Google+

The impact of turnovers on winning football games

Want to win games in the NFL?  Avoiding fumbles and other turnovers can go a long way.

Want to win games in the NFL? Avoiding fumbles and other turnovers can go a long way.

With offseason programs behind us and training camp still a month away, this is truly the doldrums off the NFL offseason. I like to use this time (well, really the entire offseason) to try and improve my football knowledge by looking at league-wide trends such as quarterback development, rookie wide receiver impacts, and the importance of home-field advantage.

I am excited to continue that tradition this year, and a logical topic submitted itself after a recent round of discussions here at WCSN. We have spent a great deal of time lately—both in articles and in comments—talking about turnovers and the impact they have on teams winning and losing games. I decided it would be fun and informative to look up just how big of an impact turnovers have on a teams’ chances of winning.

The answers that I got were not exactly what I expected, but did provide some insight.

Season-long impact

I began my inquiry by looking at teams’ records over a 16-game NFL season and how they were influenced by their turnover differential (the number of takeaways minus the number of giveaways). I looked at all results from the last five years, which gave me 160 data points.

Unsurprisingly, there was a positive correlation between turnover differential and wins (r2=0.36), which you can see in the graph below. For comparison’s sake, the correlation between complete quarterback rating and team wins is 0.33, so turnovers have just as much impact on a team’s season as the quarterback’s performance.
turnover wins

Figure 1. There is a positive correlation (r2=0.36) between an NFL team’s turnover differential and its regular-season win/loss record.

Note, however, that teams have been capable of having successful seasons while struggling with turnovers, highlighted by the 2011 Pittsburgh Steelers, who went 12-4 despite having a -13 turnover differential. Conversely, teams can struggle to win games despite having a positive turnover differential; this happened to the Kansas City Chiefs in 2008 when they had five more takeaways than turnovers but still went 2-14. Clearly, turnovers are important to a teams’ season-long success, but they are not the only thing that matters.

Game-by-game impact

I next wondered how much of an impact turnovers had on an individual game. In order to examine this, I looked at every game from the 2012 season. There were 256 games played in the regular season. One ended in a tie, while 51 featured the same number of turnovers from both teams. That leaves 204 games which featured a winner and one team turning the ball over more than the other. The results are broken down in the table below.

TO DiffWinsLossesWin %
> +255395
+2431081
+1633068
0515150

In those 204 games, the team that won the turnover battle went 161-43, a winning percentage of 79%. Splitting those numbers up to account for the size of the turnover differential could be useful here. Teams that had a turnover differential of +1 won 68% of the time (roughly two out of three games), a number that increased to 81% for teams that had a turnover differential of +2. When teams had a turnover differential of +3 or better, they won 95% of the time. Looking at the playoffs over the last five years (46 games with a non-zero turnover differential), those numbers are fairly comparable, albeit slightly lower at most levels, although that could be due to a smaller sample size.

NFC North application

One interesting thing we can do in order to apply this to the NFC North is to look at the expected wins for each team in the last two years based on their turnover differential in each game using the probability of winning a regular season game based on their turnover differential alone — so 95% if a team has a turnover differential of +3 or greater, or conversely 5% if its turnover differential is -3 or worse).

The table below shows the expected results for all four NFC North teams over the last two years, compared to their actual number of wins.

TO DiffChicagoDetroitGreen BayMinnesota
> +26361
+25653
+165910
05477
-13436
-26222
<-21803
Expected Wins18152116
Actual Wins18132614

We can see that the predicted results are not very far off from what actually happened. Chicago exactly matched their predicted win total, Green Bay came in five games higher than projected, and Minnesota and Detroit both won two fewer games than they were expected to.

Conclusions

For those of you who didn’t bother reading the article, I hope you’ll at least take the time to stop here to see the conclusions before going down to the comments. I present them here in easy-to-digest bullet point:

  • Roughly a third of a teams’ season-long success can be explained by its turnover differential.
  • The team that turns the ball over less will win the game roughly 80% of the time.
  • Odds of winning a game range from 68% for a turnover differential of +1 to 95% for having a turnover differential of +3 or greater.
  • These odds are fairly steady in the regular season and playoffs.

Feedback

What did you think of this article? Do you have any questions about my data or conclusions? Was there an important factor I forgot to consider? Do you have an idea for another topic I could look into? Let me know in the comments below, or send me an email.

1890 comments
MIBearFan
MIBearFan

The Aaron Hernandez situation could be a mess.

G & G
G & G moderator

"My team will destroy the rest of our division!!"

4-12 team fan

 ✔

The June 2013 study from the American Journal of Orthopedics looked at 10 confirmed cases of ruptured hamstrings in the NFL from 1990 to 2008.

While nine of the 10 players in the study returned to play the next season, only five of the 10 played more than one game following the injury.

In addition, two of the 10 players returned and had major injuries occur to the same leg. One suffered a torn Achilles tendon and another re-ruptured the same hamstring (the player was a linebacker). On this note, it is worth noting that Bishop has already been dealing with a strain in the same hamstring he injured in August.

LKP
LKP

@Green and Gold Maybe the biggest fluke of all right there. Green Bay is so fucking lucky they won that game

LKP
LKP

@MIBearFan That plus Gronk plus their lack of WRs. If Brady can pull that off, I'll be impressed

LKP
LKP

@Green and Gold 10-6 previous year and 4-12 only due to fluke and injuries

2012 really went like this without the flukes

Detroit 9-7 4-2 division

Green Bay 9-7 3-3 division

Minnesota 9-7 3-3 division

Chicago 9-7 2-4 divisioin

LKP
LKP

@Mr._Horse How do you feel your LB pass coverage will be with Hawk, Jones and Perry? Hawk -2.5, Perry -2.3 and Jones +4.6 last year. Jones -2 plus the previous 2 years.

MIBearFan
MIBearFan

@Mr._Horse  

Buyer beware. You had to figure the Packers wouldn't dump him if they didn't have concerns about his health.

LKP
LKP

@Green and Gold Durham running the wrong route, Hanson missing a field goal tied 17-17. Such crap that Green Bay won that game

MIBearFan
MIBearFan

@LOLebron @MIBearFan  

I agree, and I assume they will remain silent on the matter until they have the facts, but this isn't Joe Blow on the street. He is an NFL player, belonging to one of the Premier franchises. The media will be swarming. If nothing else, it creates a hell of a distraction. I guess it's better it happened now while things are relatively quiet.

 ЯD
ЯD

@MIBearFan I don't think so. Silence in any legal situation is always the easiest and most productive option. Answering questions is simply a no-win scenario: either you're not forthcoming enough, or you say something you later regret. Everyone will always suspect your motives.

 ЯD
ЯD

@MIBearFan It would be just as bad to assume he's guilty and find out later he was innocent all along. Silence is the best option right now.

LKP
LKP

@MIBearFan @2012 Was a Fluke is Fact @Green and Gold No they are not. The plays were not normal. You see throwbacks all the time? I've never seen Pettigrew fumble before. That Bell fumble was ridiculous leading to the pick on 4th down even more ridiclous, 2 special teams back to back games never happened before, and off Schefflers hands plus a block in the back and Staffords fumble was the flukiest play ever.

Nothing like that has ever happened before

MIBearFan
MIBearFan

@2012 Was a Fluke is Fact @MIBearFan @Green and Gold  

Have you ever watched any football other than the Lions? If you have, than you should know that games are decided by plays you call flukes all of the time. Good teams just end up on the winning end of these game s more often than bad teams.

LKP
LKP

@Green and Gold @MIBearFan Weak year is a fact in 2010. Edwards, Henne, McNabb, Kolb, Stanton, Hill, Troy Smith, Kitna, Sanchez. Congrats on nothing Green Bay

LKP
LKP

@MIBearFan @2012 Was a Fluke is Fact It's not always a linear progression. Jones and Hawk have had several years of bad coverage and Perry is very raw still missing most of last season. Sorry the high probability is that they are bad in coverage. It's not trolling, it's truthing

MIBearFan
MIBearFan

@2012 Was a Fluke is Fact @MIBearFan  

That's just trolling bullshit. They should naturally improve over last year due to experience. Also, improvements on the D-line causes increased pressure in the passing game, and that always helps the pass coverage.

LKP
LKP

@MIBearFan But you had to make your Lion comment. Pack LB coverage could very well suck. Can't wait to find out. Stupid offseason

MIBearFan
MIBearFan

For the record, I'm not a Packer fan, so I couldn't give a crap one way or the other.