Anyone ever pay attention to the computer rankings in the BCS?

[mstatefan88]

It seems like some of these polls are based purely on SOS and nothign else. Sagarin has Auburn ranked 9th, ahead of Virginia Tech, Oklahoma, and Oregon. Another poll has Auburn receiving no points for the Top 25.Sagarin also has undefeated Stanford ranked 21st, behind teams like Texas Tech, Penn State, and Arizona St. I know they haven't played anyone great, but come on. Behind Arizona St?

Almost every poll has Wisconsin ranked behind Auburn, Houston, Penn State, and Georgia. I know they lost a heartbreaker to Michigan St, but Wisconsin is a good football team, and Iwould put money on them to beat all 4 of those teams easily.

I don't understand how some of these polls can come to these conclusions. I think the BCS is pure garbage, and any poll or computer system that has Auburn ranked 9th right now should be shot. They have 3 losses, and even though they are to good teams, there is no way in hell Auburn would touch Wisconsin or Oklahoma.I just don't think anything will ever be better than passing the eye test.

 

[QuaoarsKing]

Only 1 of the 6 is run by a person with a mathematical/statistical background, and it's the only 1 of the 6 that has a public formula.<div>Sure enough, that one had a mistake in last year's standings that was caught by an observant fan. Who knows if the other 5 have had unchecked mistakes.</div><div>
</div><div>I used to think the computer polls were cool, but then I read the chapter of "Death to the BCS" that completely flipped my opinion.</div><div>
</div><div>
</div><div>ETA this Dan Wetzel articlehttp://www.nytimes.com/2010/10/17/sports/ncaafootball/17score.html</div>

 

[Incognegro]

I never read anything that makes the BCS system actually sound like it's a good idea. How can something thatblatantlyhave so many errors and bias still be acceptable to date?

 

[RebelBruiser]

Computer polls and really any mathematical ranking, which is all a computer poll is, need a significant amount of data to make them reasonable measures.

Think of them as comparable to the RPI. That's all they are with different factors placed on wins and losses, schedule, etc. Think of what the RPI looks like about 10 games into the season every year. It's a joke. You have teams like SMU ranked in the Top 10, and other crazy stuff. It's still a really wacky looking poll about 1/4 way through conference play. The RPI in mid-January is about as accurate as any postseason computer football poll could ever be.

It's ridiculous that they're even included in the BCS, much less for 1/3 of the vote.

Human polls with a lot of voters involved are by far the best way to rank football teams, given the sample size. Granted, there are inherent flaws with the human polls we use. We use a coach's poll that relies on coaches who only see 2 teams every weekend. Still, even with those flaws, the human polls have always been the ones that looked right in every BCS controversy. The computers have always been the ones that messed it up.

 

[Incognegro]

I think the computer rankings could be done properly but the way the BCS is set up is absolutely ridiculous. I didn't know they weren't even allowed to take into account of point differential in games. I'm not going to claim to be an expert mathematician, but the little bit I do know you can't just leave things out like that. You also have to have some equations that factor in proponents from the immediate season prior as far as their average points scored/allowed and their performance against teams of particular ranks and positional ranks, and even take into account how many players they're going to lose and gain and what those new additions/subtraction will project for their team performances. You can't have this work properly using "simple" equations.<div>
</div><div>You'd also have to have some recursive functions that will change particular teams rankings when we come to find that the teams that they played either weren't as good as we once thought or much better than what we figured. Even if we don't have the BCS in the next couple of years, I wouldn't mind seeing computers factor into the rankings, but only if they are at least done by people who really give a ****. This right here is stupid.</div>

 

[patdog]

(or basketball for that matter). But too many computer polls are Garbage In Garbage Out. Has anyone actually looked at these polls and determined how accurate they are at predicting games? That's the real test of whether any poll is worth anything or not. And just arbitrarily deciding that the computer polls can't take margin of victory into account at all is asinine. I can guarantee you the human polls take that into account. If they're going touse computer polls in the BCS, they should not put any constraints on what factors the computers can consider, then pick 1 or 2 based on which ones are the most accurate at predicting games.

 

[dawgs.sixpack]

at least the best and worst computer ranking for ech team is tossed out for the BCS rankings, so stanford's 21 spot and auburn's 9 spots are both knocked out.

that said, people complained about margin of victory factoring too much into the rankings. i understand that, you don't necessarily won't to encourage a team throwing TD passes with starters in the game in a 50 point W. however, the answer isn't to completely strike out margin of victory, but to limit it. say any W by 28+ points merely is recorded as a 28 point W for computer ranking purposes. it allows enough of a margin of victory factor to split apart a poor showing from a dominant showing, but without motivating coaches to win my 52 instead of settling for the 45 point W.

i've made a similar argument regarding SoS in both the RPI for CBB and the computer rankings for CFB. to many team have better SoS numbers because they played a somewhat less ****** ****** team. the RPI shouldn't come down to which CBB program played the 250th rated team in stead fo the 286th rated team. likewise, the BCS shouldn't come down to who played the 88th best team instead of the 101st best team. a ****** team is a ****** team. in CBB, i think any team rated below 200 should count equally against the schedule. in the computer rankings, i think any team rated below 75 or 80 should count equally against the computer rankings. if you are in a position where your computer rankings matter, you should beat every team by 28+ ranked below that threshhold and splitting hairs over which ****** team is more or less ****** is a terrible way to go about determining who deserves to play for a national title or in a BCS bowl.

 

[dogmatic1]

Div. II and III have had playoffs in place for years, so there's no legitimate argument for Div. I not having it other than preserving bowl money and the relationships that money secures, yet not one soul connected to the BCS has the guts to say so. Using one poll would be ********, they say, then tell us taking the average of 6 polls provides chicken salad, not simply average ********. The BCS ranking would be a great way to seed a playoff, but for picking a one-off championship game, it or any other formula is just ********.

 

[Incognegro]

Not factoring in margin of victory hasn't stopped teams from running the scores up 60+. You'll find that a majority of the coaches have a philosophy of "You put your defense on the field to stop my offense. I won't stop them for you." Factoring in margin of victory is not going to stop this philosophy. If you have coaches that leave their starters in a blowout, then you have an idiot for a coach since he's more worried about ranking than the possibility of his starters getting hurt.<div>
</div><div>The main reason I think margin of victory is really important is because it helps determine how much better a team performed against another. If LSU beats Bama in a couple of weeks 7-6 then in the computer rankings, Bama probably won't drop too far down, but let's just say hypothetically that Bama beats LSU 70 - 6, still... in the computer polls, LSU won't drop that far down but clearly, when we're seeing this, we see that there's a huge gap between the 2 teams. I just think if you're going to do margin of victory, you need to do all or nothing.</div>

 

[sleepy dawg]

I think it is really too early for to be reading into the computer polls. The computer rankings always appear ridiculous this time of year. Wait until the end of the season, and they will appear much more realistic. At this point in the season, teams still have a good chunk of their tough games left (especially sec teams), with most easy games out of the way. A teams true strength of schedule hasn't had a chance to kick in yet. For example, Alabama and LSU's SOS should shoot way up once they play.

Every year people complain about how ****** the BCS is around this time of the year. However, once the time comes to actually use the BCS, it is usually a pretty accurate indicator of the most deserving teams.