Deep Dives: The Myth of “Leaving the Right Guy”

NBA teams do not “leave the right guys open.” For every time (and it is demonstrated every time) research demonstrates a team has little influence on opponents’ 3FG%, the first rejoinder is generally along those lines. But across any reasonable sample, it just isn’t so. Certainly over the course of one or even a few games, the bounce of the ball and flow of play might result in the threes given up by a team finding their way to the worst shooters. But across a season, not so much.

My preferred method of demonstrating this, looking to “open” threes allowed[1. Using SportVU definitions, I’ve found that closest defender distance of 6 feet is a good line of demarcation between “open” and “contested” shooting for three pointers, and this differs somewhat from the “contested” and “uncontested” definitions used in NBA.com’s player tracking boxscores.], is sadly not readily available, after detailed SportVU shot logs were taken offline at NBA.com in January. However, there are other reasonable proxies available.

Possibly the best other place to look is catch-and-shoot data. As I’ve argued before, catch-and-shoot shots tend to occur when the defense has broken down in some way. After all, if the player who catches the ball is well-covered, he won’t likely shoot! This season, there is decent variation in opponents’ catch-and-shoot 3FG% across the NBA, ranging from Indiana’s 34% to Phoenix’s 39.8%:

Note that even Indiana’s “League leading” percentage allowed still results in an above average look for the offense.

Moving on, we’ve seen time and again that three point shooting is one of the areas most affected by the variance of “make or miss league.” Players can shoot above or below their “true” talent level for quite a long time, and clusters of makes and misses can be oddly timed, especially in the short run, which from a statistical standpoint, and NBA season largely is with respect to the volume of catch-and-shoot threes. To illustrate, consider how much narrower the range is between the top and bottom teams in terms of expected catch and shoot threes – that is to say, the weighted average[1. For sake of clarity, this means that if a team playing the Clippers gave up 9 catch and shoot 3s to JJ Redick and 1 to DeAndre Jordan’s, Redick’s season-long percentage on those shots would count 9 times as much as Jordan’s in the calculation.],  of catch-and-shoot 3FG% of opponents taking all of these shots:

Not only is the range far narrower, but almost 20% of the variation that exists is explained by the team’s own catch-and-shoot 3FG% – that is to say, San Antonio and Golden State benefit greatly in this measure by not having to defend themselves! This isn’t to say there is nothing teams can do to better defend the three. Being closer to the players who do end up shooting is better than leaving them more open, but better still is running players off the line completely, as there is a strong relationship between the proportion of threes a team forces to be taken off the dribble as opposed to catch-and-shoots and their overall 3FG% allowed.  But for the most part teams simply aren’t able to pick and choose which players get open, in large part because allowing anyone to get open for rhythm threes is a defensive breakdown.