Turnover Margin and Wins: Is there a Correlation? Analyzing College Football Turnover Margins Since 2006
If you ever read articles from Phil Steele, he has a theory that a team that has forced double digit turnovers in excess of the times they turned the ball over will either equal their win total the following year, or will decrease their wins. As a corollary, a team who has turned the ball over double digit times more than they forced turnovers, will at least equal their win total the following year, or will increase their wins.
If you are like me, you are at least skeptical of this theory. So I went ahead and ran the numbers, going back to the 2006-2007 season. We looked at the team’s win total in 2006, their turnovers in 2006, and their win total in 2007. If the team had net double-digit turnovers, either to the positive or the negative, they came into the population study. Then a comparison was made between whether the team should have improved or at least stayed the same. Note that when looking at win totals for each year, I am only including regular season win totals, so no bowl games or conference championships are included.
If you want to see all of the charts since the 2006-2007 season, visit my website. For sake of brevity, I’m only going to include the 2010-2011 chart here. I’ll cover the 2011-2012 chart in a few weeks.
| Team | Conference | Net | 2010 Wins | 2011 Wins | Win Difference | Correct? |
|---|---|---|---|---|---|---|
| Virginia Tech | ACC | 19 | 11 | 11 | 0 | YES |
| Tulsa | C-USA | 17 | 9 | 8 | -1 | YES |
| Army | Independent | 16 | 6 | 3 | -3 | YES |
| Ohio State | Big Ten | 15 | 11 | 6 | -5 | YES |
| Maryland | ACC | 15 | 8 | 2 | -6 | YES |
| Wisconsin | Big Ten | 14 | 11 | 10 | -1 | YES |
| Oklahoma | Big 12 | 14 | 10 | 9 | -1 | YES |
| Iowa | Big Ten | 13 | 7 | 7 | 0 | YES |
| Oregon | PAC-12 | 13 | 12 | 10 | -2 | YES |
| Stanford | PAC-12 | 13 | 11 | 11 | 0 | YES |
| Alabama | SEC | 12 | 9 | 11 | 2 | |
| Connecticut | Big East | 12 | 8 | 5 | -3 | YES |
| Oklahoma State | Big 12 | 12 | 10 | 11 | 1 | |
| Hawaii | WAC | 12 | 10 | 6 | -4 | YES |
| Northern Illinois | MAC | 11 | 10 | 9 | -1 | YES |
| Missouri | Big 12 | 11 | 10 | 7 | -3 | YES |
| Toledo | MAC | 11 | 8 | 8 | 0 | YES |
| Miami (OH) | MAC | 11 | 8 | 4 | -4 | YES |
| Middle Tennessee | Sun Belt | -19 | 6 | 2 | -4 | |
| Cincinnati | Big East | -15 | 4 | 10 | 6 | YES |
| Memphis | C-USA | -13 | 1 | 2 | 1 | YES |
| New Mexico | Mountain West | -12 | 1 | 1 | 0 | YES |
| Texas | Big 12 | -12 | 5 | 7 | 2 | YES |
| SMU | C-USA | -12 | 7 | 7 | 0 | YES |
| Fresno State | WAC | -11 | 8 | 4 | -4 | |
| UCLA | PAC-12 | 11 | 4 | 6 | 2 | YES |
| Eastern Michigan | MAC | -11 | 2 | 6 | 4 | YES |
| Duke | ACC | -11 | 3 | 3 | 0 | YES |
| Central Michigan | MAC | -11 | 3 | 3 | 0 | YES |
| Michigan | Big Ten | -10 | 7 | 10 | 3 | YES |
Since the 2006-2007 season up through the 2010-2011 season (five seasons), I reviewed the turnover differential. Here are the results:
58/76 with double-digit turnovers to the positive either won less games or stayed the same (76.3%).
55/64 with double-digit turnovers to the negative either won more games or stayed the same (85.9%).
113/140 total followed trend (80.7%)
In case you were wondering the stats from the 2010-2011 season:
16/18 with double-digit turnovers to the positive either won less games or stayed the same (88.9%).
10/12 with double-digit turnovers to the negative either won more games or stayed the same (83.3%).
26/30 total followed trend (86.7%)
February 7th, 2012 at 9:50 AM ^
"If you ever read articles from Phil Steele, he has a theory that a team that has forced double digit turnovers in excess of the times they turned the ball over will either equal their win total the following year, or will decrease their wins. "
Maybe I'm reading this wrong, but shouldn't this say increase instead of decrease? Aren't you saying that a team with a double digit turnover margin in the positive will equal their wins or increase them in the following year?
February 7th, 2012 at 9:57 AM ^
It reads correctly.
Basically, the idea is a team who forces a net double digit turnovers to the positive in one year will not force as many the next year and their win total will decrease. For example, say Michigan State forced a net 12 turnovers this year; they would expect to decrease their regular season wins from 10 (or, in the best case scenario for them, maintain the same).
By contrast, a team who turned the ball over a net double digit amount should expect their win total to increase. For example, say Michigan State gave up a net 12 turnovers this year; they would expect to increase their regular season wins from 10 (or, in the worst case scenario for them, maintain the same).
Obviously the theory isn't 100%, but there is a pretty strong correlation.
February 7th, 2012 at 10:43 AM ^
I see. So it's more or less suggesting a regression to the mean? Thanks for clearing that up.
February 7th, 2012 at 10:48 AM ^
Not a problem. Probably should have just said regression to the mean.
February 7th, 2012 at 10:35 AM ^
This isn't terribly shocking, but it's nice to see it comfirmed. I'd be interested in seeing what the data looks like at various other points too. I don't imagine there's anything magic about the number 10 - the effect should remain at a -9 net turnover stat, just less so, right?
Would be nice to see how much less.
February 7th, 2012 at 10:47 AM ^
Thanks. Once I get some time I'll run the numbers. Its hard to believe that there would be a difference between 10 and 9, but at some point, there has to be a difference. The question is where there ceases to be some sort of correlation.
February 7th, 2012 at 3:10 PM ^
I started looking at TOs in 2009 and Phil Steele is wrong on several accounts.
http://mgoblog.com/diaries/turnover-analysis-part-2-do-turnovers-turnar…
1. Turnovers are not primarily luck. Many teams have positve double digit TOs for several years in a row and many others have negative double digit TOs for several years in a row (I'm looking at you M in 2008-2010). Luck does not explain this.
2. The teams Steele isolates (those with double-digit turnovers) are the teams whose TOM is primarily due to performance and not luck. Therefore his basic premise is incorrect.
Even if the double-digit TOM was primarily due to luck, there is no cause and effect. The percentage of teams that “turnaround” the next season is approximately the same when TOM is completely ignored.
From 1999 to the present, 72% of all FBS teams that had a winning record of at least +2 (7-5 or better) had the same or weaker records the next year regardless of TOM. This includes approximately 50 teams each year. (Steele: 77% for the teams with double-digit turnovers.)
From 1999 to the present, 74% of all FBS teams that had a losing record of at least -2 (5-7 or worse) had the same or stronger records the next year regardless of TOM. This includes approximately 43 teams each year. (Steele: 80% for the teams with double-digit turnovers.)
Basically, it is very difficult for winning teams to keep on winning at the same rate and very difficult for losing teams to keep on losing at the same rate regardless of what TOM happens to be.
February 7th, 2012 at 4:11 PM ^
I am not sure if this really says anything.
113 out 140 followed the trend, but so what? We are not exactly talking about 50/50 to begin with. The key here is that he predicts the teams to do worse or THE SAME. It would be pretty shocking if that was not true as most of the teams who had high TO margin also had pretty good records. Just factoring in the fact that they are likely to regress to average would say that statistically they are MUCH MORE likely to do worse or the same.
That is like saying sky will likely to be the same color as today. Not exactly very enlightening or useful.
What would be actually interesting is if you compare the result to the teams with similar winning percentages, but who did it WITHOUT high turnover margin and see if they perform any differently than the ones with high turnover margin.
Without that comparison, I don't think this study proves anything other than everything regress towards average.
February 7th, 2012 at 4:11 PM ^
The theory works out better if scheduling is taken into account. A team will average +1 in turnover margin for every 23 points it is favored by. You might want to re-run the numbers after taking this into account.
February 7th, 2012 at 5:54 PM ^
Not to pick nits here, but facts are facts, not theories. Steele (and Maize-in-Spartyland) are not "theorizing," they are pointing out a fact: teams with double digit TO margins tend to lose more games the next year. This is not something you can argue with, assuming they have compiled the data properly (which I assume they have).
A theory is an explanation for that fact. Enjoy Life's post is a bit confusing but in essence proposes the notion that overall team quality is the main issue, as opposed to luck, and that therefore this year's TO margin is either a) generated by team quality or b) irrelevant for next year's outcomes.
I don't think anyone would argue that there are other things going on beyond TO margin when it comes to predicting wins and losses. On the other hand, no one would seriously argue that - whatever the cause - a highly positive TO margin helps add to your win total.
The problem with Enjoy Life's argument is that he mistakenly writes off the potential importance of TO margin. In fact, it doesn't matter, as he seems to argue, whether TOs are a result of performance or luck.
It stands to reason that if you are either really good OR really lucky with TO margin in a given year, you are unlikely to be either than good or that lucky next year. Thus, beyond your other issues, given the resulting regression to the mean for TO margin specifically, you would expect life to be harder...ie, less wins. This is really not even arguable: when you get less TOs, you will, on average, win less games. The question then boils down to: what causes TO margins to be positive, quality or luck?
Of course, it will always be tough to answer that question for two reasons. First, if TO margin can be influenced by team quality, as Enjoy Life argues, then there will always be a a correlation between quality and TO margin, making it hard to discern whether wins are due to quality or TO margin. Second, since we measure team quality by wins, there is a circular argument problem: if you only measure quality by wins, then your independent and dependent variable are the same thing: not good science!
Comments