Reigning BCS champion Florida State returns seven offensive starters from a top-five offense, six defensive starters from a top -five defense, the Heisman Trophy-winning starting quarterback and its head coach, and the Seminoles have had top-10 recruiting classes over the last four years. With all of that going for the Seminoles, how could they be anything but the No. 1 team in a system that uses all of that information?
That system is the College Football Power Index (FPI). Introduced last year, the method has been improved to account for all of the above information because, well, those things matter. You as a fan know it, and the numbers support it.
FPI not only says the obvious -- such as Florida State being No. 1 -- but it also gives nuance, such as the Seminoles having just a 39-percent chance of going undefeated through the ACC championship game, despite how good they are.
It also says the surprising -- such as Marshall having the second-best chance of going undefeated this year, thanks to the easiest schedule in college football. If Marshall climbs the polls this year, remember that pretty much any average team would have a similar chance to go undefeated against the Thundering Herd’s slate. FPI ranks them 54th, roughly average across FBS teams, but this means that an average team would have a pretty good chance to go undefeated against that schedule. Only a road game against Old Dominion does the Thundering Herd have less than a 60 percent chance of winning. If they go undefeated, the College Football Playoff selection committee shouldn’t look at them seriously if it values a team’s strength of schedule.
FPI captures a lot but might not capture everything. Last year, Michigan State fans hated FPI because it never saw the Spartans as a top-five or even a top-10 team. The reason was an offense that got into the end zone a total of three times against weak teams Purdue, Western Michigan and South Florida. FPI kept predicting Michigan State to lose or to at least not cover the spread. And FPI kept being wrong.
But we improved FPI this year, and it now predicts games at about 75 percent over 10 years (and does better with last year’s Michigan State team). Because it predicts games well, it can simulate every game for every FBS team for the entire season and produce these kinds of prognostications for 2014.
Oregon has the third-best chance of going undefeated at 6 percent. That’s not very likely, so don’t get mad at FPI if it doesn’t happen.
Realistically, there will be only one undefeated team through this season’s conference championships.
Auburn has a better FPI rank than South Carolina, but South Carolina’s chances to win the SEC are higher than Auburn’s. This is because Auburn’s schedule is quite a bit harder, playing seven teams in the FPI top 25 (four on the road) compared with South Carolina’s five (two on the road).
The four most difficult schedules in the country are in the Pac-12. Every Pac-12 team has a schedule rated in the top 40. Five Pac-12 teams play only one team with an FPI under zero, where an FPI of zero represents a team that is exactly average. No other team in any other conference plays fewer than two games like this. Pac-12 teams don’t schedule easy games.
UCLA has the most difficult schedule and, as a result, projects to lose three games despite being the fourth-best team in FPI. They’re projected to be so high in FPI because they return 17 starters from what was a very good team last year.
The loss of Ohio State quarterback Braxton Miller lowered the Buckeyes’ FPI rank from sixth to 12th. It likely adds a projected loss to their record and lowers their chances to win the Big Ten from about 40 percent to about 30 percent. That 30 percent chance, though, still leaves them as the favorite in FPI’s eyes.
That brings us back to Michigan State. Michigan State’s projected record is only about 8-4, and the Spartans have just an 11-percent chance to win the Big Ten. There are a couple of reasons FPI doesn’t like Michigan State as much as (seemingly) everyone else. It returns only five to its vaunted defense. Its offense hasn’t been particularly good over the last few years. And people forget that they were a very mediocre 7-6 team in 2012 in a weak conference.
FPI doesn’t account for any specific strength that the Spartans coaching staff has for rebuilding the defense, so it could be wrong. FPI doesn’t account for Connor Cook being some analysts’ favorite dark horse quarterback, so it could be wrong. But FPI has seen a lot of seasons, a lot of coaching staffs, a lot of quarterbacks, and it’s learned to ignore the hype. So, yes, FPI hedges its bets and says the Spartans have about a 1 in 9 chance to win the Big Ten.
It’s not ruling it out. And it might change its mind if the Spartans start the season destroying teams. Just like on-air analysts, FPI is allowed to incorporate new information.
One thing FPI won’t be particularly good at, though, is identifying which teams should be in the Playoff. This is because we, the fans, and the College Football Playoff selection committee don’t know what we want in the Playoff. The polls and the committee will reflect something different than who will win the next game. Polls often reflect a season résumé, including accomplishments such as a conference championship, a win-loss record in combination with a strength of schedule (something we put into a metric called “Strength of Record”), as well as in-game dominance.
FPI is about looking forward. As a result, it’s perfect for this time of year.