Monday, November 7
Under Review: The Colley Matrix
More computer weirdness for you. Last week I detailed some of the bizarre rankings found in Jeff Sagarin’s computer rankings. This week, I’ll take a quick and dirty look at the Colley Matrix, another computer ranking system used by the BCS.
Unlike Sagarin, Colley doesn’t offer two sets of rankings, one with MOV and one without. He only offers one formula, and it does not include margin of victory as a ranking factor.
Colley’s rankings for this week were just released, and the top 10 might surprise you a little bit.
I find it rather bizarre that a computer system in which only winning and losing matters can have a one-loss Penn State team ahead of both Alabama and USC. It’s simply confusing. The Big 10 is clearly ranked as the big time heavyweight conference, which would also explain the bizarre showing of Wisconsin in the top 10. But then again, Oregon at #6 surely means the Pac 10 is ranked pretty decently, too. If so, why is USC not ahead of Penn State? And hey, isn’t the Big 10 0-2 against the Big 12?
I don’t have any profound conclusions to offer here. I’ll just say that it’s a rather wacky top 10. Wisconsin has no business in there, and USC should be no worse than #2. I’m all for objective measurements of teams, but it sure looks like these formulas are producing some wacky results. The human voters, I’m beginning to think, may be the more fair and reliable source for ranking teams.
I don’t want to dismiss computer formulas and conclude they're worthless. But my early, rudimentary examinations of them have left me scratching my head.
Back to Hornsblog Home Page.
--PB--
Unlike Sagarin, Colley doesn’t offer two sets of rankings, one with MOV and one without. He only offers one formula, and it does not include margin of victory as a ranking factor.
Colley’s rankings for this week were just released, and the top 10 might surprise you a little bit.
| Rank | Team |
|---|---|
| 1 | Texas |
| 2 | Penn State |
| 3 | USC |
| 4 | Va Tech |
| 5 | Bama |
| 6 | Oregon |
| 7 | Miami |
| 8 | Ohio St |
| 9 | Wisconsin |
| 10 | Texas Tech |
I find it rather bizarre that a computer system in which only winning and losing matters can have a one-loss Penn State team ahead of both Alabama and USC. It’s simply confusing. The Big 10 is clearly ranked as the big time heavyweight conference, which would also explain the bizarre showing of Wisconsin in the top 10. But then again, Oregon at #6 surely means the Pac 10 is ranked pretty decently, too. If so, why is USC not ahead of Penn State? And hey, isn’t the Big 10 0-2 against the Big 12?
I don’t have any profound conclusions to offer here. I’ll just say that it’s a rather wacky top 10. Wisconsin has no business in there, and USC should be no worse than #2. I’m all for objective measurements of teams, but it sure looks like these formulas are producing some wacky results. The human voters, I’m beginning to think, may be the more fair and reliable source for ranking teams.
I don’t want to dismiss computer formulas and conclude they're worthless. But my early, rudimentary examinations of them have left me scratching my head.
Back to Hornsblog Home Page.
--PB--
Comments:
<< Home |
Yeah, it is weird. I would agree with the top 8 of those teams being in the top 10, but of course, not in that order. Kind of strange that a team that beat Va Tech is 4 spots below it.
The biggest problem with computer rankings isn't MOV, its that it can't properly address real-world discrepencies.
A great example is how injuries affect the national title picture. ASU and Georgia were excellent teams in the first half of the season. They're not after losing their no. 1 QBs.
Naturally those should be judged on how they respond, but what about their opponents? Should LSU and USC be docked points because they beat a strong ASU team, that lost a bunch of games without their QB after the fact? Should Florida be rewarded for winning against a good, but toothless Georgia team?
Computers should be involved in the process, but in a consulting capacity. Much like the basketball and baseball committees consult the RPI (ignoring for the moment that the baseball formula is ridiculously out of whack).
A great example is how injuries affect the national title picture. ASU and Georgia were excellent teams in the first half of the season. They're not after losing their no. 1 QBs.
Naturally those should be judged on how they respond, but what about their opponents? Should LSU and USC be docked points because they beat a strong ASU team, that lost a bunch of games without their QB after the fact? Should Florida be rewarded for winning against a good, but toothless Georgia team?
Computers should be involved in the process, but in a consulting capacity. Much like the basketball and baseball committees consult the RPI (ignoring for the moment that the baseball formula is ridiculously out of whack).
# posted by 
: 11/08/2005 2:38 PM
: 11/08/2005 2:38 PM
I think that's a fair assessment. Concordingly, the BCS uses a two part human poll, one part computer poll combination to get their final ranking. It's logical and, I think, reasonably fair.
A selection committee for football would make sense if you had a playoff format, but I haven't given it much thought.
--PB--
A selection committee for football would make sense if you had a playoff format, but I haven't given it much thought.
--PB--
Colley's matrix method for ranking college football teams is explained in detail, with many examples and explicit derivations. The method is based on very simple statistical principles, and uses only Div. I-A wins and losses as input --- margin of victory does not matter. The scheme adjusts effectively for strength of schedule, in a way that is free of bias toward conference, tradition, or region. Comparison of rankings produced by this method to those produced by the press polls shows that despite its simplicity, the scheme produces common sense results.
That is from Colley.. Penn State has a strength of schedule near the top 10 in the nation... they are ranked higher than any of the top 25 BCS teams and out rank USC substantially, it is because of the quality of opponents PSU beat that gives them the slight edge over USC and that is why a team like Wisconsin can make the top ten.. research his site more.. as for the 0-2 Big 10 vs Big 12 Iowa St beat an Iowa team that started off slow and the Texas/OSU game was a classic so its not like the Big 12 owns the Big 10
That is from Colley.. Penn State has a strength of schedule near the top 10 in the nation... they are ranked higher than any of the top 25 BCS teams and out rank USC substantially, it is because of the quality of opponents PSU beat that gives them the slight edge over USC and that is why a team like Wisconsin can make the top ten.. research his site more.. as for the 0-2 Big 10 vs Big 12 Iowa St beat an Iowa team that started off slow and the Texas/OSU game was a classic so its not like the Big 12 owns the Big 10
# posted by : 11/09/2005 10:50 PM
: 11/09/2005 10:50 PM
The rankings are "wierd" because the computer program is set up to judge what has already happend to a team, and the entire season has yet to happen. Once the entire season has happend, the teams in the same conference will start to be more appropriatly favored. If USC, Texas and Penn State all win out, I'm pretty confident that USC would move ahead of Penn State. The whole problem is that everyone worries about the in the middle stuff with these systems.
Post a Comment
# posted by : 12/01/2005 1:59 AM
: 12/01/2005 1:59 AM << Home |
