The Enigma of NFL Turnovers


There are so many cliches about turnovers in the NFL. “Win the turnover battle and you win the game.” “The game is decided by turnovers.” “If you turn the ball over, you’re going to lose.” Those are just a few, there are tons more. You hear them from the announcers during every single game. You hear them from the talking heads before and after every game.

What if I told you that turnovers, both takeaway and giveaways, have almost no correlations to wins and losses? You’d probably reply that I’m crazy. The belief that turnovers are huge factors in determining win and losses is so ingrained that we don’t even question it.

Take it one step further. Good teams take the away from other teams, and bad team tend to turn the ball over a lot. You agree with that, right? I’m just stating the obvious here. Except that I’m not. It turns out that in fact turnover are almost completely uncorrelated to winning. Check out this graph of the data from the last 10 years:

Check out that r-squared value. 0.184. Yikes. It doesn’t get much clearer than that. There simply isn’t any correlation between turnover differential and wins.

Multiple people on twitter suggested that either takeaways would correlate to wins, or giveaways to losses. Basically suggesting that one of the other would work, but not both. Since for every giveaway there’s a takeaway, and for every win there’s a loss, it just doesn’t work that way. But, since people have already suggested it, and I know how stubborn people can be, I decided to run the numbers anyways. Have a look:

I know that by using total wins and losses and turnover differential masks the effect on any one game. Unfortunately, I can’t run the stats on a per-game basis at this point. I’m working on that, and will do so in the future once my data base has that info.

Still, the idea that good teams force turnovers, and bad teams give the ball away a lot, and that those turnovers have a huge influence on wins and losses, is an idea that we need to put to rest. As counter-intuitive as it seems, it appears that turnovers have little influence on a wins and losses for a team.