Can College Football’s “Final Four” Be Selected Based on Stats?

You can already tell after Tuesday night’s formal announcement of the initial rankings of the playoff selection committee that the evaluation process of college football teams is going to be front and center from this point forward in the 2014 season.

Parts of the country which are happy with the rankings are going to make their case. Parts which are furious will make theirs. And, far too many voices will be cherry-picking results, facts, and stats to make their case the way lawyers would for their clients.

What if you had to act as a fair judge truly trying to determine the best four teams? Are there statistical categories that are so good as “indicators” that they should be used automatically? The illusion that “computers” should do the job was dashed in the past when it was realized that very few computers programs found agreement. (And, some were so out-of-touch that it was nonsensical! Too many computer programmers don’t know football!)

Can stats lead the way? Can stats save the day?

I want to deal with that topic today because of timeliness, and because there really weren’t many “misleading” final scores last week. A few here and there. But, those came in largely irrelevant games that won’t impact conference races or the rankings. So, let’s bypass weekly boxscore developments this time around and see if there’s a way to contribute to the championship process.

First, a few problems…

College Football Stats are Polluted By:

• Small sample sizes

• Widely varied schedule strengths

• Widely varied offensive styles (with play calling and pace)

• Widely varied approaches late in blowouts

• The tendency for pass-heavy teams to compile more yardage than run-heavy teams

• The tendency for elite offensive teams to win by bigger margins than elite defensive teams

• The lack of connectivity from different parts of the country

Most of you reading this are college football handicappers or bettors, and you know exactly what I’m talking about. You can develop a sense of teams, and an understanding of how styles match up against each other. But, if you’re trying to build a totem pole that accurately ranks teams, the margin of error is so large that you can’t honestly be certain. You can separate “very good” from “good,” and “mediocre” from “lousy.” The methodology just doesn’t exist yet to fully pin down differentials inside those categories. You can make educated guesses. Take two sincere analysts making good-faith efforts, and those educated guesses won’t align perfectly.

And, a factor that would become important if “stat performance” ever became a qualifying element, head coaches would start playing to the categories rather than just playing football! We know this because coaches and athletic directors are constantly saying variations of “We want to win a National Championship, so we’re going to do whatever it takes to win a National Championship.”

If the field of analytics found a great indicator stat that was adopted into the evaluation process, like yards-per-pass, third down conversion rate, or red zone defense…coaches would pull out all the stops to make sure those categories graded out well. In blowouts, starters would rush back onto the field for third down and three on offense, or to get a late red zone stop when up 48-7 with two minutes to go. We already have too many “running up the score” situations because coaches are worried about earning style points with voters. We don’t need anyone trying to run up yards-per-pass-attempt stats, or punting on third downs so they don’t get charged with a failed conversion.

Unfortunately, we’re already looking at “stats are very polluted,” and “emphasizing stats would just mess the sport up anyway.” If a set of stats ever is determined to be ideally suited for evaluation, the committee will have to keep them a secret to preserve the integrity of the game!

An intelligent use of stats should be part of the process though. A Final Four can’t be picked solely on won-lost records and scoring margins because those are already polluted by schedule strength, stylistic, and sample size issues. The dominance of the SEC over the past several seasons (and Florida State from the Southeast quadrant of the country last season) has shown that pass-happy, run-up-the-score, dominate-soft-schedule scoreboard darlings tend to get blown off the point of attack when facing balanced physical teams with great athleticism. We want to keep crowning real champions.

Some suggested starting points for getting stats into the mix:

• Only count stats accrued when teams are within 21 points of each other

• Reward balance over one-dimensionality

• Reward defense as much as offense even if it isn’t as pretty

• Reward finding the end zone over moving between the 20’s

• Reward playing tough schedules over playing weak ones

• Reward offensive TD’s vs. quality defenses more than others

• Reward defensive impact plays…takeaways, tackles behind line of scrimmage

• Focus on “per-play” rather than raw volume to equalize issues involving pace

In short, focus on the stuff that matters MOST in a way that best represents equal footing. Trim out superfluous stats attained when a decision was no longer in doubt. Be wary of one-dimensional teams whose one-dimension helps them create illusions vs. non-challenging schedules.

Honestly, no matter WHAT you do, the difference between #4 and #5 is going to be negligible…and everyone connected to #5 is going to be furious. But, if it were an 8-team tournament, the same would be true when comparing teams #8 and #9. A sweet 16 just pushes things down further. Dick Vitale blows a gasket every March a few dozen spots further down the ladder in college basketball when bubble teams are snubbed.

Focusing on REAL FOOTBALL being played AS A RESULT IS BEING DETERMINED will at least help create the most accurate estimates of team quality. Scrub out as much pollution as possible. Perfect ain’t going to happen.

See you again next Wednesday.

---

Jeff Fogle is a freelance writer living in Austin, Texas. He writes about college and pro football, college and pro basketball, and MLB on his blog StatIntelligence. You can follow Jeff on Twitter @JeffFogle.

---