and I my opinion this mathmatically really represents a true linear algebraic approach to the problem because they feel it should be completely non subjective. To Epics point this is or can be a bad approach. In essence Calpreps uses no human input after I think week 4( they do take into account player graduation, returning players etc if the coaches respond to their request for information and/or otherwise guess).

Massey is a much more extreme treatment of data applying linear algrebra but then combining minimization techniques to account for anomaliestic behavior in the mathmatical solution which yield unexpected results as the season progresses. The mathmatical treatment of data for both systems is attached below. I use a lot of this crap daily in my work but it is still a lot of hand waving and assumptions in my opinion. From a mathmatical standpoint Massey is a friggin genius.

http://www.masseyratings.com/theory/massey97.pdf

http://calpreps.com/ratings.htm


I have looked at these two for years and still don't understand how a team can be slotted so low in one compared to the other.

Does anyone know a good shorthand way to explain the difference?
 

Penlight

All-Conference
May 29, 2001
2,089
1,039
113
Cap'n

If you are looking for a serious answer I can refer you to both their explanations. Calpreps uses Freeman to do the math and I my opinion this mathmatically really represents a true linear algebraic approach to the problem because they feel it should be completely non subjective. To Epics point this is or can be a bad approach. In essence Calpreps uses no human input after I think week 4( they do take into account player graduation, returning players etc if the coaches respond to their request for information and/or otherwise guess).

Massey is a much more extreme treatment of data applying linear algrebra but then combining minimization techniques to account for anomaliestic behavior in the mathmatical solution which yield unexpected results as the season progresses. The mathmatical treatment of data for both systems is attached below. I use a lot of this crap daily in my work but it is still a lot of hand waving and assumptions in my opinion. From a mathmatical standpoint Massey is a friggin genius.

http://www.masseyratings.com/theory/massey97.pdf

http://calpreps.com/ratings.htm
I must confess, I was day dreaming the day that they taught stuff.
 
May 18, 2015
1,666
845
0
Thanks jwar...I have some reading ahead of me tonight. Love to geek out on statistical modeling.

Not sure if you are a fan of HBO comedy Silicon Valley, but Season 1 ended on the single funniest discussion of mathematical theory I have ever witnessed. Not exactly for all audiences due to the topic matter, but a Stanford guy wrote the white paper that explains the equation. The link below takes you to the white paper. Killer math comedy.
http://gizmodo.com/an-actual-mathematical-explanation-of-silicon-valleys-e-1589569331

Cap'n

If you are looking for a serious answer I can refer you to both their explanations. Calpreps uses Freeman to do the math and I my opinion this mathmatically really represents a true linear algebraic approach to the problem because they feel it should be completely non subjective. To Epics point this is or can be a bad approach. In essence Calpreps uses no human input after I think week 4( they do take into account player graduation, returning players etc if the coaches respond to their request for information and/or otherwise guess).

Massey is a much more extreme treatment of data applying linear algrebra but then combining minimization techniques to account for anomaliestic behavior in the mathmatical solution which yield unexpected results as the season progresses. The mathmatical treatment of data for both systems is attached below. I use a lot of this crap daily in my work but it is still a lot of hand waving and assumptions in my opinion. From a mathmatical standpoint Massey is a friggin genius.

http://www.masseyratings.com/theory/massey97.pdf

http://calpreps.com/ratings.htm