BCS Computers and Their Impact on College Football

Dec 30, 2012, 12:53 AM

After reading many of the ongoing debates here and elsewhere concerning the SEC's "dominance" versus the rest of college football I took it upon myself to try and find out exactly what helps the BCS determine their rankings.

The first thing is the simplest. They take the Harris Poll and the ESPN/USA Today Coach's Poll each week and assign descending point values to each vote received by each team. So the number 1 gets 25 points, number 2 gets 24, and so on until number 25 gets 1. They then multiply the number of voters, 115 in the Harris Poll and 59 in the Coach's Poll, by 25, resulting in what would be a perfect score (in the case that all 115 or 59 voted the same team as number 1). They then tally up the point values for each team and divide them by the perfect score (2875 for the Harris Poll and 1475 for the Coach's). The result is a number of 1.000 or less. So that covers the human polls.

Now on to the computers. The BCS employs the work of 6 computer formulas, only one of which is made public. They average the computer ratings and finish with a number of 1.000 of less again. At this point we have each of the three determinant values that the BCS uses to generate a teams average, which is ultimately its overall BCS rating.

Just to be thorough, a given team's average is determined by adding the three ratings from the Harris Poll, Coach's Poll, and BCS computer average and dividing the resulting number by 3, thus resulting in an average of 1.000 or less.

Now that we have covered the basis of the rating system does anyone see the glaring flaw in all of this? I sure do!

First and foremost, the human polls are much too subjective to outside influences and bias. Team and conference preference and margin of victory are both major players in a given human's perspective of a certain team.

Second, we are allowing computer formulas, 5 out of 6 which we have no clue what inputs are being used or how the formula works, to have a 1/3 vote in our ranking system!

Finally, at least one formula is known to take conference into account. I can't possibly be alone on this one!

Right now the only formula that has been made public is the Colley Matrix. You can find detailed information concerning the matrix here:

http://www.colleyrankings.com/matrate.pdf

Though it was a bit difficult for me, since the mathematical equations are a bit over my head, I read the piece from end to end, even doubling back a few times for clarification. Now, I'm no math whiz. I don't have a college education. The highest math I ever took was precal in high school. Even still, Colley's explanations are relatively simple if you have a basic understanding of calculus, so I was able to come away with what I feel is a good understanding of his system.

Colley only takes into account the following data for input:

1. Wins
2. Losses
3. Games Played
4. Strength of Schedule

He does not, in any way, emphasize conference, adjust for margin of victory (which was done away with by the BCS anyway), or where the game is played. Colley's rating system relies simply on the final outcome of each FBS game played. He doesn't reward any points in any way shape or form for a win or loss against a non-FBS school.

The other five computer formulas are secret. Only a couple of the others discuss the inputs which are used for their formula. One of the gentlemen even goes as far as to keep the entire process completely secret, save for a basic methodology. Another admits to using preseason rankings for one of his inputs!

Now that I have read as much as I can about the computer rating systems and formulas, which actually ended up being much less than I thought, I must say... THIS IS LUDACRIS!!!

How can anyone tell me that the current rating system is unbiased and accurately produces a ranking of 25 FBS teams when things like preseason rankings and conference affiliations are being taken into account?

I have yet to develop a better system, but even without a college education I can tell you this is bologna! I for one would rather go back to the old way of doing business. At least then we could preserve our tradition!

flag link

military_tech

thumb_down

thumb_up

I'm not trying to say that the ACC doesn't suck, since it makes a case to the contrary week in and week out. All I am trying to do is point out a major flaw in the makeup of college football today. With computers using things like conference affiliation and preseason rankings to determine the weekly BCS rankings, it was only a matter of time before one conference began to dominate.

If conference affiliation is taken into account, then you might as well say that a team in Conference A is better than a team in Conference B with identical credentials simply because another team in Conference A is undefeated! Makes no sense.

It's just as bad to have preseason rankings involved. Look at teams like Auburn and Arkansas. Both teams were ranked to begin the season. Both teams were utterly disgusting to watch this season. Preseason rankings are based on human projections. We all know how good humans are at predicting the future. I mean heck, we're all still here. The end of the world is 9 days late at this point!

BCS Computers and Their Impact on College Football

Re: BCS Computers and Their Impact on College Football

Re: BCS Computers and Their Impact on College Football

Re: BCS Computers and Their Impact on College Football

TigerBoard POTD