Determining the 'most deserving' teams

If college football were like pro sports, this whole “playoff” thing would be pretty easy: Just take the teams with the highest win percentages. Or take only conference champions, like division winners in the NFL. Why do you need a committee to pick teams when you could just use the conference standings and the “only stat that matters” – the win column?

As you know, selecting teams for a college football playoff is different. Level of competition can vary dramatically across teams and conferences, so performance has to be evaluated in the context of the schedule played. And with each team playing a dozen or so games each, how teams look in achieving their results – “style points” – can also help separate teams beyond their records.

To help determine the “most deserving” teams for the College Football Playoff, ESPN’s Stats & Information Group has developed some in-depth analytical tools to evaluate résumés. These metrics look “backward” at what a team has accomplished to date and are therefore categorically different from our Football Power Index. FPI is a forward-looking system that evaluates who is “most powerful” and helps predict specific matchups as well as the rest of the season. Some of ESPN’s own break down the difference between “best” (which we measure using FPI) and “most deserving” (where the résumé metrics come in) here.

As we get further into the season and the true playoff contenders start to come into focus, the two main metrics we’ll be using to rate teams’ résumés are Strength of Record and Game Control. These metrics are uniquely designed to evaluate how each team’s accomplishment stacks up against what everyone else has done to the same point in the season.

Strength of Record

Aptly named by “College GameDay” coordinating producer Lee Fitting, Strength of Record (SOR) measures how strong a team’s record is, given its schedule. It’s based on the chance of an average Top-25 team having the team’s record or better, given the opponents the team has played (and where they played) to date. A higher Strength of Record indicates a more impressive accomplishment – it means that the team’s W-L record was more difficult to achieve.

It should be fairly obvious that going 5-0 against Marshall’s schedule is much easier than going 5-0 against Mississippi State’s or Auburn’s schedule. That’s exactly the type of thing Strength of Record helps separate: Marshall ranks 26th; Mississippi State and Auburn are first and second. Several teams with one loss have a “stronger” record than Marshall’s 5-0, with each of the Thundering Herd’s wins coming against opponents ranked worse than 85th in FPI.

Strength of Record captures a team’s “quality wins” and “bad losses” from an opponent-strength perspective, but it’s not concerned with the team’s point margin or other ways of measuring dominance or luck. This particular metric doesn’t care if you had the game "in hand" but lost to your rival in a historically flukish way, or if you scored 36 points in the fourth quarter to squeak out a win on a Hail Mary – just whether you won or lost.

Game Control

But to some, it’s not just about winning and losing, but how you looked – again, “style points.” Final scores have often been used for this purpose, but those can be misleading.

With play-by-play data for all games, we can go much more in depth and look at each team’s chances of winning on every play of every game based on the score, time remaining and other parts of the game situation. By looking at those chances across all plays, we can calculate a team’s average in-game win probability, a measure of how exactly it went about winning (or losing) throughout games that goes beyond just the 12 or 13 final results.

As with going from basic W-L to Strength of Record, each team’s average in-game win probability gets translated to Game Control based on how hard it would be for a top team to achieve it, given the schedule. Game Control also ends up on a 0-to-100 scale, measuring how well a team controlled games from start to finish, accounting for the difficulty of the games it has played to date.

An interesting case from the early part of this season is Florida State, who had the highest Game Control last season. Although the Seminoles are undefeated again through five games, they have fallen behind and really struggled a couple of times, needing rallies to win against Clemson and North Carolina State. Game Control sees this and ranks the defending champs 21st in the nation so far.

Both Strength of Record and Game Control will become more informative as more games are played. Given the relatively few games in a college football season, the rankings should be pretty fluid from week to week. Check the Playoff Picture for the most updated résumé rankings along with FPI, AP Poll and Playoff Committee rankings (once they are available) throughout the season.