MEASURING TURNOVERS

Interesting article from BC web site on turnovers written a few years ago but applicable today...especially today.

Inside the Numbers: The Cost of a Turnover
By Mike Zarren

An area of great concern to Celtic coaches and fans this preseason was turnovers. The Celtics were 28th in the league in turnovers given, ahead of only the Knicks and the Jazz. One newspaper article noted that in a game, the Celtics had given up 21 turnovers, but that the Raptors had "failed to capitalize" on Celtic mistakes, "scoring only 10 points" off of the Celtics turnovers. This "points off turnovers" stat is presented in the boxscore as if those 10 points (or however many) were the only price the team paid for its turnovers. However, this stat is often misleading.

When a team commits a turnover, the scoring crew records the turnover. On the following opponent possession, if the opponent scores, the scoring system credits that opponent with a "point off a turnover." This is the case regardless of the type of turnover - for example, if Ricky Davis steals the ball and dunks it on the other end, the Celtics are credited for two "points off of turnovers" just the same as if the opposition threw the ball out of bounds on the other end, called a timeout, and then forced the Celtics to use the entire shot clock before an Al Jefferson last-second hook shot. Both these plays would be scored as two points off of turnovers, but only in one of them did the scoring really result directly from the turnover.

"Points off of turnovers" is thus a misleading statistic. In the Al Jefferson example above, the 2 points are credited "off of a turnover" even though the team might have run the exact same play, with the exact same Jefferson hook shot, even if the opponent had scored on the previous possession instead of throwing the ball out of bounds.

This is why Celtic coaches do not rely on the "points off turnovers" statistic to determine the cost of turnovers to the team. Rather, they focus on the true costs of turnovers, which are twofold:

First, the biggest cost of a turnover is that the team who turned the ball over loses an offensive scoring opportunity. This, and not the opponent's next possession, is the most important cost of committing a turnover. Teams only get about 94 offensive possessions per game, and the loss of one of these scoring opportunities, much like an out in baseball, is not something you can get back. The Celtics average about 1.21 points per possession when they don't commit a turnover; so each turnover, on average, costs the team one of these opportunities.

Second, the opponent often gets an improved scoring opportunity off a turnover. This varies by type of turnovers, however; an offensive foul after which the defense is able to set up will not improve the opponent's chances of scoring against that defense - they are as likely to score as they would have been had they scored and called a timeout. In contrast, a steal leading to a high-percentage fast break can be a particularly costly turnover. On average, Celtic opponents score 0.08 more points following our turnovers than they do when they get the ball at other times.
So the true "points off of turnovers" for Celtic opponents is about 1.29 points per turnover, though obviously this will vary on any given night depending on the type of turnover and when in our offensive possession it occurred.

Preseason turnover levels generally are not good predictors of how good a team will be at protecting the ball in the regular season. Turnovers are magnified in the pre-season as guys get used to playing with each other. In addition young players, like many of those on the Celtics' bench, are generally more turnover-prone than older players. And, of course, a certain level of turnovers is good; for example if Orien Greene never made any risky passes, he would never make any spectacular assists. It is worth noting that despite the team's higher overall level of turnovers this preseason, the Celtics' starting 5 averaged only 14.3 turnovers per 48 minutes. This is not statistically different from last year's league average of 13.9.

Finally, for all the focus on Celtics turnovers, it is also worth noting that the Celtics also gain points by forcing opponents' turnovers. The team forced 16 opponents' turnovers per 48 minutes this preseason, a rate that would have been good enough for fourth in the league last season.

Mike Zarren is the Celtics' Basketball Operations Analyst, responsible for using analytical tools to help improve team decision making.