I know I should be on topic with Global Warming, but this is a subject that irritates me, so I decided to write an off topic article on how error prone the BCS system is at picking the National Championship Game for College Football. I was going to start a cause, but it seems that many people have already jumped on this and many pages exist to promote a boycott of BCS football games. For me it means I will not watch any BCS football games this year and perhaps even in the future. I wash my hands of the BCS.
Now to show just how bad the BCS system is at choosing the National Championship Game (NCG). I will do this by using the Sagarin computer ranking of the NFL and comparing how accurate it is at picking the Superbowl. I pick this method because the NFL uses a play-off system that pits the top 6 teams from each of two conferences, where the winner of each conference plays in the Superbowl.
The NCAA is the only group that doesn’t have a play-off system to chose who will play for the championship. To put it in a politically correct term, they are flaming idiots. They have a convoluted method where they simply pick two teams from a pool of 100+ teams and then whoever wins the game is the champion.
I use the Sagarin ranking because it is used as part of the BCS selection method and it also shows rankings for the NFL using the same methodology. The rankings are available back to 1998 for the NFL and that is the period I used to show how poor a predictor it is. The only real issue I see is that the Sagarin rankings include the play-off and the Superbowl so the final rankings include that final game. After I was done crunching the results it hardly mattered except for that there was a correlation between the winning team and the Superbowl winner. Since the purpose is to predict the game, that doesn’t really cause a problem.
I will start off by showing the final top two teams by the Sagarin ranking from 1998-2010.
Right away there is an issue with the idea of using a computer to choose the Superbowl. 5 out of 13 years (38%) the top two teams have come from the same conference based on the computer rankings. This is similar to the result this year where both LSU and Alabama are the top two teams based on the computer ranking. Of course in the NFL it is not possible for two teams from the same conference to meet in the Superbowl. There have been many great NFC and AFC championship games over the years as a result, but in the end each team comes from a different conference.
Of the 8 times that the top two teams have come from different conferences, only 3 times have the top two teams met in the Superbowl. That leaves a 23% rate of predicting the Superbowl in general, but even when that is only used to compare for the 8 times the top two teams have been from different conferences, it is only a 38% accurate predictor for who plays in the Superbowl.
When it comes to picking the Superbowl, the rankings system is an abject failure as it has only managed to pick the Superbowl teams a dismal 23% of the time. That is a reasonable estimate of the accuracy for the BCS system. It gets it right about one quarter of the time and this year is not one of the good years. This year it is certainly possible that the Packers and 49ers will be the top two teams and they will have a chance to play in the play-offs.
If the BCS rankings were used to pick the top 12 or top 16 teams to have a play-off, that would be a useful application of the BCS. Instead what we have is a BCS system is broken and a play-off is needed. It is time to take action and boycott the entire system until we get a play-off. I understand that the students and alumni of their respective schools will participate, but for the rest of us. Let the boycott begin.