How does Turnover History correlate to the turnovers in the next game?

The contrarian view is that turnovers are one of the least understood pieces of handicapping data, and often going against the team that has benefitted from turnovers thus far can be a successful road to wagering profits.

A key point to assess then to answer these competing perspectives, is whether indeed the previous turnovers forced and committed have much to say about what happens in future games.

What do the numbers mean? Well, basically it's telling us that there is virtually no correlation between what has happened before with turnovers and what is likely to happen this week! The correlation values can range from -1 to +1 with a +1 being a perfect correlation (what happened in the past will perfectly predict what will happen this week), and 0 being no correlation whatsoever. At ranges of .02 to .06 we are seeing next to no correlation.

Ah, but lest we end the article before we've really begun it behooves us to point out that predicting turnovers in say week two with just one game played might seem a tad different than predicting them in week ten when a team has had many more chances to show us tendencies. Therefore we'll look at the above on a week by week basis:

Do the correlations improve the later in the season you get? Well, not much. The lesson to be learned then is that 'turnovers are not to be trusted' -- in other words, a team's turnover history won't be much of a giveaway as to how they will perform in this area in the future. At the same time there may be some other effects to turnovers: perhaps a team that has thrown many interceptions will become more cautious in the passing game, etc and these could have some fairly profound implications. Still, that's a turnover study for another day...