- Peter Keating
Bowl predictions are as common this time of year as leftovers, and you may want another Fiesta Bowl call as much as you need one more plate of mashed potatoes. But here's something different: a completely objective set of BCS projections, based on a mathematical analysis of the simplest data available -- scores and schedules for every FBS team.
Theoretically, of course, the BCS computer rankings should provide valid statistical ratings for college football teams. But the BCS "computer" simply averages six different analytical methods, most of which are confidential, and none of which are allowed to include teams' margin of victory in their calculations. And its results aren't expressed in units that anyone can understand; Stanford finished .090 ahead of Oregon, but .090 what?
So here's a better way to apply silicon circuitry to college football: the Simple Rating System (SRS). SRS rates a set of teams by their margin of victory, adjusted upward or downward by their opponents' strength. That's all. SRS simply answers the question: How strong would every team in the country have to be for all their scores, week by week, to come out the way they did?
SRS ratings are expressed in points: LSU, the No. 1 team in the country, is at +34.8, meaning the Tigers are 34.8 points per game better than an average team. And they can be separated into various components -- LSU, for example, has an SRS rating of +2.4 on offense and +32.4 on defense.
The SRS isn't just easy to understand, however (for a description of the history of the formula, see the sidebar at right); it's also a better predictor of bowl game matchups than the BCS standings. If you pick bowls based on which team is ranked higher, you'll barely break even. But SRS chooses bowl game winners correctly 65 percent of the time -- and has accurately called an impressive 80 percent of BCS bowl games correctly over the past five years.
In this year's BCS bowls, SRS projects results that are mostly in line with the national rankings -- but that offer some interesting contrasts with the point spreads that have emerged and essential intel on the true strength of each team. Here is our analytically correct guide to the BCS bowl games:
The low-scoring game of missed opportunities that these teams played in November left most of the country thinking they were essentially even, or perhaps that Alabama was the superior team (more total yards, several missed scoring opportunities versus LSU). SRS, on the other hand, sees the Tigers as slightly but measurably better.
Peter Keating provides a primer on the Simple Rating System, a formula that has proved to be an accurate predictor of bowl games, and offers up some knowledge on each of the BCS bowl games.