BCS

[Raven_tiger 5]

Linky

[tgr4lfe 220]

So, do you think the computers will fry when AUBURN beats 3 top 10 teams over the next 5 weeks???

[LegalEagle 22]

We're 0.3210 and LSU is 0.7078

Why do we even bother to go? Just forfeit I say.

Sadly, when we beat them, the BCS will say LSU lost to a nobody and therefore was overrated - result? We'll stay at 0.3210 and LSU will be dropped to 0.300!

[GalensGhost 0]

We're 0.3210 and LSU is 0.7078

Why do we even bother to go?  Just forfeit I say.

Sadly, when we beat them, the BCS will say LSU lost to a nobody and therefore was overrated - result?  We'll stay at 0.3210 and LSU will be dropped to 0.300!

189834[/snapback]

What I find intriguing is the computer's disdain for Notre Dame. Average computer ranking? 25. Harris/USA ranking? 11th.

Amazing.

[runswithscissors 33]

We're 0.3210 and LSU is 0.7078

Why do we even bother to go?  Just forfeit I say.

Sadly, when we beat them, the BCS will say LSU lost to a nobody and therefore was overrated - result?  We'll stay at 0.3210 and LSU will be dropped to 0.300!

189834[/snapback]

i think you're right about the BCS deal.... low BCS percentage is due partly to the strength of schedule i think.

[mcgufcm 4,107]

biggest joke of the poll? YOOOOOUR penn state nittany lions fresh off a loss to two-loss ohio state and STILL in the top ten.

[tgr4lfe 220]

The computer rankings are crap...Sagarin is an idiot plain and simple. Everybody knows that computers can be told what to output. They all rate SOS differently. Some may calculate (or whatever the crap they call it) SOS based on what the ranking of the team was when played and go week to week, others may do it different to where the SOS changes based on how that team does each week. For instance, with with the Bama win over UF. At the time UF was ranked #5. Now one ranking will lock in that for the whole year. It doesn't matter what UF does from here on out, they were #5 when played. Others will change based on UFs performance and rankings from here on out. If UF wins out things get better for bama whereas if they lose things get worse.

They need have one, if they are going to use, which I think is a total waste of time. It already tells us what we already know b/c it uses the same info we already have, the human polls adn some weird made up formulas.

[tgr4lfe 220]

Peter Wolfe ratings

You got to read this if you haven't!!! This is Wolfe's explanation for the rating system. I willpose questions below after you read.

lot has been written on how to rate football teams.   This is not surprising, because a good rating system has applications far beyond sports.  (A good bibliography is kept by David Wilson here.)   

Sports like college football with short seasons and many teams are much harder to rate than sports with many games and few teams.  If teams in a league play a balanced schedule, then winning percentage is all you need to rank the teams.  College football schedules are far from balanced, though.  About 700 college football teams can be linked by schedules, but because each team plays only about eleven games, there is no direct way to compare each team with the vast majority of all the other teams.  Using winning percentage as the benchmark for comparison doesn't work.  (Was 2001 division III champion Mount Union at 14-0 as good as 2001 division I-A champion Miami FL, 13-0?  Each had a perfect winning percentage.) 

A significant but hard-to-measure factor in comparing teams is sportsmanship.  Running up the score is generally looked on as evidence of bad sportsmanship, behavior which should not be encouraged or rewarded.  With this in mind, the BCS has chosen computer systems that use only won/loss data (and not scoring margin) to compute ratings.  We have developed such a system that provides reasonable results.  We rate all varsity teams of four year colleges that can be connected by mutual opponents.  Games played against club teams, junior varsity teams, or junior or community college teams aren't counted, but game location is taken into account.

The method we use is called a maximum likelihood estimate.  In it, each team i is assigned a rating value πi that is used in predicting the expected result between it and its opponent j, with the likelihood of i beating j given by:

πi / (πi + πj)

The probability P of all the results happening as they actually did is simply the product of multiplying together all the individual probabilities derived from each game.  The rating values are chosen in such a way that the number P is as large as possible.  This is often called a Bradley-Terry model, and is described in papers listed at Wilson's site (see Bradley and Terry 1952, Ford 1957, Elo 1986, Keener 1993). 

All I say is I went thru 3 years of Engineering at AU so I have a halfway understanding of the formula itself, BUT wth...now where does he say how each team is given their rating value?? Predicting the expected result? I also noticed this guy is from UCLA....hmmmmm!

Another reason this is a stupid a$$ system.

[BigSammyK 171]

biggest joke of the poll? YOOOOOUR penn state nittany lions fresh off a loss to two-loss ohio state and STILL in the top ten.

189846[/snapback]

Actually they beat the Buckeyes and lost to Michigan. Go up north and confuse the two and you might find yourself in some trouble.

[runswithscissors 33]

the only real way to do it is go to a playoff style system to see who gets to play for the championship. i don't think anyone will argue that the BCS system makes sense. until that happens, there will always be a problem like there was with last year. it may very well happen this year with USC/VATech/Texas.

[SouthLink02 8]

I am surprised VATech and Texas are so far apart

[Streyeder 0]

Peter Wolfe ratings

You got to read this if you haven't!!!  This is Wolfe's explanation for the rating system.  I willpose questions below after you read.

lot has been written on how to rate football teams.  This is not surprising, because a good rating system has applications far beyond sports.  (A good bibliography is kept by David Wilson here.)   

Sports like college football with short seasons and many teams are much harder to rate than sports with many games and few teams.  If teams in a league play a balanced schedule, then winning percentage is all you need to rank the teams.  College football schedules are far from balanced, though.  About 700 college football teams can be linked by schedules, but because each team plays only about eleven games, there is no direct way to compare each team with the vast majority of all the other teams.  Using winning percentage as the benchmark for comparison doesn't work.  (Was 2001 division III champion Mount Union at 14-0 as good as 2001 division I-A champion Miami FL, 13-0?  Each had a perfect winning percentage.) 

A significant but hard-to-measure factor in comparing teams is sportsmanship.  Running up the score is generally looked on as evidence of bad sportsmanship, behavior which should not be encouraged or rewarded.  With this in mind, the BCS has chosen computer systems that use only won/loss data (and not scoring margin) to compute ratings.  We have developed such a system that provides reasonable results.  We rate all varsity teams of four year colleges that can be connected by mutual opponents.  Games played against club teams, junior varsity teams, or junior or community college teams aren't counted, but game location is taken into account.

The method we use is called a maximum likelihood estimate.  In it, each team i is assigned a rating value πi that is used in predicting the expected result between it and its opponent j, with the likelihood of i beating j given by:

πi / (πi + πj)

The probability P of all the results happening as they actually did is simply the product of multiplying together all the individual probabilities derived from each game.  The rating values are chosen in such a way that the number P is as large as possible.  This is often called a Bradley-Terry model, and is described in papers listed at Wilson's site (see Bradley and Terry 1952, Ford 1957, Elo 1986, Keener 1993). 

All I say is I went thru 3 years of Engineering at AU so I have a halfway understanding of the formula itself, BUT wth...now where does he say how each team is given their rating value?? Predicting the expected result? I also noticed this guy is from UCLA....hmmmmm!

Another reason this is a stupid a$$ system.

189855[/snapback]

Sounds like you need to go pull out that old Random book.