Stats That Matter: Turnover Margin

There is nothing controversial about "turnover margin" as a statistic that matters, except that it came in a fairly anemic 14th place among the 27 stats we studied in the offseason. That just doesn't sound right, does it? Shouldn't it be higher?

In a well-researched piece by Tom Dienhart ("Coaches pick the most important statistic"), during which he interviewed 29 head coaches and/or coordinators, 20 of those coaches identified turnover margin as at least one of the most important stats. Among fans I'd expect the ratio of fans that believe that turnover margin is the most important stat would be even higher.

What is going on here?

For starters, it would be possible to have a negative, positive, or neutral turnover margin for a season, yet also identify turnover margin as the major factor in a team's win/loss record. Turnovers are obviously a stat that matters, but you won't appreciate their full effect by looking at the aggregate numbers for a season.

In 2010, Miami and Duke, in the aggregate, had identical -0.67 turnover margins. Miami went 7-6, 5-3 in the ACC, while Duke went 3-9, 1-7 in the ACC. From this perspective, turnover margin had little bearing on winning and losing.

On the other hand, in Miami's wins it had a positive turnover margin of 1.57, but had a turnover margin of -3.17 in its losses. If you were to ask anyone on last year's Miami's staff what was the difference between winning and losing, I can all but guarantee "Turnovers" would be the response.

In fact, as you go through the ACC in 2010, not a single team didn't fare worse in turnover margin in losses than it did in its wins.

Basically what we are tracking here is the flawed nature of examining football stats in the aggregate generally, and aggregate turnover margin numbers specifically. On the other hand, we lose our way if we decide that "flawed," equals "useless." There's a lot I think we can learn from looking at the aggregate data, even when it comes to turnover margin.

For example, Phil Steele includes a factor in his magazine every year that he calls "Turnovers = Turnaround." He has looked at teams every year since 1996 that had, on the season, a positive double-digit turnover ratio and found that those teams had the same or weaker record 74.4 percent of the time, and 64.2 percent had weaker records. Conversely, he's found that those teams who had a negative season double-digit turnover margin to have the same or stronger record 81 percent of the time, and 69 percent improved their records.

The implication is that turnover margins that are either exceptionally high or exceptionally low will regress to the norm and have an impact – positive or negative – on the following season's record.

Virginia Tech, however, appears to be one of the exceptions.

Over the last five years the ACC has had five top 25 finishes in turnover margin for a season against BCS teams: Virginia Tech has recorded three of them. In 2010 Virginia Tech notched the best season-long TO margin against BCS teams of any BCS team, recording a full 2.0 turnovers more per game than its BCS foes.

Even by Virginia Tech's standards, that's an extremely high turnover margin. The program's second best mark came in 2008, when the Hokies recorded 1.18 more turnovers per game than their BCS opponents. I think it is reasonable to expect some of Steele's regression to the norm with Virginia Tech next year.

Surprisingly, as much as UNC's defense has improved over the past several years, the turnover margin has not.

The Tar Heels have had only one positive turnover margin in the past five years, and all of their four negative turnover margins were in the bottom half of the 320 recorded turnover margins by the 64 BCS teams in the past five years.

Check back on Friday for the final installment in this series ...

Inside Carolina Top Stories