Flawed computer polls facilitating Kent State's possible BCS run
Kent State is ranked No. 17 in the BCS standings and is on the verge of a BCS bid
Golden Flashes benefit from computer polls that don't reveal the algorithms used
The Colley Matrix doesn't take into account the opponents each team lost against
Oklahoma appears headed for a BCS at-large bid, most likely a trip to the Sugar Bowl. While it hasn't been a perfect season in Norman, the Sooners have a 9-2 record with losses to Kansas State and Notre Dame, two top-10 teams. They blew out Texas in the Red River Rivalry. Moreover, they will survive the Big 12 with a single conference loss if they can beat TCU on Saturday. There are four at-large slots available for teams that did not win their conference, and OU will line up fourth behind Notre Dame, Florida and Oregon should Kansas St. win the Big 12 on Saturday with a win over Texas.
But Oklahoma could lose its BCS bid to Kent State.
The BCS makes it possible for a team from a non-automatic qualifying conference to make a BCS bowl under a few special conditions. First, that team must finish in the top 16 of the BCS standings. Second, it must finish ahead of the champion of an AQ conference. The 11-1 Golden Flashes are currently No. 17 in the BCS standings and ranked ahead of every Big East team. However, Kent State also lost 47-14 to a Kentucky team that didn't win a single SEC game. How can it be in a position for a BCS berth?
Part of the answer lies in the problematic computer polls, and most of the issues are well known. To begin with, five of the six computers do not reveal the details necessary to reproduce their results. Millions of BCS dollars are at the mercy of five black-box algorithms. Second, the BCS forbids the computers from using margin of victory in their calculations. It doesn't matter that a 33-point loss to Kentucky says something much different about a team than a one-point loss to a top-10 opponent. In the name of sportsmanship, the BCS will not give teams the incentive to run up the score.
At the very least, you'd think the BCS computers could factor in the results of each game. For example, they could take into consideration which team handed Kent State its lone loss. Losing to 2-10 Kentucky would presumably hurt the Flashes' ranking. If they had lost to 9-2 Rutgers instead, the computer poll could assign Kent State a different, and presumably higher, ranking.
However, the Colley Matrix, the one fully transparent computer poll, does not use this game-specific information. The system considers a team's win-loss record and strength of schedule; yet, the results of each individual game are not counted as an input. The method doesn't care whether Kent State's loss came against Kentucky or against Rutgers.
It doesn't require an advanced math degree to verify this. Wesley Colley, the creator of the ranking method, has set up a web page where you can alter the results of games and see how the rankings change. If Kent State loses to Rutgers (25th in Colley's rankings) but beats Kentucky (98th), common sense indicates its rank should improve. However, the Flashes actually drop one place. In this scenario, Rutgers has a 10-1 record instead of a 9-2 mark and therefore moves ahead of Kent State.
With the help of a computer poll that ignores who a team loses to, Kent State is on the brink of knocking Oklahoma out of a BCS bowl. If it beats Northern Illinois in the MAC title game, it could move up far enough in both the human and computer polls to end up No. 16 or higher in the final BCS standings. In the process, Kent State would redirect millions of dollars from the Big 12 to the MAC.
It makes two things very clear: Sooners' fans should be cheering hard for Northern Illinois on Friday. And the rest of the MAC should be cheering hard for Kent State.
Ed Feng runs the sports analytics website The Power Rank.