APBRmetrics

kjb · Joined: 03 Jan 2005 Posts: 842 Location: Washington, DC

I wondering if anyone has done some work evaluating the value of steals. The Wizards lead the league in steals, and are among the league leaders in forcing turnovers. Yet their defense ranks well below average. The argument from some fans in another forum is that steals are "the best" defensive play there is, and that it's value will show up on the offensive end (THEORY: steals lead to easy baskets). My contention (in part drawn from the defensive data I've been collecting regarding the Wizards) is that the inveterate gambling for steals costs them about as many pts defensively (in open shots and help fouls) as they get in easy baskets from successfully stealing the ball and converting.

Thoughts?

Ben · Joined: 13 Jan 2005 Posts: 264 Location: Iowa City

I think Dan Rosenbaum could speak to this. I believe steals come up as quite valuable in his study. Perhaps you could have something there, but on the average steals still serve as a good proxy for an aspect of defensive ability.

kjb · Joined: 03 Jan 2005 Posts: 842 Location: Washington, DC

Dan Rosenbaum · Posted: Thu Jan 27, 2005 3:19 pm Post subject: Re: Value of a steal

kjb · Joined: 03 Jan 2005 Posts: 842 Location: Washington, DC

Interesting Dan. My thoughts on this are related specifically to watching the Wizards -- a team that leads the league in steals, but has a poor team defense. I've been tracking the defense possession by possession for the past 12 games, and what I'm seeing is rampant gambling for steals that result in open shots for the opponent and fouls on teammates in help defense.

Maybe what I'm seeing is limited to the Wizards, whose coaching staff places a lot of emphasis on stealing the ball. My close observation of the Wizards over the past 12 games tells me something different than your findings.

I'd be interested to see what your methods say about the Wizards this season.

As I re-read your post, I think I'm seeing you say that steals are valuable IF the player does others things well too. Is that correct?

Dan Rosenbaum · Posted: Thu Jan 27, 2005 3:52 pm Post subject:

kjb · Joined: 03 Jan 2005 Posts: 842 Location: Washington, DC

I suspect that steals have tremendous value when they're coupled with the ability to force missed shots. The Wizards lead the league in steals and are 4th in forcing turnovers, BUT are 26th in efg, and are 27th in opponent offensive rebounding %. The data I've collected over the past dozen games suggests that the high efg may be related to the number of open shots stemming from failed steal attempts (as well as a preposterous zone defense they run sometimes). I'll try to track this specifically over the next few games and see what I find.

Kneepad · Joined: 19 Jan 2005 Posts: 16

Kevin Pelton · Posted: Fri Jan 28, 2005 12:50 am Post subject:

Perhaps Roland will come along and offer some more insight, but he sent me the following chart last summer when I asked about the value of steals on offense:

Kevin Pelton · Posted: Fri Jan 28, 2005 12:53 am Post subject:

Mike G · Joined: 14 Jan 2005 Posts: 3293 Location: Delphi, Indiana

If someone (Kevin B?) is tracking Missed Steal Attempts, and their ensuant effect on team defense, we'd have an idea of the net value of a steal attempt. Kevin P suggests a diminishing value of steals registered when a player (Hughes?) might just be pursuing a record or a stat title.

Such situations are similar to a FGA. While this category is one of the officially-recorded ones, it overlaps with FG. However, a shot attempt is either/or: it either goes in, or it doesn't. Good or bad. 2 or 0.

Dan R earlier published positive value for FGA, FTA, and 3FGA. I thought a more revealing regression would be for Missed FGA, etc. Surely one wouldn't arrive at a value >0 for a Missed shot, even when there's a residual benefit from an undifferentiated Attempt.

Just as a missed FGA may lead to an offensive rebound, a missed Steal may be followed by a defensive stop by a teammate. A team might even anticipate the occasion and adjust for it. This is analagous to getting rebound position when you anticipate a shot attempt.

Missing a steal also might take the defender out of position to rebound. A guy can rebound his own shot when his man is elsewhere.

kjb · Joined: 03 Jan 2005 Posts: 842 Location: Washington, DC

I'm going to take a stab at tracking missed steal attempts in upcoming Wizards games. I think that KevinP's comment may encapsulate what I'm seeing from the Wizards. Their defensive strategy is predicated on stealing the ball and forcing turnovers. It may be that the pressure defenses they're running would "naturally" generate X steals, but that the extra gambling it takes to achieve Y steals ends up hurting the team defense. This is only theory at this point, of course.

Different topic: What are the guidelines for official scorekeepers in crediting steals? Is there a manual somewhere that I could obtain? I'm curious about this for a few reasons, the biggest being differences between the steals I'm recording and the steals being credited in the official box. What I think I'll do is jot down a narrative of the possessions where I'd record a steal and compare it to the play-by-play.

Kevin Pelton · Posted: Fri Jan 28, 2005 10:46 am Post subject:

gabefarkas · Joined: 31 Dec 2004 Posts: 1255 Location: Durham, NC

xteve · Joined: 14 Jan 2005 Posts: 6

My contention (in part drawn from the defensive data I've been collecting regarding the Wizards) is that the inveterate gambling for steals costs them about as many pts defensively (in open shots and help fouls) as they get in easy baskets from successfully stealing the ball and converting.

Thoughts?

I'm with you WizardsKev. I've often thought that steals in and of themselves are an overrated stat. Oftentimes guys with high steal numbers are playing the lanes instead of their man. Ron Artest in his Chicago days was a prime example. He had great steal numbers but I can't even count the number of times he'd get burned playing 3 feet off his man trying to cherrypick. It would make for an interesting study should Roland choose to finish it! I would not think the steal had as much offensive value as his numbers seem to suggest.