If one University of Michigan graduate student had his way, the Missouri football team would be playing in a postseason tournament.</p><p>Brady West, a doctoral student in survey methodology, developed a rating system for college football teams and &ndash; don’t tell the BCS &ndash; the top two teams in the country are Texas and Southern California.</p><p>More surprisingly, Missouri was No. 3. Florida State was fourth, followed by Mississippi, Texas Tech, Penn State and Clemson. As for the participants in Thursday’s BCS championship game, Florida was tied for 12th (with Nebraska) and Oklahoma was 20th.</p><p>“That could be a function of the system,” West said of the top-ranked Sooners, “but I think it’s the defensive numbers.” </p><p>Oklahoma is allowing an average of 359.1 yards per game this season, and that is key in West’s predictive model.</p><p>West, whose findings first appeared last year (using the 2007 teams) in the Journal of Quantitative Analysis in Sports, took all 68 bowl teams from the 2008 season and played a round-robin tournament, using a number of predictors.</p><p>“All these predictors were assumed to have a linear impact on the games,” West said. “For example, scoring margin. With every additional one unit difference in scoring margin &ndash; not one point, but one standard deviation difference between two teams &ndash; the predicted outcome would go up by 10, and it would go up by 10 with every additional standard deviation difference between the teams. “This year I went in and refined the model. What I really want to do is try getting a better fitting model with every additional year of data I collect.”</p><p>The key predictors were scoring margin and strength of schedule.</p><p>“Those two guys are just really, really important,” West said.</p><p>The next two significant predictors were turnover margin and defensive yards allowed per game. Interestingly, teams did better with a negative turnover margin.</p><p>“Turnover margin actually had a curve linear effect,” West said. “The relationship with turnover margin actually looked like an upside down U. ... The team with the better turnover margin is actually expected to lose.</p><p>“That pattern has held up where turnover margin actually seems to have a detrimental effect. It’s something that has no science behind it, but it’s like 'Why is that happening?’ ”</p><p>West has the same question with the BCS. He wonders why there is no playoff system, and would like to see an eight- or 16-team competition. But for someone who crunches numbers, West is not interested in merely having computers pick the field. He believes the polls should play a part as well, and gives the BCS credit for using the polls in its rankings.</p><p>Speaking of polls, Missouri was ranked 25th in the latest Associated Press poll. How could they be third in West’s calculations?</p><p>“They had a very high scoring margin and had a high strength of schedule from the competition that they played,” West said. “Pretty much if you match up Missouri against all the other teams, except the big names like Texas and Oklahoma, they’re going to be expected to win those games. ...</p><p>“It’s that human factor that for whatever reason they couldn’t put it together in some of those big games. Basically they had a good team on paper, but when they played those big matchups, they couldn’t quite put it together.”</p><p>West admitted that his model for ranking the teams was in its infancy, considering he started his work last year as a hobby. There are predictors he’d love to use such as how teams play in weather, coaching experience, etc. He also has had to rely on finding team statistics online, and they usually included bowl games.</p><p>So he had another student, Madhur Lamsal, remove the statistics from the bowl games from a team’s final numbers before plugging them into his past models. That slowed the process.</p><p>“I would love to have data going back to the 1980s or earlier,” West said, “so that I could pick up some better patterns of what’s actually happening in these bowl games. As the paper said, I’m only explaining about a quarter of what actually goes on with these predictions. There’s just so much you can’t predict objectively like injuries or players dealing with agents, all the stuff that can happen.”