Varsity Numbers Talks Turnovers

by Bill Connelly

Through the years, Phil Steele [1] has carved out a niche as one of the most unique, thorough college football writer-slash-prognosticators in the country. He is known for taking risks with his picks. Some pay off, some don't (his pick of Auburn winning the SEC West in 2008, for example, backfired just a bit), but he is not afraid to look at his numerous power rankings and take some chances.

Steele is possibly most well-known at this point for his identification of certain year-to-year factors that signal improvement or regression, the most significant being his work with turnovers. Turnovers are arbitrary enough that, if your turnover margin is too low or too high in a given year, chances are that the numbers will turn around the next year, as will your fortunes in the win-loss column. (Football Outsiders' work with recovered fumbles at the NFL level is indicative of the same sentiment.)

Well what if we dial in on turnovers even further? What will that show us? In the second ever Varsity Numbers column [2], we introduced a measure called Turnover Points Margin.

[C]ounting the significance of two values -- the team's field position when the turnover happened, and the opponent's resulting field position -- gives you a much better view of a turnover's true costliness. And that's what we're going to try to do.

[3]

Let's look at two turnovers:

• Turnover 1: Team A is about to score. Their running back carries the ball inside the 1, then fumbles. The ball bounces into the arms of a safety, who carries it 100 yards for a touchdown.

• Turnover 2: Team A's quarterback fumbles at his 45-yard line, and Team B recovers. Team B then immediately goes three-and-out and punts the ball back to Team A.

Using my numbers, Turnover 1 was worth 12.62 points (5.62 points for being at the opponent's 1, 7.00 points for being returned for a touchdown). Turnover 2 was worth 4.26 points (1.92 points lost/prevented, plus 2.34 points given/taken). Is that not a much more accurate read of which turnover truly impacted the result of the game and which did not?

In theory, turnover margin is predictive despite the fact that it doesn't differentiate between truly costly turnovers and minor ones, like an intercepted, end-of-half hail mary. So if you came up with a more descriptive turnover measure, wouldn't it be even more predictive? To find out, let's dive into some numbers. Below are the top and bottom teams according to turnover points margin in 2007, along with their proceeding 2008 margins.

All ten of these teams saw a worse turnover points margin in 2008, while all but three fell to +3.0 points per game or worse. Six saw their margins swing by double-digit amounts. Think about how much turnovers made a difference to a team like LSU, who won the 2007 national title benefiting from a ten-point advantage, then threw a series of well-publicized pick-sixes in 2008 and saw their overall turnover points margin fall by a full two touchdowns.

As you would expect, we see a pretty similar trend looking at the bottom teams. All ten at the bottom improved, six by double-digits, and five all the way into positive margins. A poor margin cost Georgia Tech dearly in 2007, as they finished 7-6 and got head coach Chan Gailey fired; in 2008, their margin swung dramatically in the other direction despite a whopping 36 fumbles (20 lost) while getting their wits in Paul Johnson's flexbone system. Meanwhile, teams like Baylor and Florida International improved rather significantly in 2008, in part due to better results in the turnover department.

So those were the extremes. Let's look at turnover points margin by different groupings of teams.

Here is where things get pretty dramatic. When you group teams by their 2007 margin, all categories regressed or improved to within +/- 2.0 points per game in 2008. Half were within +/- 0.3 points per game. This very clearly suggests that, on average, teams will regress to zero from one year to another. If you were +5.0 points per game in 2008, you can probably count on regressing by about five points this year. It's the same story if you were -3.0 points per game or +9.0 points per game. Clearly this will not happen to every team, but overall things will even out pretty significantly.

There is still solid regression to the mean here, but you can possibly see a trend emerging. If you believe that some turnovers really are somewhat caused by talent or a lack thereof, then you will find it interesting to note that the top 20 teams according to S&P+ did not regress to zero as much as others, while the bottom 20 teams did not improve to zero as much. (The middle 80 teams were basically a wash.) This does support the theory that the best teams will usually finish a bit ahead of the game (by about two points, apparently) in terms of turnovers, while the worst teams will always be fighting an uphill battle -- that's why the best are the best and the worst are the worst.

With that in mind, let's take a look at last year's turnover winners and losers and, with their 2008 S&P+ rankings in mind, explore what it might mean for 2009.

First of all, it is interesting to note that 2008's highs and lows were not as extreme as those from 2007. With more data, we will be able to better determine which was the outlier of the two years.

With the general perceptions of Beamer Ball, one can assume that Virginia Tech is probably quite strong in the turnovers department more often than not, but being best in the country in turnover points margin is probably not a good sign for the following season, and even if Beamer Ball results in quite a few turnovers, the Hokies will likely see a regression of six to seven points, at least, in 2009. For a team that played seven games determined by a touchdown or less in 2008 (they went 4-3), that could swing things against them and hurt their hopes for a third straight ACC title.

You see quite a few high-S&P+ teams on this list -- Oklahoma, Florida, TCU, Alabama, Ohio State, Penn State, USC, Boise State -- and while the margins weren't as extreme and teams like Penn State and USC probably will not fall much (then again, if Matt Barkley is as interception-prone as he was his senior year in high school, then USC's margin could swing violently in the wrong direction), this does suggest that turnovers might ding Oklahoma and Florida more than they did last year.

It might also be important to note the top-ranked teams that did not benefit greatly from turnovers: fifth-ranked Texas (+0.9/game in 2008), 10th-ranked Missouri (+0.3/game), 11th-ranked Oregon (+3.3/game), and 12th-ranked Georgia (+0.1/game) might benefit just from not regressing much in this category (then again, winning the turnover margin certainly didn't help Oregon much last night in Boise).

Some pretty pathetic teams populate the top (or bottom) of this list, but scattered among the lowly teams are a few other squads to note: Nebraska improved from minus-9.5 points per game in 2007 to just minus-5.4 in 2008, but they were still on the wrong end of the spectrum, and after going just 1-2 in games decided by a touchdown or less last season, they could be on the right end of close games this year. Meanwhile, Arkansas' already-strong turnaround potential is given a five-point boost by their presence on this list, and LSU and South Florida join Nebraska as teams with terrible turnover margins that still managed to squeeze into the S&P+ Top 25.

In all, it wasn't hard to predict that the year-to-year change in turnover points margin would average out like it did last year, but it was still impressive just how much it evened out. Turnover points margin ascribes a specific point total to the likelihood of a turnaround, something that should excite fans in Lincoln, Fayetteville, Madison, Baton Rouge and elsewhere, while striking at least a little fear in the hearts of fans in Blacksburg, Norman, Colorado Springs, and Gainesville.

College football fans: Don't forget our new college football game discussion board [5] for talking about games in progress all day Saturday.