Three Point Plinko and the Minnesota Timberwolves

We still don’t know an awful lot about the process of shot and offense creation in the NBA. We’ve gotten pretty good at describing the inputs in terms of what various players bring to the table, and the outputs in terms of team level offensive success or lack thereof. But in between, how a given set of players interact with an offensive system to arrive at the shots actually taken, remains something of a black box, at least from an analytic view.

From a qualitative standpoint, coaches know a great deal about what kinds of actions will tend to produce the best shot, but even in that frame of reference it resembles a game of plinko. The structure of a team’s offense and personnel determine the placement of the pegs, and the play call is choosing which slot to drop in the pellet. But from there, the series of bounces, the intricacies of the offensive and defensive execution of any given play, determine the outcome in a not entirely predictable way.

This doesn’t mean we should just throw up our hands and say things like, well:

From one standpoint, what Mitchell is saying is inarguable. For the most part, three pointers DO have to come within the flow. It’s why three pointers are far more effective coming in catch & shoot situations – 3FG% off no dribbles was 36.3% in 2014/15 vs. 31.1% off the dribble[1. Steph Curry’s ridiculous ability to shoot off the dribble from distance was the single biggest reason for his MVP season in 2014/15 and was a driving force behind the Warriors’ overall offensive success.], and why the overwhelming majority[4. 84.2% across the 2014/15 regular season and playoffs per play by play logs.] of made threes were assisted. Open threes are usually the end result of good things happening throughout the offensive process. Which leads me to the problem with Mitchell’s statement: Minnesota’s process, their “flow”, stunk.

Getting back to the plinko analogy, one season of low three point rates might be an unfortunate happenstance of roster construction or random variation. But Flip Saunders’ offenses (as well as those of his disciple, Washington’s Randy Wittman) have always produced three pointers at well below league average rates and long twos at higher than average rates[5. The one exception being Saunders’ first season in Detroit, when unleashing Rasheed Wallace to bomb away helped the Pistons to dramatically improve their offense from the last season of the Larry Brown regime.]. That “flow” Mitchell was speaking of, from which threes must naturally arise is not an accident, an exogenous force with which the Wolves must deal. That “flow” is the result of design. And that design consistently produces the worst shot in basketball, the long two.

By way of illustration, consider the following breakdown of each team’s catch & shoot proclivities in 2014/15. I choose catch & shoot attempts, because these will strongly tend to be the result of good design and execution. If you catch the ball and you are open, you shoot it. Something good has probably happened in a possession to lead to that point. Further, to rule out, as much as possible, confusion with situations where a big man shoots a jumper out of a pick-and-roll, I’ve further limited the sample to players playing at a wing position at the time of the shot, per the positional playing time data I worked through last week.

As the chart below shows, Minnesota’s wings took, by a decent margin, the most catch & shoot two pointers of any team in the league. Largely as a result, only the jump shot-challenged Hornets were less efficient on plays where they got open jumpers for their wing players:

And to the common refrain, that this reflects the talents of the players, well not so much. Andrew Wiggins isn’t yet the finished article as a three point shooter. But he did hit respectable 34.5% on catch & shoot threes. As compared to an execrable 31.3% on catch & shoot 2 pointers:

Again, if the “flow” is generating those kind of shots, this isn’t the flow you are looking for.