Nylon Calculus: Golden State’s offense before and after All-Star Weekend

Mar 21, 2017; Dallas, TX, USA; Golden State Warriors guard Stephen Curry (30) in action during the game against the Dallas Mavericks at the American Airlines Center. The Warriors defeat the Mavericks 112-87. Mandatory Credit: Jerome Miron-USA TODAY Sports
Mar 21, 2017; Dallas, TX, USA; Golden State Warriors guard Stephen Curry (30) in action during the game against the Dallas Mavericks at the American Airlines Center. The Warriors defeat the Mavericks 112-87. Mandatory Credit: Jerome Miron-USA TODAY Sports /

Heading into the All-Star Game, the Warriors were scoring points at a record-setting pace — averaging 118.2 points per game and 114.2 points per 100 possessions. That’s an even more lethal attack than they had during their historic 73-9 run last year. In fact, it would have been the best offensive rating in recent NBA history (earliest available data on NBA.com was from 1996-97).

Since the break, however, Golden State’s offensive machine has slowed — as they’ve generated just 108.3 points per game (sixth-best in the league) and 109.1 points per 100 possessions (ninth-best) in the last 17 games. In a broad sense, the decline in productivity is attributable to the absence of Kevin Durant, who has played only 67 minutes during that stretch. But, we can use play-type data from NBA.com to put the Warriors’ offense under the microscope and examine how, precisely, things have changed for them since the All-Star Game.

Before the break

Even when at full strength, the Warriors’ scoring efficiency is predicated not solely on their offensive weapons; but also on the way in which they’re deployed. Coach Steve Kerr and his staff put Durant, Stephen Curry, Klay Thompson, and their teammates in positions to succeed with a fast-paced, pass-happy style of attack. Golden State’s exceptional brand of basketball is characterized by the play types they use most often: fast breaks, spot-up jumpers, off-ball screens, and cuts. Specifically, the Warriors finish more of their plays off of cuts and screens and in transition than any other team in the league.

With Durant in the fold, the Warriors were also more efficient with each of these play types than any other team in the league.

Leading up to All-Star weekend, the Golden State Warriors were using cuts, off-ball screens, and transition more often and more effectively than any other team in the league.

In general, the transition and cut play types are the two juiciest ways to score in the NBA. Taken together, the Warriors were using these most-productive plays about 30 percent of the time before the All-Star break, compared to the league-average frequency of 21 percent.

On the other end of the spectrum, post-ups, isolations, pick-and-rolls finished by the ball handler, and other miscellaneous shots are generally the four least-efficient play types. Prior to All-Star weekend, the Warriors were finishing only 29 percent of their plays with one of these options compared to the league-average frequency of 38 percent.

The frequency and productivity of play types used by the Golden State Warriors compared to the average for the rest of the league, prior to the All-Star Game.

In the plot above, the Warriors’ green blocks are wider and taller than the rest of the league — that’s because — with Durant’s help — they were using these play types more often and more efficiently than other teams in the league.

After the All-Star activities

Since the All-Star Game, the Warriors have been forced to use a different distribution of play types on offense without Durant. They have reduced their reliance on off-ball screens and doubled down on pick-and-roll sets finished by the ball-handler. As a team, Golden State has also experienced a big drop off in efficiency for several of their favorite play categories, including fast breaks, cuts, and off-ball screens.

Since All-Star weekend, the Golden State Warriors have been less effective with their favorite play types — scoring less readily in transition and off of cuts and off-ball screens.

Durant is the Warriors most prolific one-on-one creator, so it’s no surprise that they’ve been running less isolation without him (decreased frequency by -1.5 percentage points). In contrast, the players that are taking Durant’s minutes on the floor — Pat McCaw, Matt Barnes, and Andre Iguodala — rely much more heavily on spot-up jumpers (spot-up frequency of 30 percent or more) than the rest of their teammates, including Durant (14 percent frequency). As a result, the team is using even more of their plays passing and kicking to jump shooters than they did at the beginning of the year. On balance, a shift from iso sets to spot-up jumpers should be a healthy indicator for Golden State’s offense, but there have been other areas of concern without Durant.

KD is a beast in transition, scoring 1.36 points per possession on the break this season (91.4 percentile). While the Warriors have been able to continue finding opportunities to get out and run without Durant (increased frequency by +1.3 percentage points), they’ve been much less efficient with these opportunities (-0.13 points per possession).

Durant is also ultra-efficient in his 2.7 possessions per game as the ball-handler in pick-and-rolls, finishing at a rate of 0.98 points per possession (84.0 percentile), a mark that bests Curry, the team’s primary pick-and-roll option. Without Durant to pitch in on the pick-and-roll duties, the Warriors have been less productive with this play type as well (-0.10 points per possession). The dropoff in efficiency is magnified by the fact that Golden State is more reliant on Curry’s high pick-and-roll sets without the benefit of Durant’s offensive versatility — as they are using pick-and rolls more often than they did before the break (increased frequency by +1.5 percentage points).

The ripple effect is that there are fewer opportunities for the team produced by off-ball screens. The Warriors have three players who have spent a lot of time running around screens this season — Thompson (6.8 possessions per game), Curry (3.2), and Durant (2.9). With Durant out and Curry forced to be on the ball more often, the Warriors are getting fewer shots from these types of plays (decreased frequency by -3.7 percentage points). What’s more, off-ball screens were another effective means of scoring for Durant (1.15 points per possession, 87.2 percentile), and the Warriors have been less efficient on these plays without him (-0.13 points per possession).

One situation where the Warriors have reaped the benefits of Durant’s absence is on the glass. Since they’re missing more shots — with the team’s true shooting shooting percentage dropping from 60.5 before the break to 57.0 after — there are more opportunities for put-backs (increased frequency by 0.7 percentage points).

Projecting into the playoffs

Due to the increased intensity, physicality, and quality of the competition, playoff basketball has its own unique style of play, quite distinct from regular season basketball.

In the playoffs, movement-based buckets are generally harder to come by, forcing teams to use cuts, off-ball screens, and transition plays less frequently in the postseason than they do during the regular season. Instead, playoff teams tend to rely more on isolations and pick-and-rolls finished by the ball handler than they would during the regular season. These tendencies were evident throughout the NBA during the 2016 playoffs, but they were especially apparent in Oakland.

The Golden State Warriors used some play types with less frequency in the 2016 playoffs than they did during the 2015-16 regular season.

Note: The “Rest of NBA” columns represent weighted averages of each playoff team, excluding the Golden State Warriors. The weights were defined by the number of playoff possessions used by each team for each play type. Freq=frequency, i.e., the fraction of all plays; PPP=points per possession.

The Warriors used each of their signature play types less often in the playoffs — dropping the frequencies of cut (decreased frequency by -1.0 percentage points), transition (-1.7), and off-screen (-0.7) plays from the regular season rates. On the other hand, the Warriors leaned more heavily on isolation in the playoffs — bumping the frequency of isos from 6.3 percent to 10.0 percent of all their plays (increased frequency by +3.7 percentage points). Unlike the rest of the league, the Warriors decreased their use of pick-and-rolls finished by ball handlers during last year’s postseason (decreased frequency by -0.4 percentage points); but, the decline was due to the fact that they played most of their first two playoff series without Curry. Indeed, Curry’s playoff injuries also help explain why the Warriors experienced such a sharp drop-off in pick-and-roll productivity last postseason (-0.17 points per possession for the ball handler).

At the start of the year, the Warriors were once again running, cutting, and screening their way to one of the best offensive seasons in NBA history. However, without the services of virtuoso scorer, Kevin Durant, the Warriors have had more trouble converting on their favorite play types, recently. Impressively, they’re still finding ways to succeed and will carry a seven-game win streak into their matchup with Houston, tonight.

Next: Jeremy Lin's return and his impact on the Brooklyn Nets

In the playoffs, the Warriors can once again expect the defensive screws to tighten and they may need to forgo their flow offense at times, like they did last year. But, this time around, the Warriors hope to have Durant at their disposal, as the league’s best fall-back plan. When the game tilts towards isolation and pick-and-roll sets in the postseason, Durant’s contribution on offense will become even more critical. And with an offensive approach that has already proven malleable under the stress testing of a challenging regular season as well as the added versatility that a healthy Durant brings to the table, Golden State should be well-positioned to start scoring like gangbusters again when the playoffs start.