To: Jim Delany (Big 10) John D. Swofford (ACC) Larry Scott (PAC 12)
An Open Letter to: Jim Delany (Big 10), John D. Swofford (ACC), Larry Scott (PAC 12)
Re: BCS Poll
You should immediately quit the BCS. It is rigged against you and your three conferences. If you read my letter then you should see from my analysis, that the “computer polls” are inherently biased (and perhaps worse than the two human polls that make up the other 2/3 of the BCS rankings).
First off, I used the Human Polls (Harris Poll & USA Today/ESPN) as the normative index. If you say this ok then you can accept my analysis. If you reject it, then you should be pitting LSU against Oklahoma State in the BCS Championship Bowl Game, because that is what the Computer Polls would have made the result if there were no human polls as a part of the BCS Index.
My analysis clearly shows that the computer polls OVERWHELMINGLY favor the BIG12 and have a strong bias in favor of the SEC too. At the same time it is OVERWHELMINGLY rigged against the BIG 10 and strongly biased against the ACC and the PAC 12 conferences.
The analysis shows that the Big 10, ACC and the PAC 12 would have to overcome a huge bias by the computer polls via the Human Polls to have any chance to reach the BCS Championship Game. You should realize that by selecting the SEC every year to play in the BCS Championship Game, you keep the bias in the computer polls and it will become a self-fulfilling prophecy each and every year. That means the BIG Money will continue to flow unchecked into the SEC (and also to lesser degree to the BIG 12) as it is a “virtuous cycle” upwards for these two conferences who get the best recruits and booster money because they are ALWAYS in the BCS Championship Game.
Now that you have given in to the precedent of two teams from the same conference in the BCS Championship Game (should be a rule against this) you will see a heavy bias to that year after year, since that is all new recruits will see and the “virtuous cycle” will persist. Also, did you realize that the computer and the human polls will emphasize the next year’s polls based upon the previous year, via the pre-season polls?
The root cause you will see is that two computer polls in particular: Kenneth Massey & Jeff Sagarin strongly overemphasize Big 12 teams and SEC teams also have a strong positive bias, while at the same time, these two same computer polls also demonstrate an under-emphasis of the Big 10 and a strong negative bias against the ACC and PAC 12. The effect is what we have seen for the last few years and culminating in this years SEC-only Championship.
If you want to keep the BCS Polls, then you will need to do five things to improve them and their perception as fair:
- Make a rule that the BCS Championship can NEVER have two teams from the same conference. This should be self evident.
- Make the remaining computer polls submit their algorithms to an audit before the season starts and a week before/after the final BCS rankings \to ensure that these computer algorithms are “bias free” from human intervention and that the same results are achieved in the before/after of the final rankings (i.e no tampering and results are reproducible i.e. no randomness).
- You must get rid of one or both of Kenneth Massey of Jeff Sagarin computer polls. The dual combination skews the biases in favor of BIG12/SEC and against the BIG10/ACC/PAC12. If you only get rid of one, then the initial removal should be Jeff Sagarin. The two computer polls show the same bias and are merely echoes of each other thus giving them an undue advantage over the other four computer polls. The Jeff Sagarin poll is merely MORE pronounced (in its biases) than the Kenneth Massey poll.
- No 4 loss or 5 loss TEAM can ever be eligible for a BCS Bowl Game. You need this rule to prevent obvious bias from contaminating the system.
- No 2 loss TEAM can play in the BCS Championship Bowl Game (substitute the next highest ranked team that does not violate rules 1 & 5).
Mind you the Anderson & Hester computer poll exhibits some bias too, but it at least it is not in COMPLETE lock step with the Kenneth Massey or Jeff Sagarin polls. Otherwise, please dismantle the BCS system and just have 4 super football conferences and take the conference champion from each and have these four teams play a semi-final and a final game to determine the national champion fairly. See the attached spreadsheet data, cut/pasted into the next page and do you your own analysis to validate my findings and see if you reach the same conclusion. Please pay special attention to TEXAS in the final rankings if you wish to be totally disgusted by the computer polls – there is no mathematics that can justify that conclusion by computers, unless there is a BIG12 bias. The computer polls would have made TEXAS, a 7-5 team, the 19th ranked team overall in the whole country and the two offending computer polls would have made TEXAS 13th in the country and eligible for a BCS at Large Bowl Game. Can you imagine? Only TEXAS and AUBURN (BIG12 & SEC) have 5 losses in the BCS Top 25. In fact there are no other 5 loss or any 4 loss teams!
Someone should commend the Richard Billingsley, Colley Matrix and Peter Wolfe computer polls for their ability to keep bias from skewing their rankings.
Anderson & Hester can and should do better in their computer algorithm.
|Comp Diff||Diff Summ|
Source: 12/5/2011 Philadelphia Inquirer Final BCS Standings
The bottom five teams were unranked in one or more computer polls making their data unfit for some of the analyses – these were not used in the bottom analysis of Skew By Conference.