The BCS, Computer Polls, and College Football

by Bryan on 1/5/2004 at 8:38 PM in Sports

I came across a post by Jason Salas concerning the BCS and it's use of computer polls to aid in determining a champion.

First a quick correction to his post, he's got the polls and their champs reversed...USC was voted #1 in the AP, or writer's poll, while LSU claimed the BCS championship, resulting in them being named #1 in the ESPN/USA Today Coaches Poll, which is contractually obligated to name the BCS champion as their champion.  The interesting thing is that three coaches didn't do this, rather they kept their #1 team from the final regular season poll (USC) intact.

The problem that I see with the computerized rankings is the same as with the human polls -- that is they are opinion based.  Sure, from week to week, someone just plugs in a bunch of numbers and the computer spits out the rankings, but in the beginning of each and every one of the computer polls, the developer had to decide on the things that are important in ranking teams and how important each of those things are in relation to the other things.  Like any other computer program, the computer polls are only as good as the people who designed and wrote them. 

As far as the BCS goes, my gripe isn't with the computer polls, though I do scratch my head from time to time, after seeing some of their rankings, wondering what they were smoking when they put their system together.  My problem is with the BCS itself, and with the AD's who insist on using it to determine a national champion, rather having a playoff as is done in every other college sport at every other level.  Personally, if we have to have the BCS, I would make these changes...1) Remove strength of schedule as a BCS component.  Strength of schedule is already factored in to each and every poll, including the human polls.  (Don't tell me that the voters don't take into consideration who a team has played, and possibly who those teams have played, when voting.)  2) Put margin of victory back in.  I just don't see how you can have one without the other.  The kids playing the games have no control over who is on their schedule, but they do control the outcome.  Shouldn't a team be able to offset a weak strength of schedule by a wide margin of victory?  I think so.  3) Finally, remove the bonus points for quality wins.  They should already be reflected by record and strength of schedule.

Congrats to USC, and to those 3 coaches who had the balls to vote against their contract.  Now let's have a playoff.



Legacy Comments

My problem with a computer system is the so-called strength of schedule variable. Here's 2 ways it doesn't work. #1. Let's say TEAM #1 has a loss to a 7-4 team DIV-1A team like Idaho, while TEAM #2 has a loss to a 7-4 team like Auburn. In the computer system, both team's strength of schedule losses are treated equal, even though we all know that Auburn's 7-4 record is going to be a much tougher one than Idaho's. #2. Let's say that TEAM #1 beats a #3 ranked team at the beginning of the season when they're at full strength. Then, that team has a rash of injuried, so they fall out of the polls. TEAM #1 is penalized, even though they beat the team when they were at their toughest. This can also go the other way, in that TEAM #1 can beat an unranked team with a bunch of starters who are ineligible for the 1st game of the season, and then those players come back and provide that team with a good record later. TEAM 31 beat a handicapped team, yet they get credit for beating the team at their ending strength. To fix #2, we need to use strength of schedule that computes wins and losses with point values based on the quality of the opposition on the day that they played them, and not how they eneded up. I also think that strength of schedule has to take into consideration the the opponent's opponent winning percentage, because a 7-4 team from the MWC is not the same as a 7-4 team from the SEC.

Posted by on 1/12/2004 12:07:06 AM