Introduction by J.R. Wilco
The way sports are analyzed is based on certain assumptions, and I think most people (myself included) are completely blind to most of those assumptions. They are the way we understand the game we watch, and the context in which we view each performance. When we say that someone played well, we've made an evaluation that's not just based on previous experience, but also guided and shaped by hundreds of principles that we never consciously consider. This dynamic also extends to the analysis of statistics and the numbers that we use to judge player performance.
Recently, I reached out to cocanat about the way he works with statistics to evaluate players, and I asked him to write about that process, which we named PRABS. In the discussion of that story, it became clear that he is someone who has a unique way of viewing things, such that some of those invisible assumptions are highlighted and questioned. To me, that is a recipe for interesting an thought-provoking writing. So once more, I asked him to share his take, this time on turnovers as a statistic. I hope you enjoy his refreshing perspective.
Here's the standings of a stat of the 2011-12 season:
1. Deron Williams
2 John Wall
3. Kevin Durant
4. Steve Nash
5. Rajon Rondo
6. Russell Westbrook
7. Kobe Bryant
8. LeBron James
9. Dwight Howard
10. Ricky Rubio
Who knew that a list made up of some of the game's greatest names could actually be a list of the league's worst offenders in what's perhaps the NBA's highest-profile negative statistical category: most Turnovers per Game?
Wait a minute! If turnovers are such an awful thing, shouldn't the leaders be from teams like Charlotte, Detroit, Cleveland, New Orleans, Sacramento, or Golden State? You know, those teams that aren't very good. The failing teams that can't seem to get it together. Not the Thunder and Celtics. Certainly not the Heat and the Lakers.
How can six players in the Top 10 be the stars of playoff teams? I think the conventional wisdom of the league is maybe missing something.
Now the first thing that comes to mind is that statisticians like John Hollinger have tried to put the turnover into perspective, by turning the number into a percentage of possessions. But, to me, that doesn't quite go far enough.
The Turnover is supposed to have some meaning as a ball-handling, possession, and play-creation stat. But the NBA itself created a problem for that, when sometime in the mid-to-late 70s, it went to a scorekeeping system that had to account for a result of every possession. In the modern era, each possession must end in either a made shot, a missed shot with rebound, a blocked shot, a foul, or a turnover. (All you detail-oriented people who immediately thought about the end of the quarter can give yourself extra credit).
So the Turnover became an official stat, and to give Official Scorekeepers something else to do, a turnover had to be assigned to a player.
But is The Turnover a the kind of stat we think it is? I wonder.
1. A player whistled for charging is charged with a turnover. Is that a ballhandling stat?
2. A player whistled for a moving screen is charged with a turnover. Is that a play-creation stat?
3. A player whistled for a 3-second call is charged with a turnover. He might not have even had possession.
4. A player holding the ball when the 24-second clock runs out is charged with a turnover. Just what kind stat are we dealing with here?
But these glitches are minor compared to my major issue with the recording and publishing of individual turnover stats which is: the actual ballhandling turnovers are recorded against the last player who clearly possessed the ball. Why is this so horrible? Please consider a few situations here. I'll give the players names, just for fun.
1. Tim Duncan passes to a wide-open Tiago Splitter who misjudges the angle, and the ball bounces off Tiago's chest and out of bounds. Tim gets a turnover even if he made a perfect pass and the loss of possession is entirely Tiago's fault.
2. Tony throws an excellent lob pass (relax, it's a hypothetical situation) to Danny Green who totally botches his jump, and loses the ball. It's such a humorous and obvious mistime of a leap that Tony jokes to Danny that he's taking it out of the playbook, But it's Tony that's charged with the turnover.
3. Manu Ginobili fakes a jumper and throws a perfect pass to DeJuan under the basket. Unfortunately, DeJuan has been faked out, too, and doesn't see the ball as it bounces off his shoulder and out of bounds. Yep, Manu gets the turnover.
I could go on and on with these examples, but obviously my point is that the turnover is often, not just occasionally, recorded to the player not at fault. Even the non-handling turnovers are often recorded to the non-offender. Examples include:
1. As Sean Elliot reminds us on nearly every telecast, when the player setting the pick is whistled for a moving screen, it is usually the fault of the dribbler, who didn't set up the pick properly.
2. A 3-second call is often the result of an outside shooter, who, due to the his familiarity with the play being run, should have known he had players at risk for a 3-second call when he faked his shot to make an extra pass or move to a closer spot to shoot.
3. The turnover on a 24-second call is awarded to whichever poor schmuck who received a pass at the 23-second mark, after his teammates passed up shooting opportunities earlier in the possession.
I have many more examples of the inequities in awarding this particular stat, but I think I've made my point.
Before I'd done, I want to make it clear that I consider TOTAL TEAM TURNOVERS to be a completely relevant stat. But, awarding individual turnovers is fraught with problems, and should no longer be relied on as a measure. In fact, I would like to see the end of the individual count. As you can see by searching the list of the players with the worst number of turnovers, the players who are charged are simply those who are the mostly guarded by the defense, or the ones who handle the ball the most.