As Ian wrote the other day, “the value of a statistic…is tied to the questions they are being used to answer.” Unfortunately, until now the data needed to answer some fairly foundational questions about basketball haven’t been readily available. Thanksfully, the introduction of detailed and granular player tracking information such as the SportVU data on NBA.com has started to provide some this raw analytic materials. Though the public is just skimming the surface of what can be learned by manipulating this new data as well as combining it with existing stats, we can start to examine bits of conventional basketball wisdom in ways not readily doable previously.
One such bit of common parlance is the notion of the “pass first” point guard. At the most reductive level, lead guards are defined as being “scorers” or “passers,” with the former group sometimes being derided as “not pure point guards.” This line of thinking perhaps reached it’s zenith this offseason when the Sacramento Kings replaced Isaiah Thomas with Darren Collison.
Though a rather clear downgrade in terms of talent, the move was sold as giving the Kings more of a “true point guard.” However, this bit of organizational spin doesn’t hold up under even passing scrutiny. Collison was actually slightly more inclined to shoot than to create for teammates than was Thomas during 2013-14:
That said, relative to shooting, Collison did pass more often than did Thomas last year, averaging 4.33 passes per shooting attempt against Thomas rate of 3.67. On that level, Collison was more of a “passer”.
But the discussion doesn’t end there. Passing and “setting up teammates” are related but not totally the same thing. If Thomas shoots more often relative to passing, but throws what might be described as meaningful passes at a similar rate as Collison, to some degree Thomas’ passing was more effective in terms of creating offense than was Collison’s.
More broadly, as Matt Femrite discovered, creating for teammates can be risky, trying to set up a shot can increase the risk of turnovers:
As there is a difference between simply passing the ball and passing “with a purpose,” it’s worth investigating if there is a relationship between the ratio of passes and shots and the efficacy of that passing in terms of creating assist chances. For the entire league, there was almost no correlation between the percentage of a players’ passes which led to teammates’ scoring attempts and their own pass-to-shot ration:
In fact, looking strictly at point guards, what little correlation could be found was actually negative. That is, the more shot-happy a point guard was, the (very slightly) greater likelihood of their passes leading to teammates’ shots1:
This aligns with an intuitive understanding that simply swinging the ball around and never looking to score isn’t necessarily the best way to get shots for teamamtes. A key part of being a “drive-and-kick” player is the drive. And in fact, driving the ball was very highly correlated with generating assist chances. For the whole league, almost 60% of the variation in the likelihood of a pass leading to a teammates’ scoring opportunity can be explained by a player’s ability to drive the ball towards the basket:
Looking just at point guards the correlation isn’t quite as strong.2, but still observable:
Again, by both intuition and observation, driving the ball is riskier than simply passing the ball around the arc. Comparing driving ability with turnover rates confirms this suspicion:’
With the correlation even stronger just among point guards:
All of this taken together implies a sort of “usage/efficiency” curve for setting up teammates. Which the data happens to support:
But bringing us back to the example of Thomas and Collison, Thomas as much more effective in terms of creating offense not only for himself, but for his teammates as well when examined through the lens of this relationship between risk and creativity:
This is still a very top-level look at the relationship between aggressiveness, unselfishness and creating team offense. However, with the new data available, claims such as this can be examined with a little more rigor. With more seasons and data, a deeper understanding over the very notion of a “pass first” point guard being a real or even desirable thing can be explored.