College football season is nearing its annual conclusion, which means it’s time for everyone to become overly invested in the College Football Playoff. Every year, the 12-person selection committee sets out to choose the “four best teams” to compete for the College Football Playoff National Championship. (This year’s picks are #1: Alabama, #2: Clemson, #3: Ohio State, and #4: Washington.)
The four teams are then paired off, and the two-stage playoff determines the champion.
The committee’s selection process of the top four teams is, as even the committee itself admits “an art, not a science.” “Art” is another way of saying subjectivity, and with subjectivity comes bias, intentional or otherwise.
The pertinent question: Could changes to the selection process help remove some of this subjectivity?
Americans love playoffs.
Playoffs in general aren’t uniquely American, but Americans certainly embrace the idea in their sports. Every major professional sport in the United States crowns its champion via a playoff system. And thousands — tens of thousands? Hundreds of thousands? — of fans fill out their college basketball brackets every March without so much as a minute of college basketball viewership during the season.
(The most notable crowning of champions-sans-playoff may be in European soccer; winners of these leagues are the teams with the best records over the course of the entire season.)
Insofar as playoffs are intended to distill a “true” champion from a group of teams, they suffer from problems similar to statistical sampling — namely, relying on a small number of outcomes may not accurately crown a “true” best team. Ultimately, commissioners of all leagues are in the difficult situation of keeping playoffs sufficiently short so as to keep fan interest high, but also sufficiently long so as to minimize as much as possible the randomness of the outcome.
To that end, given that playoff structures are inherently compromised, what could be done to tweak the current College Football Playoff structure so as to achieve a more sound process of crowning a champion?
A better system is possible.
The Bowl Championship Series — the ranking process that predated the College Football Playoff — tried to take a step in this direction by including computer rankings in the overall score generation process. Many fans seem to have an inherent distrust of computer rankings which I believe stems from the usually-protecting algorithms used to generate them; nevertheless, relying solely on a computer ranking eliminates a committee from the decision-making process.
I have long been a proponent of using the Gus Rankings to determine the rankings for college football teams, mainly because it would create a lot of favorable incentives that would allow college football teams themselves to iron out the features of current-day college football that fans find frustrating.
In brief, under Gus Rankings, teams gain points equal to the wins of the teams they beat, and lose points equal to the losses of the teams that defeat them. In college football, the four teams with the highest Gus scores at the end of the season could go into the playoffs.
Many of the same advantages of the Gus Rankings can also be achieved by utilizing a sports-specific version of an h-index. An h-index is a common metric for academics—an individual’s h-index indicates that she has h papers cited h number of times.
For a college football team, a similar idea (say, a w-index) would denote teams that have w wins against teams that themselves have won w games. This particular index would seem to be too broad to be useful — i.e., most teams would have a ranking between 2 and 8, leaving only seven distinct values for 128 FBS teams — but iterated versions of this ranking can help address that concern. If you’d like more detail, see this working paper.
Better systems would mean better incentives.
How would utilizing these ranking systems impact college football? The largest impact is clearly on scheduling.
Consider the incentives for scheduling at present. The goal is to be perceived as a strong team by a 12-person committee. As such, teams already viewed as strong have little incentive to risk losing this perception; the upside of scheduling and winning a risky game (confirming strength) is small compared to the downside of losing (revealing weakness).
Teams viewed as weak face the opposite scenario—they strongly desire a risky game, as the upside of winning (proving strength) easily outweighs the downside of losing (confirming a weak perception). As such, we see many teams in a position of perceived strength — think major conference members — schedule in a very conservative manner.
Using the Gus Rankings and/or an h-index throws all of this on its head. The most valuable teams to play under these systems are those teams whose record most overstates their true ability. The exact argument major conference teams use against including non–major conference teams in the playoff is that the true ability of these teams is masked by a gaudy win-loss record.
Instead of engaging in playoff keep-away with non-major conference teams, major conference programs would be wise to search out and schedule as many of these valuable teams as possible — that is, unless major conference teams are simply posturing and truly believe these teams to be strong. In any case, the on-field ability of non-major conference teams relative to major conference teams (and the beliefs of these major conference teams) should be revealed by the incentives of these ranking systems.
Consider: At the time of writing, Alabama and Western Michigan are the lone remaining undefeated teams in major college football at 12-0. Both teams would then offer 12 points in the Gus Rankings (or 12 wins in the h-index) to a team that could beat them; the same team would lose zero points for losing to either of them. With all due respect to the Broncos, which team would likely be the preferred opponent?
As with so many dynamic systems in society, developing the proper incentive structures leads individuals pursuing their own self-interest to generate desirable outcomes for the entire group. College football is no different.