College Football

Bowl Championship Series
The Man Behind The Math

A few days after the first Bowl Championship Series ratings came out on Oct. 26, college football's official math geek lamented that they had been released. "I was the conservative," SEC commissioner Roy Kramer said. "I don't think the ratings are tremendously accurate until we get eight or nine games in."

College football now anxiously awaits the release of the ratings on Monday afternoon. This week Ohio State and Tennessee are in, UCLA is in the wings, and Kansas State continues to study the Sugar Bowl brochure.

Kramer, who spearheaded the development of the formula that will determine which two teams play in the Fiesta Bowl for the national title, rattles off terms like "adjusted deviation" and "quartile" as if they were part of the football lexicon—which, thanks to him, they are. It's no mystery why the commissioners of the conferences in the Bowl Championship Series chose Kramer to come up with the rating. He is a former chair of the NCAA Infractions Committee and a onetime football coach (Central Michigan, 1967-77) who headed the Division I men's basketball tournament selection committee, and he has a reputation for fairness and honesty, even though his tendency to hoard power causes colleagues to smirk. One commissioner, asked why the Bowl Championship Series committee shouldn't just lock itself in a room and choose two teams for the title game, said, "Then Roy wouldn't be in charge."

"I bet he would," says Dave Cawood, the former NCAA official who's now a vice president at Host Communications, the company that does the NCAA's marketing. "He has such great integrity that people believe he'll be fair in whatever is done. People know that he does his research and thoroughly studies an issue before he takes a position. This is a case in point."

Kramer, along with his top assistant at the SEC, Mark Womack, and conference media officer Charles Bloom, spent three or four hours a day for three months developing the formula. They looked at more than 40 computer ratings before choosing those produced by The New York Times; Michigan-based computer whiz Jeff Sagarin, whose rankings appeal-in USA Today; and The Seattle Times. Not only do the three ratings complement each other, but their geographical diversity is also politically correct. "Using Seattle helped with the Pac-10 a great deal," Kramer says.

He and his aides tested a dozen or so formulas by applying them to the past 10 seasons. They knew the answer—e.g., last season Michigan and Nebraska should finish first and second—and had to come up with a formula that would produce that answer. "We had to be able to get to a place that you could defend," Kramer says. The method on which they settled weighs equally a team's average standing in the AP and USA Today/ESPN polls and its average standing in the three computer ratings, then uses the BCS's own strength-of-schedule rating as a tiebreaker.

Among the years that proved the formula viable was 1989, when Colorado went 11-0 in the regular season and Alabama, Miami, Michigan, Nebraska, Notre Dame and Tennessee finished with one loss. The formula rated eventual national champion Miami, which had beaten Notre Dame in the last game of the regular season and had lost only to eventual No. 5 Florida State, as the team to play Colorado in the theoretical title game.

Kramer says he has been flooded with mail from math professors stating that the formula incorrectly uses adjusted deviation (don't ask) to average the three computer ratings. He's experimenting with an alternative formula that may be more accurate (so far, he says, it hasn't contradicted the Bowl Championship Series rankings). Get ready to widen your football vocabulary next year: "You know what a trimean is?" Kramer asks.

Rumblings Out West
The WAC Whacked?