Two thirds of the BCS formula are human-controlled. The Coaches Poll, voted on by the population segment most-biased and least able to sample a broad array of games, is one-third. The Harris Poll, a selection of people with no business voting parroting the Coaches Poll, provides the second third.
Computer polls provide the final third. The formula factors in six polls of varying statistical validity, neutered by their inability to account for margin of victory. Some pollsters run more reflective versions of their own polls they submit to the BCS. The BCS does not even independently verify the data. The computer polls serve little purpose, beyond creating a number to add an aura of objectivity and to offer a convenient scapegoat should the public dislike the two teams chosen.
The BCS formula is biased, flawed and illogical. It places two teams into a game. But, any one of us could do that. For the formula to provide a pro-competitive benefit, it needs to ensure those are the best two teams. Dr. Thomas J. Reynolds, a Professor Emeritus in the University of Texas system, assessed the BCS empirically against a simple statistical model.
Reynolds created the “R Methodology” to rank college football teams. Broken down to essentials, the R methodology measures wins (with a small, square-rooted margin of victory component) rated by opponent quality, based on the opponents’ wins. To provide a pro-competitive benefit, the BCS formula should perform better than this basic statistical model. It doesn’t.
Looking at data from 1998 to 2009, the R methodology outperformed the BCS formula predicting regular season outcomes (78.2 to 76.7 percent), all bowl outcomes (58.3 to 57.0 percent) and BCS bowl outcomes (59.6 to 55.8 percent). The BCS formula performs worse than a basic statistical model. It has not even proven it performs better than subjective polling. “There is no empirical evidence that their model is better than the prior versions (Media and Coaches Poll)” Dr. Reynolds told The Big Lead.
Placing two teams in a title game isn’t the pro-competitive benefit. It is the method of placing two teams in the title game fairly. The BCS formula, statistically invalid, can’t claim to do so, thus it isn’t a pro-competitive benefit and the BCS can’t claim it for exemption from the Sherman Antitrust Act. This would force the DOJ to look at the anticompetitive elements, such as the cartel intended to funnel a disproportionate amount of the revenue into the AQ conferences, and probably force the BCS to be amended or disbanded altogether.
How do we replace this? I would argue for a playoff, but Dr. Reynolds presents an alternative solution that would better determine a champion while enhancing the meaning, the interest and presumably the profitability of bowl games.
Reynolds would replace the BCS formula with a transparent, unbiased statistical formula, an optimized version of his R methodology. Bowl games would factor into the final rankings, which could be used to determine a champion. It could also use the bowl games as a de facto playoff to determine the two participants in a plus-one national title game.
The Reynolds system adds interest at the top by maximizing the number of title contenders. In 2008, his system would have left six potential national champions after the regular season. This makes multiple bowl games directly relevant for determining the national title, more if there’s a plus-one. It also injects meaning at the bottom. Results of lower-rung bowl games would affect the ratings of the teams that beat them. A mediocre game such as Syracuse vs. Kansas State in the Pinstripe Bowl could have national importance.
This would also improve the regular season. Teams’ goal would be maximizing win value rather than inflating win-loss records to dupe Harris Poll voters. “It encourages teams to play the best teams possible, rather than patsies to increase their number of wins,” Reynolds said. “Beating a crappy team doesn’t do much at all.”
Dan Wetzel’s 16-team playoff may be an impractical Utopia. However, virtually any reform in that direction, even just using better math, could improve a college football postseason that is neither competitive nor compelling and costs schools an inordinate amount of money.
[Photo via Getty]