Don't follow Massey so I am not going to say anything. However, I will repeat some MWC poster comments and hopefully those posters are correct. All info taken off MWC Board/Thread.

1. Massey Ratings are still heavily influenced by last year.

2. MWC dropped 3 points from the previous week. AAC dropped 5 points. SBC gained 1 point and MAC dropped by 2 or 3 points. Big Ten dropped by about 5 points.

About a 30 point difference between BCS conferences and NonBCS conferences.

1. SEC 30.27
2. P12 38.73
3. B12 45.84
4. ACC 46.64
5. B10 51.36
6. AAC 81.14
7. MWC 83.74
8. CUSA 87.50
9. MAC 94.24
10.SBC 99.16

The Massey Ratings are very useful if understood properly. One thing that is helpful is to look at the Median ratings and not the Average ratings, which give equal weight to every pollster, including the less-reputable or less-accurate ones.

(09-09-2014 10:41 AM)Native Georgian Wrote:  The Massey Ratings are very useful if understood properly.

IMO, any computer that factors in last year is useless for this year. Which means they are all useless until about week six or so.

(09-09-2014 11:05 AM)quo vadis Wrote:  

(09-09-2014 10:41 AM)Native Georgian Wrote:  The Massey Ratings are very useful if understood properly.

IMO, any computer that factors in last year is useless for this year. Which means they are all useless until about week six or so.

Because human pollsters do not factor in previous years or a schools general reputation. As long as the programmer has taken steps to account for returning talent, there is no reason that previous season results given less weight are still not part of the equation. It is imprecise to be sure, but no less so than a human poll voter looking at team ABC and saying oh they returned 17 starters from a 9-3 team they are going to be pretty good.

(09-09-2014 11:22 AM)sundodger Wrote:  

(09-09-2014 11:05 AM)quo vadis Wrote:  

(09-09-2014 10:41 AM)Native Georgian Wrote:  The Massey Ratings are very useful if understood properly.

IMO, any computer that factors in last year is useless for this year. Which means they are all useless until about week six or so.

Because human pollsters do not factor in previous years or a schools general reputation. As long as the programmer has taken steps to account for returning talent, there is no reason that previous season results given less weight are still not part of the equation. It is imprecise to be sure, but no less so than a human poll voter looking at team ABC and saying oh they returned 17 starters from a 9-3 team they are going to be pretty good.

I don't think they factor in returning players. And this is college, not the pros, where you turnover and develop players on a lot quicker pace.

(09-09-2014 11:22 AM)sundodger Wrote:  

(09-09-2014 11:05 AM)quo vadis Wrote:  

(09-09-2014 10:41 AM)Native Georgian Wrote:  The Massey Ratings are very useful if understood properly.

IMO, any computer that factors in last year is useless for this year. Which means they are all useless until about week six or so.

Because human pollsters do not factor in previous years or a schools general reputation.

I don't think much (anything) of human pollsters the first several weeks either, for the same reason.

(09-09-2014 11:29 AM)bullet Wrote:  I don't think they factor in returning players. And this is college, not the pros, where you turnover and develop players on a lot quicker pace.

Who is the they you refer to in you sentence, the programmers, or human poll voters?

It seems pretty clear to me that human poll voters look at both returning strength and the overall historical strength of the program to determine pre-season/early season rankings. I actually like the idea of the Harris poll for not coming out until games are played, as there would appear to be less pre-season bias to overcome. Unfortunately even that suffers to some degree.

My point was that if the programmers were to factor in returning talent it would not be that different than what we ouirselves already do. And that weighing last years games less than this year helps to overcome the fact that it was last season. Failing to account for returning talent, and providing equal weigh to last season is a problem and will skew the numbers in favor of those teams that had better results last season. Not being any of the computer poll programmers I cannot speak for what they do, but I would expect them to do one or both of those things to correct their numbers, if they are using last season as part of their factors.

(09-09-2014 11:29 AM)bullet Wrote:  

(09-09-2014 11:22 AM)sundodger Wrote:  

(09-09-2014 11:05 AM)quo vadis Wrote:  

(09-09-2014 10:41 AM)Native Georgian Wrote:  The Massey Ratings are very useful if understood properly.

IMO, any computer that factors in last year is useless for this year. Which means they are all useless until about week six or so.

Because human pollsters do not factor in previous years or a schools general reputation. As long as the programmer has taken steps to account for returning talent, there is no reason that previous season results given less weight are still not part of the equation. It is imprecise to be sure, but no less so than a human poll voter looking at team ABC and saying oh they returned 17 starters from a 9-3 team they are going to be pretty good.

I don't think they factor in returning players. And this is college, not the pros, where you turnover and develop players on a lot quicker pace.

Poll voters factor in returning players - eg, they wouldn't have made FSU preseason #1 if Winston was gone and an untested player was starting at QB - but they don't do it thoroughly, they do it more casually as suggested in Sundodger's comment.