The Numbers Behind NBA Chemistry

Is it possible to quantify chemistry on a basketball court?

For years, the analytical world has tried to tackle that question. This has been the blind spot of the statistical community for decades, and really, the basketball community as a whole. The box score can only go so far.
 
But now we have passing data that helps us understand how players mesh on the basketball court. Thanks to the NBA's recent partnerships with SportVU and Second Spectrum technologies, we can see which players tend to pass to each other and even which players shoot on those passes. For example, did you know that Lonzo Ball shoots just 29 percent on 2-pointers when he receives a pass from LeBron James, but 51 percent when he's fed by other teammates? Well, now you do.
 
So, which stars pass to each other the most? The least? We can take a step further and see how those player-to-player interactions change when the lights get brightest. If you want to see which players a coach trusts, don't look at the starting lineup; look at the finishing lineup. In the same way, we can perhaps take a glimpse into which players trust other teammates when the game is on the line. For that, we'll look at player-to-player assist rates overall and when the game enters clutch time (that is, final five minutes and game within five). 

Let's get to it.
 
Washington Wizards
Back in 2013, Bradley Beal was still a teenager when he started next to John Wall for the first time. Nearly six years later, we're not quite sure if they're a good match. Beal recently told NBC Sports Washington's Chris Miller that the two All-Stars are opposites, but in a peanut-butter-and-jelly kind of way. So, that's nice.

The numbers aren't shy about it: Wall loves passing to Beal. Wall has 51 assists to Beal, the most among any tandem in our study through Monday's games. And it's not just because they play a lot of minutes together. Wall assists Beal once every 12.3 minutes on the floor, one of the most frequent rates in the league. What's more, that rate actually accelerates in crunchtime, as Wall has assisted on four Beal buckets in clutch situations (9.8 minutes per assist). 

In the other direction, Beal has been better at finding Wall this season. Last season, he assisted Wall every 67.4 minutes on the floor together, but that has quickened to once every 41.7 minutes, which is better than a lot of star guard-to-guard interactions. In the clutch, Markieff Morris has been Wall's top feeder, generating three assists to the Wizards' point guard.

Really, Wall and Beal is not the partnership to worry about. Otto Porter Jr. is another story. The Wizards' highest-paid player made his first clutch bucket in Monday's win over the Houston Rockets, having entered that game with only four shots (all misses) in 23 minutes of clutch action. Wall assisted Monday's floater, but Porter needs to look to score much more if the Wizards want to make the playoffs.  (Coach Scott Brooks hasn't minced words when it comes to Porter's aggressiveness).

Bottom line, if you think Wall and Beal don't have on-court chemistry, the numbers disagree.

Philadelphia 76ers
Jimmy Butler just arrived, but he's showing far more chemistry with Ben Simmons than Markelle Fultz ever did. The Fultz-Simmons pairing registered a putrid offensive rating (go watch the BIG Number!), but the Butler-Simmons tandem boasts a spectacular 115.0 offensive rating together. Simmons has already assisted Butler 16 times, sprinkling multiple beautiful dimes to Butler in transition. Sixers fans will be happy to know that two of Butler's five clutch buckets were fed by none other than Simmons.

Embiid's chemistry with Butler? That's ... still a work in progress. Embiid has assisted Butler twice in 159 minutes on the court, which pales in comparison to Embiid's 14 assists to Fultz in 331 minutes, per NBA.com tracking. Butler is shooting two of 10 (20 percent) off of Embiid's passes whereas Fultz shot 33 of 71 (46.5 percent) this season.

When you peak behind the curtain, the most jarring stat of all is that Jimmy Butler is scoring just 15.9 points per 36 minutes with Embiid on the floor. That number soars to 30.1 points per 36 minutes when Butler isn't flanked by Embiid on the court, per NBA.com/stats data. That might be Butler just trying to defer to Embiid (Butler's field goal attempts slice in half when he plays with Embiid in Philly), but the Sixers should look to improve that tandem's chemistry going forward. If they want to make it to the Finals, they need Embiid and Butler to be firing on all cylinders. Get this: Despite all the turmoil in Minnesota, Butler and Karl-Anthony Towns' assists to each other occurred more frequently this season than Butler and Embiid. Keep an eye on that.

Portland Trail Blazers
Like Wall and Beal, the Damian Lillard-C.J. McCollum duo is the rare backcourt with both players scoring above the 20-point mark. But unlike the Wizards, they don't rely on each other to get those numbers. Wall assists Beal twice as often as Lillard-to-McCollum (every 12.3 minutes compared to every 30.6 minutes). The same goes for Beal-to-Wall (every 41.7 minutes) and McCollum-to-Lillard (every 80.4 minutes).

One contributing factor: McCollum has shot 33.3 percent from 3-point land on Lillard's passes compared to 40.8 percent from all other teammates. That same lopsided split showed up last season, too.

But Portland fans shouldn't fret too much just yet. The Portland offense has been far healthier than the Wizards' outfit. Portland's 113.4 offensive rating with Lillard and McCollum on the floor suggests they don't need to reinvent the wheel here. Furthermore, Lillard isn't a natural distributor like Wall, so setting up McCollum all the time isn't a huge surprise. 

The chemistry between McCollum and Lillard isn't super inspiring when it comes to crunchtime. In 28 minutes of clutch situations this season, Lillard and McCollum have combined for 13 field goals between them, but interestingly enough, no assists from each other. Lillard has fed Al-Farouq Aminu and Jusuf Nurkic for buckets twice, but nothing to McCollum. 

This isn't a new thing either. Last season, Lillard assisted just six of McCollum's 37 crunchtime buckets and McCollum assisted just one of Lillard's 39. The fact that Lillard only scored once off of a McCollum pass looks worse when you consider that McCollum had 14 clutch assists last season. The lesson here is that Lillard and McCollum don't need each other to put up big numbers. But they might need to if they want to take the leap in the playoffs.

Oklahoma City Thunder
Paul George re-signing in OKC was one of the bigger surprises of the summer, and he's quietly playing some of the best ball of his career. George is averaging a career-high 24 points per game, but it's not because Westbrook is doing all the work for him.

In fact, it's quite the contrary. Backup point guard Dennis Schroder has assisted George 24 times compared to Westbrook's total of 17. Westbrook delivers an assist to George once every 19 minutes this season, which may seem like an impressive figure, but then you consider that in the 2015-16 season, Westbrook assisted Durant a whopping 250 times, or once every 8.3 minutes. On the other hand, Durant may be the best scorer ever. So, duh.

But what this really points to is that the Thunder are spreading the ball this season, which is a good thing. In years past, the offense became too Westbrook-dependent and it's probably good to be more egalitarian. Last season, Westbrook assisted 181 of George's buckets and the next highest assister had just 28 (Raymond Felton and Carmelo Anthony were tied). This season, Billy Donovan has given Schroder a lot of run alongside the two stars, which eases the burden on Westbrook.

The funny thing? George is shooting just 38.5 percent off of Schroder passes and 35.9 percent off of Westbrook's. Maybe George should never give up the ball and just take it for himself.

Golden State Warriors
Let's get to the good news first. Of the 34 teammate pairings in this study, one Warrior tandem bested everyone in fewest minutes between assists. Which assister-assistee showed the most chemistry by this metric? Draymond Green to Klay Thompson. Through Monday's games, Green had assisted 36 Thompson buckets in 356 minutes, or once every 9.9 minutes. That's absurd. A big part in that: Thompson is shooting 50.6 percent off of Green's passes and 44.8 percent on 3-pointers, easily his best marks for any of his high-volume passers. 

Amazingly, Thompson has assisted Durant just five times in 584 minutes, representing the most minutes in between assists of any pairing in this study. In fact, Thompson has dropped more dimes to Jonas Jerebko (6) than Durant (5). None of Durant's 42 field goals in the last three games came from Thompson.

This season's lack of Thompson-to-Durant cohesion is especially bizarre considering they've had to play large minutes without Curry or Green this season. It gets even weirder when you discover that Thompson assisted Durant 30 times in 1,690 minutes last season, or twice the rate he's doing this season. 

Perhaps that's just randomness. But then again, Thompson is gunning like never before, taking 70 field goal attempts over his last three games with just two assists. That's Klay being Klay.

OK, now the bad news. Remember that blow-up between Green and Durant? I've written here about how Draymond had a point. But Durant's beef about Draymond hogging the ball? That may be spot-on, too. Green has zero assists in 40 minutes of clutch play. Durant has eight. I mean, Kevon Looney has two. I repeat: Green has zero. 

Last season, we didn't see this oddity. Green led the team with 17 clutch assists in 85 minutes, or one every five minutes. Maybe this changes with a larger sample size. Maybe it's bad luck. But get your popcorn, this is something to watch.

Boston Celtics
The Celtics' offense is a mess this season as they re-introduced Kyrie Irving and Gordon Hayward into the mix. But looking strictly at the All-Star pairing of Irving and Al Horford, something interesting emerges. Last season, Irving assisted Horford 80 times and Horford assisted Irving 80 times -- an equal distribution. But this season, Irving has assisted Horford twice as often.

Most of this change is due to Horford shooting a blistering 61.9 percent on 2s and 37.1 percent on 3s off of Irving's passes this season, far better than the 2017-18 campaign. But one of the reasons why the Celtics outperformed expectations and why Irving's name was in the MVP race last year was the amazing chemistry between these two. 

Last season, Irving shot 51.8 percent from the floor on passes from Horford (a ridiculous 48.2 percent on 3s) but just 43.0 percent on passes from all other players. That magic between Irving and Horford isn't happening this season as Irving is shooting just 43.7 percent on his passes. In other words, it's leveled out. Everyone on the Celtics has to shoot better, but rekindling the Horford-to-Irving interaction can help turn things around.

Toronto Raptors
Rather than running it back with Kyle Lowry and DeMar DeRozan yet again, Raptors general manager Masai Ujiri flipped DeRozan to the San Antonio Spurs for Kawhi Leonard. How has that turned out? Not bad so far.

Leonard and Lowry hit it off right away. Leonard has been assisted by Lowry once every 17.6 minutes this season, which is actually more often than the Lowry-to-DeRozan connection last season (every 19.8 minutes). 

Leonard doesn't find Lowry (every 76.2 minutes) nearly as often as DeRozan did (24.7 minutes), but the Raptors' offense is humming right along. With Leonard and Lowry on the floor, the Raptors score an incredible 118 points per 100 possessions, which is a tad below the Curry-Durant partnership (118.7), which led this group of tandems. Last season, the Lowry-DeRozan lineups churned out 112.5 points per 100 possessions. They went from good to great.

Speaking of the Spurs, LaMarcus Aldridge and DeRozan have been doing just fine. DeRozan and LaMarcus Aldridge have a 109.3 offensive rating on-court, but they only have one assist to each other (it was DeRozan to Aldridge) in 49 minutes of clutch time. Meanwhile, DeRozan has three clutch assists to Rudy Gay, his former teammate in Toronto. With a healthy Leonard, the Spurs were the best clutch team in the NBA, but they're still searching without him.

Memphis Grizzlies
Marc Gasol and Mike Conley have been playing together since John Wall was playing at University of Kentucky. To be honest, the two vets could probably play the two-man game with their eyes closed.

And the numbers bear this out. Conley assists Gasol once every 10.1 minutes on the floor, which is the second-best rate in this whole group of stars (only Green-to-Thompson ranked higher). That's a big improvement compared to last season when he assisted Gasol every 14.7 minutes on the floor.

Gasol returns the favor, assisting on 13 Conley buckets, the most for any of Conley's teammates. With about a decade of making music together, this is chemistry at its finest -- the new Tony Parker and Tim Duncan.

Houston Rockets
This might be the most interesting team of the bunch. Chris Paul and James Harden perennially rank among the league leaders in assists, so it would make sense that they'd find each other a ton in Houston, right?

Well, they almost never assist each other. This season, Paul has assisted Harden just four times in 274 minutes on the floor together. Harden has assisted Paul just three times, or once every 91.3 minutes on the floor, the second-worst connection in this group of All-Stars and scorers. 

The only tandem that assisted each other less often was Thompson to Durant. But Thompson averages 1.7 assists. That's understandable. Harden averages 8.2 assists. Only three of his 131 assists have gone to Paul. That's astounding for two top playmakers.

These two scorers were devastating in one-on-one isolations last season and operated mostly in silos on the offensive end. The hero-ball tendencies resulted in the best offense in the league. Not passing to each other isn't the issue. The issue is that they haven't been hitting shots. Paul is shooting 44.1 percent from the floor, his lowest figure since 2006-07. Harden is shooting a mere 4-of-13 (30.8 percent) on catch-and-shoots this season. The result is that Harden and Paul have the worst offensive rating for any duo in this study (104.4 points per 100 possessions).

That's a far cry from their 117.0 offensive rating last season. Last season, Paul assisted Harden 29 times while Harden assisted Paul 19 times. They have a lot of catching up to do, which could be said for the team as a whole.

* * *

Numbers, of course, are not the only way to measure chemistry. Coaches might draw up plays that artificially shape these numbers; players might be on a hot or cold spell on passes from certain teammates; players can also have on-court chemistry that doesn't show up in passing data. All that might be true.
 
But for decades, analysts, coaches and fans have been yearning for hard passing data, and now we have it. We're getting closer to finding out which teammates vibe with each other on the floor, which players put the blinders on and who players like to target in clutch situations. Now we can test what our eyes tell us and what our gut senses when we watch the action. And coaches can alter their game plans.
 
Now we know that Green is twice as likely to find Thompson for a bucket than he is Durant. Now we know that even though Westbrook, Wall and Harden are 20-and-10 threats every night, they arrive at those big assist totals very differently. And while you might think Wall and Beal are the Eastern Conference version of Lillard and McCollum, they feed off each other far more than their Portland peers. After all, the box score can only tell us so much.
 
So next time you see Harden and Paul feed one another for a bucket, gather all your friends, family and followers and scream from atop a mountain. It's basically a once-in-a-lifetime moment.

To see the full breakdown of this study, click here.