The 2015-16 Golden State Warriors were without a doubt one of the greatest NBA teams of all time. The Warriors won a regular-season-record 73 games, and had their season ended five (in-game) minutes earlier, would be back-to-back champions. Had their fate taken a tiny turn — or had two-time MVP Stephen Curry not been affected by lingering injury — they might still be the darlings of sports nerdom, ready to take aim at the 1995-98 Rodman-era Chicago Bulls for the most dominant three-peat run in NBA history. Instead, after losing three straight championship-clinching games, they stand alongside the 2007 New England Patriots as one of the most dominant teams in sports history not to win a title. As it was for those Patriots, their record-setting achievement became their shame.
And then things got really crazy.
Signing 2014 MVP Kevin Durant away from the Oklahoma City Thunder during the offseason — as covered here and virtually everywhere — seems, at a minimum, like a rich-get-richer acquisition. Although fans of the league’s 29 other teams and/or fans of parity are likely (and perhaps rightly) fuming, championships aren’t decided by a want-off — collecting multiple superstars is pretty much how you win in the NBA. By some measures, this isn’t even the most dramatic team-up. When LeBron James joined the Miami Heat in 2010, Box Plus/Minus pegged him and new teammate Dwyane Wade as the top two players in basketball. Last season, Curry and Durant were, by comparison, a relatively pedestrian first and fifth, respectively. And, as noted by my colleague Neil Paine, “super teams” often underperform expectations. (We’ve certainly seen some super-duds.)
Of concern for the Warriors is that when combining forces in the NBA, a paramount consideration is typically NOT duplicating skills. A team can only take one shot or get one rebound at a time, so duplication leads to diminishing returns. And Curry’s shooting was so good last season that he should arguably be taking more shots rather than sharing — even with an MVP-caliber shooter. And it’s not like the Warriors didn’t have to make sacrifices — departed big men Andrew Bogut, Marreese Speights and Festus Ezeli combined for more than 3,000 minutes last season.
The Warriors’ first outing this season obviously did not go well; the San Antonio Spurs kicked the ever-loving heck out of them on opening night. Golden State followed that up with closer-than-they-should-have-been wins over New Orleans and Phoenix. So the possibility of having fixed what wasn’t broken has taken on a more material threat than we even might have guessed.
On the other hand, the Warriors aren’t trying to get more balanced or plug holes here and there: They’re trying to break the game.1
The Warriors are betting on the belief that certain skills — if stacked beyond a certain, typically unattainable threshold — can disrupt the balance of a game so completely that instead of producing diminishing returns, they produce increasing ones. We’ve seen this dynamic at work in the past — for example, when the Spurs stumbled into the No. 1 overall draft pick and took Tim Duncan, pairing him with perennial All-Star center David Robinson (forming the “Twin Towers”) and kicking off a dynasty. Or in the NFL, when the Minnesota Vikings added Randy Moss to an already loaded wide-receiving corps and tore up the league’s offensive record books.
Durant is a great NBA player from a lot of different angles, but what’s truly remarkable about the Warriors’ acquiring him is the extent to which it doubles down on the team’s greatest (or GOATest) strength: shooting the rock. They’ve added one of the game’s best shooters to what was already the best shooting team in NBA history — without sacrificing defense or all-around skill at his position. We’ve never seen this concentration of shooting talent on one team. Actually, we’ve never seen anything even close to it. There may be stumbles along the way, but this could also be something entirely new.
So let’s break it down.
Shooting Dwayne Johnson
Using NBA player-tracking data for the 2015-16 season, I brewed my own model for expected points for a given shot based on its location, the position and height of the nearest defender, how many dribbles the player took, how long he held the ball for, and how much time was left on the shot clock. Using this, we can calculate points contributed relative to expectation for each shot taken and thus how much value players added from shooting ability alone.2
Note that this doesn’t account for shot creation or any of the myriad other ways that a player can help his team score more efficiently — it tells us just, for a given situation, how good the player is at putting the ball into the net. (For instance, a player who gets to the rim at will but performs slightly below average on his shots once he’s there is still hugely valuable, because shots near the basket are valuable. But that’s not captured fully here; this measure doesn’t show how a shot was created, only the circumstance when it was taken and what happened.) It also doesn’t count shots during which the shooter was fouled — again, it’s about pure shooting ability.
Here are charts for how much value each player in the league with at least 100 (unfouled) 2-point or 3-point shots in the 2015-16 data contributed from this skill:
OK then. Curry’s outlandish outlier-ness from 3-point range is well-documented. That Durant has been nearly as much of an outlier from inside the arc is not as well-known. Curry at least has a few players pushing him on the efficiency front (albeit over far fewer shots); Durant is pretty much all alone.
Of course, Durant is no slouch from downtown. Although he took a much higher percentage of his shots from 2-point range than “Splash Brother” Klay Thompson did for Golden State, Durant’s 3-point value added (0.17 points per shot) was only slightly below Thompson’s (0.20). Durant shot “only” 39 percent on threes last season, compared with Thompson’s 43 percent, but remember: The “points added” we’re charting here are added to expectation, and Durant’s average shot was expected to score 0.98 points compared with Thompson’s 1.08, meaning that on average Durant’s threes were far more difficult than Thompson’s. In fact, Durant’s shots were more difficult than even notorious chucker Curry’s (1.01). Meanwhile, Harrison Barnes (the man Durant is replacing at starting small forward) shot 38 percent — good for a max contract from the Dallas Mavericks — but did so with an average shot value of 1.15 points — on which he shot just slightly below expectation.
The Warriors’ hoarding tendencies are even more stark when we combine 2- and 3-point value added:
For combined points added from shooting efficiency (size of circles in the chart), Curry and Durant are the top two shooters from 2015-16 overall — and by a crazy margin: Curry added 380 points, Durant added 297, and third-place Kawhi Leonard of the Spurs added 172. It would be like if your team already had Lionel Messi and Neymar and then you went out and got Luis Suarez just for the heck of it.
At what cost?
Imagine you’re the Warriors. You’re loving this idea of having three amazing shooters who can shoot with a defender in their face, can open up shots for others, and can push back the perimeter. Great. But at what cost? Barnes did other stuff. He defended 616 shots!
OK, I can’t keep that up. Barnes gave up about 1.02 points per shot. Not terrible, but not good. Meanwhile, Durant defended 810 shots, giving up — get this — 0.86 points per shot.
As with shooting, defensive stats have to be considered in context, especially because they are so heavily affected by the different kinds of defensive assignments a player gets. Looking at players’ “expected points” per shot defended can be valuable, although that should be done cautiously, as it doesn’t reflect everything that led up to the shot (like the defender getting close to his opponent in the first place or being 7 feet tall). But let’s compare Durant to Barnes: The typical shot that Durant defended had about 0.04 points greater expectation than the shots Barnes did, meaning that Barnes was “forcing” his man into what should have been tougher shots. But the shooters Durant faced shot around 0.14 points per shot below expectation against him, while Barnes’s opponents shot 0.06 points above expectation against him.3
Although I don’t think the “defense” skill is as reliably measured as the “launching the ball through the hoop” skill, it’s worth noting how much of an outlier Durant is. If we calculate defensive “anti-shooting” value added — by combining expected points added or lost per shot with volume (as with shooting above) — we can estimate at least one aspect of defensive contribution.
Durant’s opponents scored 116 points below expectation last season, second in the NBA behind, yep, his new Warriors teammate Draymond Green (whose opponents shot 168 points below expectation on 1,307 shots). Durant may not have drawn the toughest assignments on the Thunder all year (those typically went to Andre Roberson or Steven Adams), but the sum total of his defensive contribution is hard to ignore.
Combining this defensive value added with the offensive value added above yields this:
Again, neither of these axes should be taken as the definitive representation of value for their respective genres, but even as rough proxies, they paint a pretty vivid picture. We’ve seen other superstar players affect the game on both ends of the floor — LeBron James and Tim Duncan at their peaks would look fearsome on this chart — but those all-around players have, to this point, had fairly all-around games that fit more traditional strategies, like attacking the rim or working out of the post. Durant can do that too, but he’s also something those other Hall-of-Famers weren’t: a dead-eyed sniper.
Normally a collection of sharpshooters would have to be balanced out by some guys capable of carrying the burden defensively (or otherwise diverting resources to defense). But Durant’s length and defensive versatility allow his team to put even more shooting capability on the floor than would normally be possible.
Cracking the game
That the Warriors have collected a lot of valuable, productive players is clear. But for them to match — much less exceed — their aggregate value, they have to be more than that.
The Warriors offense is built on the fact that Curry is virtually impossible to defend. That is, even his bad shots are pretty good shots, and his good shots are great shots. If every Warriors possession included a typical Curry three (on which he shot over 45 percent), they would score somewhere around 140-150 points per game (depending on their pace and offensive rebounding). Thus defenses have to devote insane amounts of attention to make sure that he doesn’t have any shots available, and the rest of his team is largely tasked with hitting the open shots defenses leave behind.
The Warriors courted Durant with promises of no more clogged lanes, no more standing at the 3-point line for entire possessions waiting for a pass that never comes, and visions of open shot after open shot. So can they deliver?
Using our player tracking data with NBA.com’s definition of “defended” (meaning a defender within 4 feet of the player when he shoots), we can compare how players shoot with and without the heat:
Note that Green and Barnes both shot better than 40 percent when left open, which was all the time.
No really. This tracking data includes 462 3-point shots by Green and Barnes, only 28 of which were defended; of those, they made six. Six. Thompson had 158 defended attempts (26 percent of his attempts overall) and made 41 percent of them.
If the premise of opposing defenses is to cover Curry at all costs, the premise of the Warriors offense is that those defenses can cover only so much ground. Eventually, they’ll lose track of one of the other guys.
Except now the “other guys” include Durant, who last season had 187 defended attempts (41 percent) and made 37 percent.
With Curry shooting 38 percent on a whopping 227 defended attempts, this gives the Warriors three players who shot better than 37 percent on a high volume of such shots last season. Factoring in chances of getting an offensive rebound, if a team could get a 37 percent 3-point shot on every possession, they’d score more than 120 points per 100 possessions. The best offenses in NBA history managed around 115.
So, yeah, Durant will be open more often than he’s used to. But he’ll also probably draw more defenders away from Curry than Barnes did because of his ability to put the ball on the floor (remember all those twos he takes, and makes?), freeing Curry for more open looks — and you can see what he does with those.
One of the thrilling — and honestly bizarre — aspects of Curry’s 2015-16 season was his sudden development into the league’s pre-eminent ultra-long-range shooter. Before Curry took long range to a new level last season, Damian Lillard of the Trail Blazers was about the only game in town. Only a handful of players have shown any serious capacity from beyond 26 feet — right around the distance where even good looks by good shooters become inefficient — and the Warriors now have three of them.
This can be a bit tricky to get at because so many shots taken from beyond 26 feet are “involuntary” — meaning that they’re taken when time is running out and a player is forced to shoot or risk not getting a shot off at all. I set this boundary at four seconds. Note that voluntary shots from long range tend to be very good looks, as otherwise the shooter wouldn’t take them. For involuntary shots from 26 to 28 feet last season, the league as a whole shot 23 percent. For voluntary shots from the same distances, it shot 34 percent. Curry, Durant and Thompson shot a combined 45 percent:
Among players with at least 30 attempts from 26 to 28 feet, Curry, Durant and Thompson have three of the top five shooting percentages. In fact, for either of Durant or Thompson, the only NBA player who has both taken more shots from 26-28 feet and hit a higher percentage is Curry. Only eight players even shot above 40 percent.
So what’s the big deal about 26-28 feet? It’s only 2 feet behind 24-26 feet, where most 3-point shots are taken,4 right? Well, imagine a 2-foot-wide arc from sideline to sideline 26-28 feet from the basket. We can find the area of that region using a little basic calculus: just over 128 square feet.
That’s an extra 128 square feet of real estate for an offense to work with! Or, put the other way, an extra 128 square feet that the defense has to defend. Against three of the league’s most willing and able shooters from that distance.
Imagine that the Warriors played a largely vanilla offense with the primary aim of trying to get any one of these three guys a decent 3-point look from anywhere inside 28 feet. Say they found such a shot on a third of possessions and had a league-average offense on the rest. That would already get them 120 points per 100 possessions5 — the highest in history by a wide margin. Again, duplicated skills with the potential to lead to exponential returns — and this isn’t even the most optimistic scenario.
None of which is to say that I think the Warriors will win 73 again or the like — certainly that took a fair amount of luck, in addition to skill. Their plan may not work at all, and even if it does, it may take them awhile to figure it out. But just as last season we saw things that were hard to conceive of beforehand — however much we suspected Curry and the Warriors were doing something revolutionary — this team should be charting new territory. And whether I think it’s fair or not, I am dying to see what happens.
Check out our latest NBA predictions.