The Enigma Of Turnovers - Reprised

Use your ← → (arrows) to browse

Almost a year ago, I began an overly ambitious project to work out what stats were the most important in determining wins and losses. It was a project that I haven’t had time to finish, but I do have some very interesting results, most of which have never been published.

Clearly the most eye-opening results were those about turnovers. Comparing wins to turnover differential was perhaps the most unexpected Null I’d ever seen. I referenced this series of articles a couple days ago, but I’d published the original results so long ago I couldn’t even easily find a link to point you them. So I decided to sum up those articlesm giving you the cliff-notes version.

In case you  aren’t familiar with what R² means, this shows that there is absolutely no correlation between wins and turnovers. This result completely busts the myth that good teams tend to force turnovers while bad team tend to turn the ball over a lot. It just isn’t true.

The problem here is that I was looking at the effect on wins from turnovers over the course of an entire season, and not on any one game. To do that, I looked at every single game played in the NFL over a 10 year span, and the results were equally interesting. Just examining the data, pre-analysis demonstrated something interesting.

While the team that won the turnover battle did win more games than the team on the other side of the ledger, it was clearly not the all-important stat that the NFL pundits would have you believe it is.

Another thing that is common heard by so-called football experts is the “fact” that turning the ball over with automatically lead to losing. That isn’t the case either. The winning team averaged just over one turnover.

The best way to determine the true importance to turnovers on the result of one game is to look at how the probability of winning is effected by them. Here is the first and only time that the data indicates that turnovers are even the least bit related to wins and losses.

Continued on page 2. click the link below to continue reading. (the best of the results are on page 2!)

Use your ← → (arrows) to browse

comments powered by Disqus