Re: Praying for a sequence...will get Donovan to rant
Posted: Sun Nov 26, 2017 9:06 am
true Cane...
But one has to explain by just what parameters a UCF would be picked over a one loss Bama, a one loss Miami, a two loss Auburn, etc. etc. And, to me, that means more than just, "well they are undefeated". Some thing more than Eric's "I think that they are better than you think" because you don't give G5 wins credit".
Parameters like good opponents beat...good opponents opponents that were beaten..etc.
A reason why the computers have them in spot #11 is that those qualifiers aren't as good as some others ranked higher.
The computers don't have a conference loyalty or bias, a regional bias...they see pure algorithms...wins, losses, opponents wins and losses, etc.
Most algorithms start with wins and losses straight up....simple winning percentage. The rub always comes in qualifying those wins vs a strength of schedule...and different algorithms do that in different ways...
While many computer rankings methodology are kept as a proprietary secret, Colley lets you see inside his...and I assume many share aspects of his method of providing an iterative correction for strength of schedule.
Now...my own internal off of the cuff method is "what team would I think would win in a match up and by how much"...
We have historic data on that...
In college football...a team favored by 7 points has won 70.3 % of the time (a huge indicator)
A team favored by 3 points has won 57.4% ( much closer to 50-50)
Soooo...if I think that a team would be favored by 7 over UCF...I put them over UCF. No exact science since it is only my opinion.
But one has to explain by just what parameters a UCF would be picked over a one loss Bama, a one loss Miami, a two loss Auburn, etc. etc. And, to me, that means more than just, "well they are undefeated". Some thing more than Eric's "I think that they are better than you think" because you don't give G5 wins credit".
Parameters like good opponents beat...good opponents opponents that were beaten..etc.
A reason why the computers have them in spot #11 is that those qualifiers aren't as good as some others ranked higher.
The computers don't have a conference loyalty or bias, a regional bias...they see pure algorithms...wins, losses, opponents wins and losses, etc.
Most algorithms start with wins and losses straight up....simple winning percentage. The rub always comes in qualifying those wins vs a strength of schedule...and different algorithms do that in different ways...
While many computer rankings methodology are kept as a proprietary secret, Colley lets you see inside his...and I assume many share aspects of his method of providing an iterative correction for strength of schedule.
Now...my own internal off of the cuff method is "what team would I think would win in a match up and by how much"...
We have historic data on that...
In college football...a team favored by 7 points has won 70.3 % of the time (a huge indicator)
A team favored by 3 points has won 57.4% ( much closer to 50-50)
Soooo...if I think that a team would be favored by 7 over UCF...I put them over UCF. No exact science since it is only my opinion.