10/2/07 - The Real Problem with the BCS

It's never too early to discuss the BCS, and any time the BCS is mentioned there is instant controversy. Many fans of college football would love to see the BCS dismantled and the Division 1 Bowl Subdivision turn to a playoff system to determine its National Champion. The biggest problem with the current BCS system, they say, is the inclusion of computer polls. But what many don't understand is that the six computer polls used in the BCS formula are stripped down, unbiased, "politically correct" versions of the computer ratings. That is to say they do not include any measurement of margin of victory or scoring margin. The problem with this is that most of the computer polls used were originally calculated and developed using some type of margin of victory. Here is the problem. Currently, the Northern Iowa Panthers look like the sleeper in the race to the BCS Championship game! Looking at Jeff Sagarin's current ratings, you'll see that Northern Iowa is ranked at #56. In case you are unaware, Northern Iowa plays in what is now called the Division 1 Championship Subdivision, or as most know it to be called, Division 1-AA. Now the fact that the Panthers, a Division 1-AA team, is rated #56 is not at all surprising. Northern Iowa is 5-0, with 4 straight wins on the road, and one of those wins coming at Division 1 Iowa State. But where things really start to get strange is when you look at what Sagarin calls his ELO-CHESS rankings, or the unbiased rankings used by the BCS. ELO-CHESS does not measure any type of scoring margin and are strictly based on wins and losses. Northern Iowa, what Sagarin's true rating considers the 56th best team in the nation, is ranked #6 when scoring margin is removed. That means the BCS would look at Jeff Sagarin's ratings and think that the Northern Iowa Panthers are the sixth best team in all of college football! Obviously, we have a problem. Now I realize that it is early in the season, and with the more games UNI plays against Division 1-AA opponents, even if they continue winning, the more they will fall down these ratings, but this should be an eyeopener that the current BCS formula is flawed. How does a Division 1-AA team who has played the 135th toughest schedule to this point look like a top 10 team to any computer? When you remove scoring margin, you are handicapping these computer polls from doing what they were originally developed to do. The BCS does not want to include any type of scoring margin due to the fear of coaches "running up the score" on their opponent. First of all, this already happens on a weekly basis. Under the current system, and any system we choose, coaches will always have the thought of scoring a few more points in order to impress the pollsters. This is going to happen whether those pollsters are human or computers. The art of "running up the score" has been around long before computer polls, and will always be a part of college football so long as there are polls. But margin of victory, with strength of schedule, is the truest way to determine the quality of teams who do not play head-to-head. For example, Team A defeats Team B handily by three touchdowns. Team C defeats the same Team B on a last second field goal. Shouldn't Team A be rewarded for playing better in a game against a like opponent? The Solution. Why not have margin of victory be included into the equation of these computer polls but with a cap or limit? That is to say, if a team wins by 20 or more points it is considered a 20+ point win. So a 20-point victory is measured the same as a 45-point win. That will appease those who are concerned about teams "running it up", but it will also reward a team that wins by 20 more than a team that wins by 1. Obviously no system is perfect, and just by adding a scoring margin element the computer polls are not going to all of a sudden be able to predict everything. But with the way these computer polls are currently derived, the BCS formula cannot possibly do what it was developed to do. Just think of a game between Northern Iowa (currently #6 in Sagarin's "unbiased" formula) and Oklahoma (38th in that same "unbiased" formula), then try to tell me that there isn't a major problem with the current BCS formula, that could be fixed very easily.