College football season kicked off 69 days ago, on Aug. 26. But with the College Football Playoff committee releasing its first set of rankings Tuesday night, the season for irrational pigskin arguments has only just begun.
Take the committee’s initial rankings: Although the top three of Alabama, Clemson and Michigan didn’t surprise anyone — they’re three of the only four remaining undefeated major-conference teams in the Football Bowl Subdivision — the decision to put one-loss Texas A&M at No. 4 over unbeaten Washington was generally seen as a head-scratcher. And that’s just the first edition of the rankings! A whole lot more can happen between now and Dec. 4, when the final version drops.
To help you sort it all out, FiveThirtyEight is relaunching our College Football Playoff prediction model for the 2016 season. In fact, this year we’ve built a continually updating page that lets you view the current favorites, as well as the not-so-favorites: Every team with at least a 0.1 percent chance of making the playoff is listed. You can also choose custom scenarios for the future, letting you see how the model changes if a given team wins or loses its next game, or if it wins out over the remainder of its schedule.
Will your favorite team make the College Football Playoff? See all of our predictions for the 2016 season »
Here’s a brief refresher on how our model works: It takes the most recent version of the CFP committee’s rankings, simulates the next week of games using ESPN’s Football Power Index (FPI), uses a version of our old friend the Elo rating to guesstimate how the committee would react to those FPI-simulated results,might not always follow historical precedent.">1 and then repeats the process until we have a simulated version of the committee’s final rankings.
Do that 10,000 times, and we can estimate the likelihood of any team ranking in the committee’s top four when the dust clears on the season. But … does it actually work? Well, for what it’s worth, the model did a pretty good job of predicting the field last season. Just don’t ask us about what happened in 2014. (Don’t worry, we also tweaked the model afterward to fix some of its first-year bugs.)
This season, Clemson, Alabama and Michigan check in at Nos. 1-3 in playoff probability — which isn’t exactly a shock. And Washington fans shouldn’t feel too bummed about their team’s snub in the first CFP rankings. Despite placing fifth in the committee rankings, our model gives the Huskies a 47 percent probability of making the playoff, the fourth-best odds of any team, and greater than a 99 percent chance of making the playoff if it wins out (though it gives the Huskies just a 25 percent chance of doing that).
Of course, the model also comes with a lot of built-in uncertainty, especially at this early stage of playoff speculation. It says there’s an 84 percent chance that at least one team outside the Clemson-Alabama-Michigan-Washington quartet crashes the CFP party by season’s end — most likely Ohio State (though the Buckeyes probably need to beat Michigan on Nov. 26 to do it), Louisville, Texas A&M or Wisconsin, all of whom have a 10 percent shot or better at the playoff.
Some of those teams need more help than others, however. The Buckeyes and Badgers could control their own destinies to some degree by winning out — Ohio State’s odds would be 96 percent if it ran the table, Wisconsin’s would be 74 percent if they did the same, and Texas A&M’s would be 61 percent. But Louisville’s chances at the playoff would only be 39 percent even if we knew they’d win all their games from here out. (Gaming out contingencies such as those are where the choose-your-own-scenario feature of our interactive comes in handy. Hint, hint.)
Even with the perfect number of undefeated major-conference teams for a four-team playoff, the crystal ball is still pretty cloudy at this point. There’s a lot of football left to be played. But our model should at least be a useful guide for making sense of the silliness that’s surely waiting in the season’s second half.
Check out our college football predictions.