Passing Fancy: Examining Assist Location

Jan 8, 2015; New York, NY, USA; New York Knicks forward Quincy Acy (4) defends Houston Rockets forward Josh Smith (5) during the third quarter at Madison Square Garden. Houston Rockets won 120-96. Mandatory Credit: Anthony Gruppuso-USA TODAY Sports

This season, we’ve looked into some gleaned from the granularity provided by SportVU data. Even with only the info released publicly through NBA.com it’s a treasure trove of detailed information about aspects of the game which could only be measured by inference , if at all, prior to the collection and release of this data. After discussing shot locations and times, rim protection and rebounding, today we’re taking a quick look at passing.

The “value” of an assist has long been a controversial topic in basketball analytics, especially since there is perhaps more variation in how assists occur than any other box score entry. A swing pass to a wide open shooter is not the same as a drive and dish for a dunk, nor is it like tossing a live shot clock grenade to a teammate who hits a bailout shot as the buzzer sounds. But all of these count the same[1. Without even discussing the issue of the “generous” home town scorer’s assist.]. We know playmaking, broadly speaking has value — assisted shots are considerably more accurate than unassisted.

Largely, this is because assisted shots tend to be more open than unassisted shots. Looking just at made shots[1. At this point in time, the public data allows a determination of how many, but not which missed shots were potentially assisted by each player. Comparing makes allows for some degree of apples-to-apples comparison.], this holds true across shooting distance and zones:

While the new information provided by SportVU doesn’t totally answer the question of how to assign credit between passer, shooter and possibly even other teammates (surely, a screener does as much if not more work on a play where a pin down screen frees a shooter than the passer), much more is available for study than before.

For starters, where are players getting their assists? Below is a chart representing the assist location for shots assisted by Ty Lawson and James Harden respectively. This will be discussed more in a moment, but Harden, (unsurprisingly given the Moreyball ethos of the Rockets) has wracked up assists on a particularly efficient mix of shots, while Lawson’s chart includes a much higher proportion of midrange attempts:

Resident data expert Darryl Blackport is even working on the ability to chart the location of the pass as well as the shot. Remains to be seen if this is doable over a large sample from public data, but here are the pass and shot locations for Manu Ginobili’s assists in the Spurs Opening Night victory over the Mavs with the pass locations in blue and the corresponding shots in orange:

So that’s cool, but not really usable yet. Back to Lawson and Harden, by using a methodology similar to that which generated Expected Effective Field Goal% (XeFG%), it’s possible to examine the theoretical efficiency (assuming league average shooters) from the mix of a player’s “assisted to” locations. Based on where on the court the shot was taken and the distance to the nearest defender, this is an estimate of that efficiency. Below are details of the top-50 in total assists this season.

What conclusions to draw from this “openness” data is unclear — at this point the detailed data on assists is only available for made shots. Thus, it’s premature to say that, for example, Josh Smith’s passing has been more or less valuable based on his top spot. But the data does show that both in Detroit and now as Harden’s teammate in Houston, he is finding players for potentially valuable shots. Along with Blake Griffin (having an incredibly underrated passing season), the two Rockets are among the three top players in the top-50 in assists with over 40% of their assists going to three pointers[2. Isaiah Thomas is the only player with significant assists who is dishing more than half his assists to threes.]. Smith also rates highly in terms of finding open three point shooters as measured by the “Expected 3%” column.

On the downside a surprising number of point guards collect large numbers of assists to lower-value midrange shots. For players like Chris Paul (passing to Griffin) and Nic Batum (finding LaMarcus Aldridge), this might not be a total surprise, but Derrick Rose, John Wall, Ty Lawson, Mike Conley and Damian Lillard also have high rates of assisting to midrange shots. A question for further investigation is the degree to which these shots are largely the result of pick-and-roll or pick-and-pop play where the point guard draws both defenders.

In the near future, I’ll look into plenty more about passing, including an examination of the high percentages teammates shoot off of passes from multi-skilled big men like Dirk Nowitzki and Al Horford.