Nylon Calculus: Shooting and shot-creation, what’s more valuable?

Oct 27, 2016; Chicago, IL, USA; Chicago Bulls guard Michael Carter-Williams (7) defended by Boston Celtics guard Terry Rozier (12) during the second half at the United Center. Chicago won 105-99. Mandatory Credit: Dennis Wierzbicki-USA TODAY Sports
Oct 27, 2016; Chicago, IL, USA; Chicago Bulls guard Michael Carter-Williams (7) defended by Boston Celtics guard Terry Rozier (12) during the second half at the United Center. Chicago won 105-99. Mandatory Credit: Dennis Wierzbicki-USA TODAY Sports /
facebooktwitterreddit

Right before the start of the regular season, the Milwaukee Bucks traded former Rookie of the Year Michael Carter-Williams to the Chicago Bulls straight up for his fellow 2013 draftee, and decidedly not the Rookie of the Year, Tony Snell. The move on the Bucks part was a pretty clear fit move likely significantly motivated by the torn hamstring injury to their arguable best player, Khris Middleton, who plays the same position as Snell.

Snell is a decent defender and a decent 3-point shooter, with a 35 percent career mark coming into the season, but doesn’t contribute in many other ways. Carter-Williams, on the other hand, is a non-shooting point guard with good defensive potential, though an uneven actual record on that end to date. In terms of overall value my projections give a slight nod to Snell for this year, but that’s again not likely the prime motivation for the Bucks, rather it’s classic fit issues, Snell plays what is now a position of need for the Bucks and his best skills, shooting and perimeter defense, are two things that complement their current players more than Carter-Williams.

In this case, for the Bucks it was shooting over play making. We all know that shooting and spacing the floor are becoming more and more valued in the NBA. But the stars of the league are still largely shot creators. So that leads me to the question, which skill is relatively more valuable?

Read More: Drivers, shooters, and the players who can do both

From a stats point of view, one way to look at this is to measure the variance between players for each skill. If there is little variation between players for a skill, there will be little difference between even the best and worst players on that basis. The wider the range between the best and worst players at a useful basketball skill, potentially, the bigger advantage to emphasizing that skill.

Defining play-making and shot creation is a little tricky. There is both creating shots for yourself and for your teammates, generally measured in assists. To explore the issue here, on the scoring side, I chose two of the most archetypal half-court shot creation play types using the NBA’s Synergry play types: Isolation and pick-and-roll ball-handler possessions to contrast with spot-up plays using last year’s data.

For the play types, I looked at the variance in scoring volume and efficiency in each. Spot-up plays are the more efficient of the three, which is important to keep in mind, adding more average spot-up plays, particularly 3-point spot-ups, has value. But the variance in efficiency is basically the same for players with at least 50 possessions finished via any of the play types when we control for position. For this, I am showing the coefficient of variation (COV), a measure calculated as the standard deviation divided by the average to make it easier to compare the variation in measures of different scales.

image-5
image-5 /

In addition, none of the play types show a strong correlation between efficiency and frequency use of the play type.

There is, however, much more variance between players on the frequency of use for the shot creation plays than there is for spot-up plays. In this measure, I used points scored per forty minutes for each play type, with players with a minimum of 500 minutes played last year. For every position other than centers there is much more variation in the creation play scoring volume than there is for spot-up plays, which can be seen in the table below.

image-6
image-6 /

The COV’s for points per 40 minutes overall were 86% for Iso’s, 88% for Handler points and 48% for spot-up points. Clearly there is a wider variation in point volume in the creation type plays than in spot-up shooting.

What that means precisely in terms of skill variation is a subset of the larger on-going debate over the trade off between usage, or volume scoring, and efficiency. There is only a weak positive relationship between volume of attempts per forty minutes and efficiency for any of the play types. But the real question is a hypothetical, for example, what would the efficiency of a lower used pick-and-roll wing or backcourt ball-handler be if used more in that role?

Even past cases of ramped up creation by players are of limited use to apply across the board. Kawhi Leonard’s explosion of use as a shot creator this year, for example, is only occurring because Leonard has continued to show growth and earned Greg Popovich’s trust. It would be folly to assume that his volume and efficiency performance could be applied to other currently lower creation wings.

But one clue in the current data could be in passing point creation, where the efficiency gap is larger than the gap for the scoring play types I looked at.

The COV for points created via assists per pass, a measure I use for my Player Tracking Plus Minus (PT-PM) metric, was 42 percent. That was larger than any of the Synergy play types efficiency variation that I looked at. The COV for assist volume, points created via assist per forty minutes, meanwhile, was 63 percent right in between the scoring creation and spot up variation.

Below is a chart with the passing efficiency separated by position.

image-8
image-8 /

The higher COV on passing efficiency within positions indicates that passing efficiency is a measure with some real separation in talent. Further, the outliers on the upper end tend to line up with the players we think of as stars. With Draymond Green, LeBron James, James Harden and Russell Westbrook all leading the league in their respective positions by this metric. The somewhat more traditional points created by assists per turnover also shows more separation and variance by position than any of the scoring play types do in efficiency as well.

In fact, Justin Willard’s research found that there is an interactive quality between scoring and passing proficiency; being a superior passer increases the impact of a player’s scoring and vice versa. The idea being that those high volume scorers and passers are typically the players putting pressure on the defense, drawing double teams and disrupting their opponent’s defensive scheme.

Related Story: The Boston Celtics' go-to SLOB play

What the variability measures do not deal with is the line-up effects of adding more shooters vs more creators. Other studies, by myself, Willard, and Steven Shea, for example, have all indicated that multiple 3-point shooters on the court has a positive effect on offensive efficiency. There have been fewer studies on the ideal number of creators sharing the floor, though the general thought seems to be that there is a quicker declining advantage with shot creators. If that’s true it would have separate implications for the value of average shooters vs average creators, even if better creators add more to the team.