According to the BCS computer ratings, 21-20 is no different from 45-0. A game decided by a last-second field goal is the same as a blowout that is over at halftime.
Putting the perception of sportsmanship ahead of the harsh reality of statistics, the conference commissioners behind the Bowl Championship Series demand that programmers whose computer rankings are used to determine which teams play in the BCS title game remove margin of victory from the equation.
The problem, as this season is showing, is that margin of victory has become another part of a postseason system stacked against Boise State, TCU or Utah playing for the championship.
“You’re throwing away information,” said Jeff Sagarin, the mathematician and MIT graduate whose computer rankings for college football have been used by the BCS since its inception in 1998.
BCS officials don’t dispute Sagarin’s claim.
But they don’t want to encourage coaches to pile up points just to boost their team’s ratings. Said BCS executive director Bill Hancock: “The commissioners decided that the price was too high in terms of sportsmanship.”
The formula for the BCS standings has changed several times, mostly to put more emphasis on the human element. The Harris poll and the coaches’ poll account for two-thirds of a team’s BCS average. A compilation of six computer ratings make up the other third. Beginning in 2002, margin of victory was removed from the computer rankings — at least the ones used by the BCS.
The true Sagarin ratings include margin of victory. That final rating is reached by combining two elements, he said. One rating does not include scores — that’s the one the BCS uses — and another does. Strength of schedule is a component in both.
Sagarin said the two ratings work together and balance out, so a team stacking up blowouts (presumably against weaker competition) isn’t overrated and a team that is constantly squeaking by (presumably against tougher competition) isn’t underrated.
For an example of the balancing act, look at this week’s Sagarin ratings for Auburn and Boise State.
When scores are taken into account, the Tigers, with several close victories, are the 15th-best team in the country. When the scores are removed, Auburn is No. 1. When the two elements are merged Auburn, comes in sixth.
Boise State is No. 13 without scores and No. 5 when scores are part of the equation. The Broncos’ true Sagarin rating is seventh.
Boise State plays a weaker schedule in the Western Athletic Conference than Auburn does in the Southeastern Conference. The way the Broncos can distinguish themselves in the computer ratings is by winning big against those softer opponents, which they usually do.
“When you’re locked into playing a lower schedule like Boise State is, and you’re in a win-loss system (with no scores), you can’t escape the balance of your schedule,” Sagarin said. “You’re automatically pulled down, but you can’t escape past a certain point, even if you’re undefeated.
“They might as well acknowledge before the season starts, ‘Hey, it doesn’t matter if the following teams go undefeated, they’re not going to get in the (championship) game.’”
By the way, Oregon and TCU are the top two teams in the true Sagarin ratings this week.
Kenneth Massey, whose ratings have been used by the BCS since 1999, agrees that a more comprehensive rating would include scores. The ratings you’ll find on his website include scores. Oregon is first, TCU is second and Boise State is third. Auburn is fifth.
The BCS version of the Massey ratings has Auburn first, Oregon second and TCU third. Boise State is seventh.
Unlike Sagarin, however, Massey said he doesn’t believe removing scores puts Boise State and TCU at a disadvantage in the BCS.
“The BCS’s primary goal is to match up the teams that most deserve to play for the national championship,” he said. “I do think it’s a little bit truer to the real goal of sports which is to win and not necessarily worry about the margin of victory.”
Every other sport uses a playoff to determine a champion instead of computer ratings.
Coaches hate talking about margin of victory — aka style points — especially Boise State’s Chris Petersen and TCU’s Gary Patterson, who face questions about whether a victory was lopsided enough to please poll voters.
Coaches generally espouse the a-win’s-a-win philosophy.
“If we would have lost 20-19 last week, we’d still be 8-1,” said Michigan State coach Mark Dantonio, whose team lost 37-6 to Iowa and tumbled from fifth to 16th in the AP Top 25. “I think the margin of (victory) is recognized in the AP vote, I’m sure, and the coaches’ vote, I’m sure.”
That is another part of the argument for keeping scores out of the BCS computers. Poll voters clearly take margin of victory into account, so it’s not as if it is completely removed for the equation.
Oklahoma coach Bob Stoops admitted he tried to hold down the score at the end of a loss to Missouri because he didn’t want poll voters to see a more lopsided final.
A score can be deceiving. A seven-point game can turn into a 21-point game because of a couple flukey plays. Because of that, the argument goes, it’s better for a human being who has a chance to watch the game — or at least read about it and see highlights — to determine the score’s true worth.
But that opens another set of problems, Sagarin said.
“Once you start letting a human jury decide the real score of a game, then why even have a computer,” he said.
Hancock notes the NCAA produces computer-generated RPIs in about a dozen sports that are used by selection committees to fill postseason tournaments — and none includes margin of victory.
But never before have so many football teams that could get hurt by leaving out margin of victory had a shot at the national title.
— AP Sports Writer Noah Trister in East Lansing, Mich., contributed to this report.