The Correlation Between Team Stats and Offensive and Defensive Efficiency, Part 3: Turnovers
Today we're continuing our exploration series on a variety of pace-adjusted team statistics and how they relate to offensive and defensive efficiency. Originally, this series was going to have five installments – steals, blocks, rebounds, fouls, and pace. But due to some reader suggestions, I’m adding a sixth: turnovers.
What makes turnovers particularly interesting in this study is that there are two very distinct types: live-ball turnovers and dead-ball turnovers. A live-ball turnover is a steal, where the other team can continue the play, usually at an advantage. A dead-ball turnover – traveling or an illegal screen, for example – stops the play, and therefore has a different effect on offensive and defensive efficiency.
We’ll look at these the same way we have the first two parts of the series, except I’ll break it down to both live and dead-ball turnovers as well. Let’s start with offensive efficiency.
How Turnovers Affect Team Offenses
I’ll give a brief reminder of how to interpret these graphs and then we’ll dive into them. The equation on the right is useful, as the number next to “x” is exactly how the team stat affects offensive efficiency. On the first graph below, a live-ball turnover in the past two years has decreased offensive efficiency (points scored per 100 possessions) by 0.01. The R² value tells us how well the linear trendline fits the graph – at exactly 1, it’s a perfect fit. The data would fall right on the line and the line would be the most meaningful. The lower the number, the less the trendline means, and the more likely it is that there is no correlation.
As you can see, there is no correlation between live-ball turnovers and offensive efficiency, at least in the last two years. The R² is essentially zero, which backs up that claim. However, when expanded to the last 10 years of data, a trend emerges.
There is very distinct trend that shows that live-ball turnovers significantly affect offensive efficiency. In the past 10 years, a single live-ball turnover has led to a decrease in offensive efficiency by 1.8. The R² value is high enough where we can infer some correlation. However, the question here is whether live-ball turnovers affect offenses, or whether it’s just turnovers in general. Let’s see what the data says about dead-ball turnovers and find out.
It’s interesting here that dead-ball turnovers over the last two years have had a more negative effect on offensive efficiency than live-ball turnovers. One would think that they would more-or-less be the same – a turnover is a turnover. The effect should really be on the defensive end, where the difference between a live-ball and a dead-ball is drastic considering the expected points on a fast break.
However, it normalizes a bit once we expand to 10 years. A dead-ball turnover has decreased offensive efficiency by 1.39 in the last decade, which is lower than live-ball turnovers, but not terribly so. The R² value of 0.1 is close enough that we can also infer correlation.
Some interesting findings so far, and we've seen similar results from both type of turnovers. Let's see if that continues when evaluating defense.
How Turnovers Affect Team Defenses
The two-year data is a little shocking – the graph shows that recently, there has been no correlation between defensive efficiency and live-ball turnovers.
The 10-year graph shows slightly higher correlation, but it’s still not as significant as we would think. A live-ball turnover in the past decade has decreased a team’s defensive efficiency by 0.24. In other terms, a team has given up 0.24 points per 100 possessions for every live-ball turnover.
As you can see, the R² value is very low and infers that even this small correlation might be overstated. It seems intuitive that giving up live-ball turnovers would really harm your defense – potentially more than any other team statistic. When an offense has the ball stolen, the defense (then turned into offense) generally has a fast break opportunity, and often a man-up advantage.
Ian Levy, in an article on fivethirtyeight.com last May, had an interesting stat: “...teams [in the 2013-2014 season] had an effective field goal rate of 61.5 percent on possessions after a live-ball turnover, compared to 46.5 percent after a dead-ball turnover.” How can we reconcile this research with our graph above?
Perhaps a live-ball turnover has a drastic effect on the following possession, but it’s limited when grouped in with a large number of possessions? The Miami Heat had 15.8% of their possessions end with a turnover of some sort, so it seems odd that 15.8% wouldn’t be high enough to make a significant difference on the defensive end.
The data gets even weirder. In the last two years, giving up a dead-ball turnover has actually increased defensive efficiency by 0.49. Or, teams in the last two years have given up 0.49 points per 100 possessions less for every dead-ball turnover.
While the data for live-ball turnovers showed a bit of positive correlation, the 10-year graph still shows none. Our R² value is essentially zero, which shows that the data is pretty much random, and dead-ball turnovers have had very little effect on which teams have had good or bad defenses in the past 10 years.
Perhaps it’s not so weird that dead-ball turnovers have recently increased defensive efficiency. Out of the possible possession outcomes – a made shot, a missed shot, a live-ball turnover, and a dead-ball turnover – the latter is the only one that does not allow a fast break opportunity.
While the correlations between live and dead-ball turnovers and offensive and defensive efficiency are arguable, maybe the biggest takeaway is the difference between the two types. The difference in the last two years isn’t incredibly large – roughly a net of 0.3 points per 100 possessions on the defensive end. However, there's still a discernible difference between the two, and this shows that, over the course of a season, a team like the Pistons last year – 56% of their turnovers were live-ball turnovers – can really be hurt on the defensive end because of the disparity.
Up next, rebounds!