A nerd's look at predicting the SEC, Part III: Turnover margin

Raise your hand if you’ve ever listened to one of those AM radio call-in coach shows.

The kind piped from SEC towns to syndicated radio stations all across the tiny, football-crazed communities in the Southeastern states.

Oh? That many of you?

Part I: Close games
Part II: Point differential

You’ve probably heard a number of SEC head coaches spew this line when asked about the important factors of any particular game (insert thoughtful, pondering tone here):

Well, Mr. Radio Guy, we face a tough matchup this week, and I have a lot of respect for the Tigers. When we play Auburn on Saturday, it’s going to be vitally important that we take care of the football. I think the team that makes the fewest mistakes has an excellent chance to win this football game.

You don’t think an SEC coach is ever going to say, “well, their boundary corner cheats up against the run on play-action, and the linebacker can’t cover tight ends, so we’ll start by attacking there,” do you? Other than the fact that those answers are dry and boring, there’s truth to them.

Across an entire season, SEC teams sometimes gain or lose an extra 15 possessions or more. That’s a huge advantage.

For example, the Georgia Bulldogs boasted a +16 turnover margin in 2014, best in the SEC. UGA also averaged an SEC-best 2.7 points per offensive possession. So, considering 16 x 2.7 = 43.2, Georgia’s ability to protect the ball and take it away on defense added a value worth more than 6 touchdowns through the course of the season.

(Points per possession stats don’t include extra points. Let’s say UGA scored six touchdowns on those 16 possessions. Now we’re talking about more than 49 points.)

That’s significant. How many other factors can cause a benefit of nearly four points per game?

To answer my own question, not many. When Georgia suspended Heisman Trophy candidate Todd Gurley a few days before the team faced SEC East rival Missouri on the road, Vegas didn’t move the betting line three points. UGA opened as a field-goal favorite.

The only problem with relying on turnovers in football is that, statistically, there’s an element of randomness to them, both in college and the NFL. Teams with an extreme turnover margin, either good or bad, are highly likely to regress to the mean the following season.

A team can be very good at creating takeaways and protecting the football, for example. But fumble recovery rates are very near 50 percent historically (more on that tomorrow), adding a strong element of chance to the overall turnover margin stat. Even interceptions tend to involve a strong variance, both for quarterbacks and defenses year-to-year.

Again, teams who enjoy a great turnover margin one year almost never do as well the next year, and vice versa.

To illustrate this statistical rule, one only has to examine the SEC teams that finished in the top three or bottom three in turnover margin in the 2012 and 2013 seasons, including ties. The results are pretty astounding.

Including ties, the 14 teams that fell into those categories followed this rule 100 percent of the time. All eight teams that finished at the top of the SEC saw their turnover margin suffer the next season. All six teams that finished at the bottom of the SEC saw their turnover margin improve.

Even more astounding is how drastically that has affected these teams. The average turnover margin among the bottom three teams in the SEC became positive the next season, while it reverted to neutral, or 0.0, for the top three teams.

NEXT-SEASON BEHAVIOR OF SEC TEAMS WITH EXTREME TURNOVER MARGINS, 2012-14

*Includes ties

Football is a difficult sport to figure out, even for the analytics movement. We know, for example, that Georgia’s turnover margin has a very, very strong likelihood of declining in 2015. But what we can’t predict is by how much.

Let’s say that UGA beats recent historical trends, and rather than reverting from +16 to +3, the Bulldogs still finish near the top of the league at +7. That’s a loss of seven possessions compared to last season. If UGA maintains its 2.7 ppp (points per possession) average this fall, the team is losing nearly 19 points just by chance.

How does this regression to the mean turnover phenomenon affect the SEC in 2015?

TOP 3 AND BOTTOM 3, 2014 SEC TURNOVER MARGIN

Taking a look at the SEC’s biggest outliers from the ’14 season, we find that the aforementioned Georgia Bulldogs and fellow SEC East team Vanderbilt lived on the extremes.

The good news for Commodores fans is that despite what could be another atrocious offense, Vandy is unlikely to turn the ball over seven times again in the season opener, as it did against Temple last year.

And Texas A&M’s defense may still remain at or near the bottom of the SEC West, but the Aggies — with a more aggressive scheme and Myles Garrett applying pressure on quarterbacks — surely will intercept more than five passes the entire season, as the team did in 2014.

Perhaps in part as a function of Vanderbilt’s expected improvement, teams like Georgia, Missouri and Kentucky may not seize the high number of extra possessions that those programs enjoyed in ’14.