A Quick Examination of Blocked Shots

facebooktwitterreddit

Mar 21, 2015; Oakland, CA, USA; Golden State Warriors forward Draymond Green (23) controls a rebound with center Andrew Bogut (12) against Utah Jazz center Rudy Gobert (27) during the third quarter at Oracle Arena. The Golden State Warriors defeated the Utah Jazz 106-91. Mandatory Credit: Kelley L Cox-USA TODAY Sports

Blocks are indeed a troublesome stat. When that number was all we had, it had to serve as a proxy for rim protection. However, with more detailed data, primarily from SportVU, a much better estimation of this ability is available. Of course, this is still merely an estimate[1. And a broad one at that.] based on publicly available data. We know there is more going on – for example, nearly a third of all blocked shots are credited to a player who isn’t the closest defender to the shooter. It stands to reason that even for shots that aren’t blocked, the presence of a second or third defender in close proximity probably affects the accuracy of the shot. We don’t know the degree to which different players receive more or less help from these “secondary” contests, nor do we know which players might be more effective when they are present as an additional defender. It’s likely to the point of near certainty that Anthony Davis is better at altering shots in this way than is his teammate Ryan Anderson, for example, but how by how much and what about less obvious cases? Still, the available SportVU data is better than nothing at all.

Most importantly, the more detailed info reduces the problem of the binary nature using blocks to denote rim protection. While a blocked shot has a zero percent chance of going in[2. In a statistical sense at least. Occasionally shots are “blocked’ in colloquial understanding – the defender deflects that ball before it gets to the rim – but it goes in any way. These aren’t recorded as blocks in the box score however.] the presence of a defender generally has some effect on even those shots he doesn’t actually touch. These altered shots have a lower chance of going in than they would otherwise. Looking at the numbers in aggregate, this is inarguable. Shot attempts from 5 feet or less are the more susceptible to defensive pressure than shots from anywhere else on the floor, largely because unguarded shots from in close are so easy. The chart below reflects the shooting percentage on such close shots by each additional foot of space between shooter and closest defender.

So, even if a shot isn’t blocked, just being there and being in the way can provide some pretty decent defensive value.[4. Not as much value as not allowing a shot near the rim in the first place, but sometimes stuff happens, even to the best defenses.] Over enough shots, a player’s general rim protection skill should shine through to a degree in the percentage of makes on the shots they defende at the rim. At least that’s the theory behind the Rim Protection metric itself. Tracking makes and misses rather than blocks and “not blocks” allows for these finer determinations of ‘changed’ as well as blocked shots. In fact, once FG% allowed is the measure being used, the “block” stat itself has very little usefulness – a blocked shot has no chance of going in. That’s plenty valuable already, and there is little need to add on an extremely blunt, often subjective counting stat to the mix on top of that “guaranteed miss”. Comparing more detailed rim protection with raw block rate helps illustrate just how imperfect the box score can be on this front:

A few notes on the chart: The fit isn’t actually that bad, about 60% of the variation in rim protection value is explainable through shot blocks. This shouldn’t be that surprising as adding an extra “0%” shot or two into the attempts a player defends can’t do anything but help overall.  Looking at some individual players, Davis is an interesting case simply because he blocks more jumpers away from the basket than just about any other big. While these are neat, they also aren’t quite as valuable, considering the expected return on a midrange pull up is about .8 points per shot as compared to about 1.5 points per shot on a layup. A hybrid big/wing like James Johnson suffers a similar fate. It’s probably no surprise to see players like Biyombo, Whiteside and, protestations of Doc Rivers to the contrary, Jordan as being mildly to moderately overrated by raw blocks, while savvy defenders like Bogut, Hibbert and Duncan appear mildly underrated. Editorializing further, surprising names with low block numbers, but high rim protection indicators like Zeller and Speights are probably decent candidates either for a sharp regression to the mean next season due to small sample sizes or are benefiting from other contextual factors making them appear more effective individually than might actually be the case.

Interestingly, one of the most common critiques of the blocked shot is one of the least trenchant. It is certainly true blocked shots result in offensive rebounds at a higher rate than other missed shots. According to numbers obtained by Nylon Calculus data whiz Darryl Blackport[2. I’ve said it before, but without Darryl, the content at TNC would be substantially worse. He’s the best.] through combined SportVU and play-by-play shot logs, 31.1% of blocked shots have been rebounded by the shooting team, as compared to 25.4% of all rebounds of missed field goal attempts. However, this is significant;y biased by where on the floor blocked shots occur. Over 55% of blocked shots are taken within 5 feet of the rim, as compared to around 14% of all missed shots. Unsurprisingly, the closer to the basket a shot is attempted, the higher likelihood of an offensive rebound[3. The dip in offensive rebounding for midrange shots is curious, but if I had to speculate, I’d guess those attempts are weighted more heavily towards big men than are three pointers and bigs shooting jumpers is

no bueno for offensive rebounding

]:

So, since blocked shots tend to be closer to the rim than the average missed shot, it stands to reason they will also be offensive rebounded at a higher rate. Controlling for this distance issue by looking at only shots blocked taken from five feet or fewer this season, blocked shots are actually slightly less likely to be rebounded by the shooting team, 33.4% as compared with 36.3%. While this doesn’t take into account shots which become team rebounds by virtue of going out of bounds[3. Certainly an issue for “statement” blocks on which the defender volleyball spikes the insolent layup attempt into the 12th row.], it does not appear there is a marked diminution of rebounding because of the blocked shot. In other words, while shot blocking is probably overrated in the grand scheme of things,[4. as compared to general rim protection.] it’s not because of rebounding.