College Football Talk

Commentary, Picks of the Week, Polls and More…

BCS Computers place far too much on SOS

with 2 comments

BCS computers are not meant to evaluate teams based on how they play.  The computers eliminate human bias by using cold calculation and the human polls add the element of actual evaluation by being able to watch teams play and determine whether Florida beating Charleston Southern by 59 is of equal or greater weight to Alabama beating LSU by 9.  It’s a balance that many, including myself, have defended for years.

I can say definitively that I’m done defending the computers.

Of course computers who don’t watch games aren’t able to evaluate the talent on the field.  But what we’re seeing this year is a ridiculous bias towards strength of schedule that seems to put more weight on playing a tough schedule regardless of the outcome than on actually winning.

Don’t get me wrong – strength of schedule definitely has its place.  A team that plays a very hard schedule and wins some close ones deserves, in my opinion, more consideration than a team that plays a bad schedule and crushes everyone.  Unfortunately, what we’re seeing this season is a system that rewards teams that play a tough schedule even if they lose.

What brought this so strongly to my attention was this week’s BCS Rankings.  You may have noticed that after losing to No. 2 Alabama, LSU actually moved up one spot from No. 9 to No. 8.  While they got some help from two top-10 teams losing to unranked opponents, the biggest thing that helped them was the computers, which moved them from No. 9 to No. 8.  They  moved up in two computers, stayed the same in three computers and only dropped in one.

Can anyone honestly say that a team deserves to move up after a loss, regardless of who they play?

And LSU is only the latest example of a team that’s benefited from playing a tough schedule regardless of the actual outcome.  The computers place two teams (LSU and Oregon) who have two losses in the top ten.  USC is just outside at No. 11.  Just behind USC is Miami and then Ohio State, also two-loss teams.  This is where these two-loss teams stand in terms of strength of schedule (according to Jeff Sagarain):

LSU 17
Oregon 6
USC 7
Miami 11
OSU 46


While it’s commendable that they’re out playing tough schedules, in some instances, their rankings seem to ignore whether or not they’re actually beating the tough teams they play.  Their records against top-30 opponents:

Team Wins Losses
LSU 0 2
Oregon 3 2
USC 3 1
Miami 2 2
OSU 2 1


LSU, the highest-ranked team among them, has played the 17-toughest schedule in the country but has yet to defeat a top-30 team.  Houston, on the other hand, is 8-1 against the 94th-toughest schedule in the country.  That includes a 1-0 record against top-30 teams.  Yet, computers have them at No. 16, a full eight spots behind LSU.  They have a better record than LSU and more wins against top-30 teams, yet they are eight spots lower.

Jeff Sagarin’s rankings, which place particular emphasis on schedule, has the biggest disparity between the two teams’ rankings – a 13-spot gap between No. 6 LSU and No. 19 Houston.

Does it seem like there might be a problem with that calculation?

In Sagarin’s power rankings, Houston sits behind two teams that have worse records and no wins against top-30 opponents, and one of which (South Florida) has played an even weaker schedule.

Sagarin’s computer even has LSU ahead of Texas, which is undefeated and 2-0 against top-30 opponents with the 52nd-toughest schedule in the country.  I’ll grant that Texas should have played a harder schedule, but what on earth is Sagarin calculating that he has an undefeated team with two wins over top-30 teams behind a team that’s 7-2 with no wins over top-30 teams?

Now, I’m not enough of a mathematician to figure out what is wrong with the computers.  But clearly, there is something horribly wrong when the computers think the team that all the humans say is No. 2 is actually No. 5.  The same computers that had Iowa as the No. 1 two weeks ago and the No. 2 last week.

Of course it can’t happen mid-season, but at the end of this year, the BCS Committee needs to take a serious look reforming the calculations being used by the computers and see if there can be a more accurate accounting of the talent of the teams.  They should be rewarded for the games they’re winning, not the games they’re playing and losing.

Advertisements

Written by Austin Swafford

November 11, 2009 at 3:05 pm

Posted in Commentary, Poll Talk

2 Responses

Subscribe to comments with RSS.

  1. Let me start by saying that I also (as I have made clear already) have no respect for the BCS. The computer polls are certainly a large part of my dislike, but probably the least of my concerns as far as problems with the BCS go. One thing you didn’t mention that is worth knowing is that a number of the computer polls use as part of their system a strength-of-conference weight as well. Try convincing me that is any good.

    However, I can honestly say that a team MAY deserve to move up after a loss, regardless of who they played. MAY.

    LSU is probably the only two-loss team with a good argument (not a perfect argument) for being where they are… instead of saying they have no wins against top-30 opponents, you could say they haven’t lost to anyone who isn’t top-3. They are further elevated by the strength-of-conference factors in some of those computers (which have the PAC-10 first, followed by the SEC). Houston’s big problem is they’ve only played one top-30 team. Sure, they beat Oklahoma State, but strength-of-schedule measurements make a big deal over the quantity of top teams played, not just the record against them. That’s why you see teams like Oregon, USC, Virginia Tech, and even Clemson ranked where they are… they’ve all faced four or more top-30 teams. Most of the other highly ranked teams have faced only two or three.

    Watch what happens after Iowa plays Ohio State this weekend… it won’t matter which team wins or loses, the computers will move the winner up significantly and the loser will not fall all that much.

    There is obviously some limit to rewarding teams for wins. I don’t think many college football fans are gonna concede an undefeated Boise State is better than, say, a one-loss SEC runner up. Some will, sure, but not most by any stretch. That’s why I said that more and more I think bowl games and the BCS in particular do not answer anything about which teams are the best. Only a playoff could do that. (Just ask FCS College Football, NCAA Basketball, NCAA Baseball, NCAA Hockey, NCAA Soccer, NCAA Lacrosse, MLB, NBA, NHL, the Olympic Committee, et al, et al.)

    Now I am convinced that not only should you do your own “human” poll… you should also come up with your own computer poll as well. You don’t need to be a mathematician, you can leave the formula’s to someone else, just come up with what exactly you want measured. C’mon!

    zeekgeek

    November 12, 2009 at 4:29 pm

    • We’ll see if I can find the time to dedicate to that. But you make a great point about the strength of conference. That’s ridiculous, and might also account for why teams are where they are even without beating anyone good (or struggling against good team, as with Oregon and Miami).

      Austin Swafford

      November 12, 2009 at 4:47 pm


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: