Stats Don't Lie: Dwight Howard Overrated Defender

Shapecity · Mar 13, 2010

Only about 20 percent of defensive plays show up in basketball box scores, which convey far less information about how points are stopped than baseball box scores do about how runs are prevented.

To get more meaning out of hoops stats, analysts have to either use statistical techniques to extract more from the existing data, or go out and get more data, even if it means creating new companies to track it. The second approach is promising -- Mark Cuban is investing in one -- but the first is far from exhausted. I want to delve into a particularly elegant example: research presented by John Huizinga and Charles "Sandy" Weil at last weekend's MIT Sloan Sports Analytics Conference.

Last year, Huizinga (who is Yao Ming's agent) and Weil presented interesting evidence that, contrary to the "hot hand" theory, NBA players are significantly overconfident after making shots. This time around, they looked at "The Value of a Blocked Shot in the NBA: From Dwight Howard to Tim Duncan." Analyzing about 1.6 million possessions from the 2002-03 to 2008-09 seasons, Huizinga and Weil found meaningful and often huge variation, in how players block shots and in how valuable those blocks are.

Offensive chances that begin in different ways -- with offensive rebounds, turnovers, inbounds passes -- each have different expected points values, as do different kinds of shots. Teams score an average of 1.54 points on layup attempts, for example, and 1.04 points on 2-point jump shots. Players do more for their own team when they stop the kinds of possessions and block the kinds of shots that are typically worth more to opponents. In other words, blocking a layup attempt is more valuable than blocking a jump shot.

What happens after (and because) a player blocks a shot is even more important, according to Huizinga and Weil. If you block a shot, the best result is a defensive rebound; you thus reduce the expected point value of your opponents' possession to zero and give the ball back to your team.

In the seven seasons that Huizinga and Weil studied, Rasho Nesterovic was able to direct 65 percent of his blocks to his own teammates, while Stromile Swift had no blocks at all that yielded defensive rebounds. If your block produces an offensive rebound -- often the result of smashing the ball out of bounds -- that's neither the best nor worst result. The other team keeps the ball, with an expected value of about 1 for its possession. If you goal-tend on a block, that's the worst; your opponent scores automatically, and occasional fouls push the expected value of the possession up to about 2.07.

Over the entire stretch of data that Huizinga and Weil examined, Tim Duncan didn't goal-tend once, while 24 percent of Dwight Howard's blocks resulted in free points for the other team.

Theoretically, there's another way for blocked shots to create value: by setting up fast breaks and increasing your own team's chances of scoring. Huizinga and Weil call this the "Russell," but nobody does it much today. However your team gets the ball back after you block a shot, the expected value of its next possession is pretty much the same.

Huizinga and Weil looked at the top 170 shot-blockers in the NBA, calculating the cumulative difference their blocks made in their respective opponents' expected points and adding the smaller effect of their Russells. (There are points-saved and points-created matrices involved here, cross-indexing how any given chance starts with the type of shot it produces to get values for various possible situations; for those, you will have to look at Huizinga and Weil's actual paper, which I will link to as soon as it's online.) The best season they found: Theo Ratliff in 2004, when he generated a monstrous 300 points off blocks. Marcus Camby in 2008 was No. 2, at 275.

The data yields all kinds of instructive contrasts, most clearly between Duncan and Howard (hence the name of Huizinga and Weil's paper). In 2008, Howard had 232 blocked shots, but he either saved or created just 124 points; Duncan had 149 blocks, but generated 167 points. On a points-per-block basis, Duncan has four of the top 10 seasons, according to Huizinga and Weil. Howard has three of the bottom 10.

This research leads to a number of interesting questions: Do players' shot-blocking skills change over time? Is it possible to predict how a player's blocks will change if you put him on a different team? Watching Huizinga present the data, I also wondered if he and Weil could expand their points-saved methodology to look at all defensive plays, or at least further types of defensive plays. In the meantime, though, it's clear we need to pay attention not only to the number of shots a player or team blocks, but to which plays are blocked and, especially, what kind of results those blocks generate.
Click to expand...

Source: SportsMetricians

44Thrilla · Mar 13, 2010

I did notice that Howard always swats it out of bounds as hard as he can, like a moron. He needs to watch some film of Bill Russell. He seemed to always come up with the ball after he blocked it.

Mr. J · Mar 14, 2010

Does Dwight's goaltending and swatting of the ball out of bounds intimidate opponents enough to reconsider taking it inside, or at least making opponents' adjust their shots and consequently increase their chances of missing? That's something to consider.

zoin · Mar 22, 2010

I agree, I guess other teams doesn't want Howard to play that way anymore because their player might loose their confidence if they were often swat like that. Too embarrassing lol

Log in or Sign up

Stats Don't Lie: Dwight Howard Overrated Defender

Shapecity S2/JBB Teamster Staff Member Administrator

44Thrilla cuatro cuatro

Mr. J Triple Up

zoin New Member

Share This Page

Log in or Sign up

Stats Don't Lie: Dwight Howard Overrated Defender

Shapecity S2/JBB Teamster Staff Member Administrator

44Thrilla cuatro cuatro

Mr. J Triple Up

zoin New Member

Share This Page

Useful Searches