Why the NBA Internet is fighting about ‘screen assists’

Tuesday night was a standout performance for Utah Jazz center Rudy Gobert. He scored 22 points, grabbed 18 rebounds, and locked down the paint as he often does. Utah picked up a 118-107 victory over the Nets, its 15th in the last 16 games.

The next morning, Jazz radio broadcaster David Locke tweeted this.

The reaction was, ahem, divisive. Jazz fans seemed to like it. Non-Jazz fans ... well, check all of these quote tweets for a brief sample. The overarching theme is that “screen assists” aren’t a real stat, and Locke’s attempts to make it so are mind-boggling, misleading, pathetic, and/or some combination of all three.

As with all social media pile-ons, each party is bringing their own baggage to the discussion. Those affiliated with the Jazz believe they aren’t getting enough respect for their success, and Gobert is their poster child. His defensive intimidation and unselfish offensive disposition is vital to the Jazz’s success. Yet when it comes time to recognize his individual contributions, he’s often passed over for peers whose skills are easier to spot.

This bothers Gobert most of all. When coaches left him off the All-Star team last year, he broke down during a media session and delivered a sour gripe for the ages.

Gobert’s Won’t someone please think of the children! pleas didn’t exactly endear him to the NBA’s Joe Q. Public outside of Utah. If winning was all he cared about, the argument goes, why was he so upset about missing out on an individual honor? In theory, Gobert had a point. In practice, speaking up about it further reinforced the Jazz’s reputation as a team and fanbase that constantly complains about not being respected by the national media and audience. Nobody likes complainers, so here we are.

Fast forward to this year. Gobert is doing his thing again, the Jazz are surging again, and he’s not even in the top 10 in the latest round of All-Star fan voting. His All-Star fate will be up to the same coaches that ignored him last year, and he may even miss out in favor of his own teammate. There was going to be another firestorm about Gobert’s value sooner or later. Locke’s screen assist tweet was simply the gasoline.

Now, toss that context aside. Let’s talk about the overarching idea Locke presented. Let’s talk about “screen assists.”

Is this actually a real thing?

It is! NBA.com tracks screen assists on its advanced stats page and offers this definition:

The number of times an offensive player or team sets a screen for a teammate that directly leads to a made field goal by that teammate

You can find it on the site’s “Hustle” section, alongside metrics such as “most deflections,” “charges drawn,” “loose balls recovered,” and many others. As with all stats on the site, you can view them as a flat number (total screen assists), a per-game ratio, or a per-minute ratio (one minute, 36 minutes, 40 minutes, or 48 minutes). You can also see how many points on average a players’ screen assists yield.

Gobert leads the league in all four categories by a wide margin, with Indiana’s Domantas Sabonis second in totals and per game. (Sabonis actually ranks third on a per-minute basis, behind Gobert and ... Tony Bradley, Gobert’s backup in Utah. We’ll circle back to this).

Can you show me a screen assist in the wild?

Sure. NBA.com doesn’t provide a video database of every screen assist like it does for other more traditional stats, but I feel safe declaring that this was logged as one.

So Gobert is getting credit for standing still while Bojan Bogdanovic did all the work? Why is this worth tracking?

Because there’s much more to setting a good screen than that, especially in today’s pick-and-roll heavy NBA. Doug Eberhardt put it best in this 2014 SB Nation article:

In reality, the nature of setting screens has changed along with the style of NBA offenses. The full-body, bone-crushing pick that frees up a jump shooter isn’t always as important as forcing the screener’s defender to make a decision that will make the other players on the defensive string also make a decision. Who helps? Who helps the helper? Do I come off that corner shooter to deal with the ball handler or roll man? It always comes back to making the defense make a decision.

Having said that, what makes a good basic screen? Most notably, the ability to establish a strong athletic position with your legs a little bit wider than shoulder-width apart. Ideally, you should be sitting down in an athletic posture with your knees bent and body stationary, allowing your teammate to set his man up and cut tightly off of your top shoulder. Your hands can either be covering your nether regions or in on your chest. The old-school, full-body screen.

When most fans think of good screens, they imagine the screener stopping the defender dead in his tracks as the ball handler comes off of the screen with speed into a beautiful drive or mid-range pull-up jumper.

But that’s the thing: you don’t always want to stop the on-ball defender. Sometimes, you want to force the defender to actually get over the screen and trigger the rest of the defense to react to your action. Rather than a lower-value medium-range jumper, you want to end up with a strong roll to the rim, a free throw, a pop for a jumper or a drive-and-kick pass for an uncontested three.

One only needs to watch the cat-and-mouse game Gobert and Joe Ingles play with defenders to understand that not all picks are the same. Ingles is not the quickest, but he uses Gobert’s skill at shifting the angle of his screen to free himself for open jumpers and drives to the basket. Those sequences can look basic, like this.

But they can also look like the defender is playing Twister, like this.

It’s no accident that Ingles’ production has exploded since moving back into Utah’s starting lineup. He’s simply not the same player running pick-and-roll with another big man. The instinct to find some way to statistically value Gobert’s screening flexibility is smart, and I applaud the effort.

So it’s a good stat, then?

Well, not really.

One obvious problem is that the player still has to make a play off the screen for the screener to get credit. This creates plenty of situations where the screen itself may be ineffective, but the ball-handler scores anyway. Should Gobert really get credit for the difficult contested three Donovan Mitchell nailed in Kyrie Irving’s face? One could argue Gobert’s efforts actually hindered Mitchell rather than aiding him.

And which of the two screeners should get credit for the “screen assists” on this Mitchell three? Didn’t Royce O’Neal’s initial screen do most of the work?

Citing screen assists as a ranking of the league’s best screen setters is also terribly misleading because it suggests the only purpose of a screen is to enable the ball-handler to score himself. The reality is that screens are set for many different purposes, ranging from distractions to creating drive-and-kick opportunities to set up a third (or fourth, or fifth) player to score. As Eberhardt noted, a good screen does its job when it helps the team score, not necessarily the individual player.

A screen like this wouldn’t be tallied as a “screen assist,” even though it triggered O’Neal’s kickout pass for Bogdanovic’s three. If the goal is to determine the league’s best screeners, how can you whitewash Gobert’s key role in the play?

There’s also a third problem: screen assists don’t do much on their own to tell you about screen-setting aptitude and efficiency. Friend of the program Dave DuFour was kidding with this tweet, but he’s actually onto something.

Recall that Tony Bradley, Utah’s backup center, was second in the league in most screen assists per 36 minutes behind Gobert. The reason why is simple: Utah’s offense is built around its centers setting a gajillion ball screens to get its drivers downhill. That means they both have ample opportunity to tally screen assists that other bigs simply don’t get, which is reflected in their high screen assist totals. The stat that is supposed to reflect the league’s best screeners instead ends up reflecting the ones who get the most chances.

In a sense, then, both Jazz centers are “screen assists stat-padding,” as weird as it may sound when describing an action as unselfish as a screen.

But can’t you say the same thing about actual assists?

Absolutely! Now we’re onto something!

This is the point Locke is hinting at when he suggested Gobert had a “triple double” with screen assists against the Nets. Screen assists may be fraught with peril, but so are actual assists. Why is one a fundamental stat used for triple doubles and the other relegated to an obscure section of the NBA’s advanced stats page?

I’m more than sympathetic to this argument. Let’s copy and paste each counterargument against screen assists from above, except with some small word changes:

One obvious problem is that the player still has to make a play off the screen for the passer to get credit. This creates plenty of situations where the pass itself may be ineffective, but the receiver scores anyway.

Sounds about right:

Citing screen assists as a ranking of the league’s best passers is also terribly misleading because it suggests the only purpose of a pass is to enable the man receiving it to score himself. The reality is that passes are made for many different purposes, ranging from distractions to creating drive-and-kick opportunities to set up a third (or fourth, or fifth) player to score. As Eberhardt noted, a good pass does its job when it helps the team score, not necessarily the individual player.

Yup.

There’s also a third problem: screen assists don’t do much on their own to tell you about passing aptitude or efficiency. ... Recall that T.J. McConnell, Indiana’s backup point guard, is third in the league in most screen assists per 36 minutes behind LeBron James and Ricky Rubio. The reason why is simple: Indiana’s offense is built around its point guards dribbling a lot and making a gajillion passes to get its other players downhill. That means McConnell has ample opportunity to tally screen assists that other less ball-dominant backup point guards simply don’t get, which is reflected in his high screen assist totals. The stat that is supposed to reflect the league’s best passers instead ends up reflecting the ones who get the most chances. In a sense, then, McConnell is stat-padding, as weird as it may sound when describing an action as unselfish as a pass.

A few logical leaps — I don’t think McConnell is stat padding, for example — but that also sounds about right. So why are “screen assists” mocked as a stat, but actual “assists” aren’t? It’s a fair question.

Great. So shouldn’t we then make screen assists mainstream?

That’s one way to put it, but I’d argue the opposite. Two wrongs don’t make a right. Whataboutism is a scrooge, even on a topic as nitty-gritty as this.

If assists are as flawed as screen assists, the answer isn’t to elevate screen assists to that level. Just as one needs to dig deeper than the assist leaderboard to determine the league’s best passers, so too do we need to think more critically about the league’s best screen-setters.

And this can happen without taking away from the initial purpose of the stat: to help us value screen-setting as a skill. Rajon Rondo has been known to pad his assist totals, but we’d all agree that Rajon Rondo was still a great passer. The same can be true of Gobert and his screen-setting, and that’s fine.

Besides, there are plenty of other ways to illustrate Gobert’s value as a player. Let’s use those instead. For example: