Steve Kerr: Lineup Wizard

Golden State Warriors Head Coach Steve Kerr has a lot of great players at his disposal and his team’s frequent big leads provide opportunities for him to experiment with exotic lineup combinations. I imagine Kerr scrolling through his roster like 10-year-old-me picking an NES Ice Hockey lineup: agonizing over the relative strengths and weakness of the fat slow pokes, normals, and skinny speedsters.  Or maybe it would be less of a stretch to think of Kerr choosing between the likes of Jammer, Mr. Doc, and Legs from NES Hoops.  Either way, Kerr is The Wizard of NBA lineups, a substitution savant.

Small Ball is Steve Kerr’s Cheat Code

The Warriors employ three basic lineup types: those with zero, one, and two traditional big men.  Here, when I refer to “traditional bigs” I’m including Warriors centers and power forwards: Andrew Bogut, Festus Ezeli, Marreese Speights, Anderson Varejão, James Michael McAdoo, and Kevon Looney as well as former big Dubs: Jason Thompson, David Lee, and Ognjen Kuzmić.  The modern hybrid big men that I’m keeping in a separate category are 6’7” Draymond Green and 6’8” Harrison Barnes.  Below is a representation of the productivity of these three Warriors lineup types during this season.  In this plot, the horizontal axis represents the minutes played per game by each lineup type, so that a wider box equates to more playing time.  The vertical axis represents the point margin per minute produced by each lineup type, so that a taller box equates to a more potent lineup type.  The area of the boxes show the point margin per game of each lineup type (point margin/minute X minutes/game = point margin/game).  Summing the areas of the three boxes yields the team’s point margin for the season, +11.0 points per game.    

As you can see, Kerr has generally left at least one big on the floor (85% of the time). Of all the possible permutations, his most used combination has been the starting lineup: Stephen Curry, Klay Thompson, Barnes, Green, and Bogut (490 minutes together, 7 mpg), which is an example of a one-big-man lineup. The small-ball lineups have been much more effective than the one-big lineups (0.53 vs. 0.23 net points per minute); however they have been employed more sparingly (15% of the time). The Warriors’ quintessential small-ball lineup has been Curry, Thompson, Andre Iguodala, Barnes, and Green (150 minutes together, 2 mpg). One-big lineups, such as the starting five, and small-ball lineups have both made important contributions to the team’s success this season (+8.6 and +3.8 net points per game, respectively). In contrast, the two-big-man lineups have struggled (-0.37 net points per minute). The most common combination of this lineup type has been Shaun Livingston, Leandro Barbosa, Iguodala, Speights, and Ezeli (42 minutes together). Generally, these two-big lineups have consisted exclusively of the Warriors reserves and they have tended to specialize in closing out big fourth quarter leads, sometimes not very successfully.

Using this technique for lineup visualization, it’s fun to see the changes in how the Warriors have employed these three lineup types over the past two seasons. Below I show four panels; one each for the 2014-15 regular season, the first three rounds of the 2015 Playoffs, the 2015 Finals, and the 2015-16 regular season. The panels all use the same scale, so that the size of the rectangles can be compared between time periods.

Throughout the 2014-15 regular season, Coach Kerr experimented with small ball, to great effect.  Although the small-ball lineups were immediately productive (0.55 net points per minute), Kerr was reluctant to lean too heavily on these units and he opted to play without a big only 7% of the time.  In the first three rounds of the 2015 Playoffs, Kerr employed the small-ball units more frequently (17% of the time) with only marginally decreasing returns (0.48 net points per minute).   In the Finals against Cleveland, Kerr furiously started pressing: UP, UP, DOWN, DOWN, LEFT, RIGHT, LEFT, RIGHT, A, I, START, which is the Small Ball© cheat code that makes Andre Iguodala (“AI”, see what I did, there) invincible.  Kerr famously decided to give Iguodala his first start of the season in Game 4 of the Finals, with his team trailing the Cavs two-games-to-one.  The Warriors never looked back, winning the next three games of the series and also the Championship.  In that series, Kerr used his small-ball units more than ever before (40% of the time) and these were the only lineups that provided an advantage for the Warriors (+9.0 net points per game).  Kerr basically abandoned the two-big-man combos for this series, playing these units for only 1.2 minutes over six games.

Two Keys Unlock Warriors Small-Ball Lineup

Not all of the Warriors’ small-ball units are created equal. There are essentially three categories of small-ball lineups:

• Small ball = 0 traditional bigs, Green at center, Barnes at power forward
• Smaller ball = 0 traditional bigs, Green at center, Iguodala et al. at power forward
• Smallest ball = 0 traditional bigs, Barnes at center, Iguodala et al. at power forward

This season, Draymond is getting a lot of well-deserved praise.  His defensive flexibility — he can bang with Zach Randolph or slide his feet to cut off a Chris Paul dribble-drive — is now widely recognized as a key element to the success of the Warriors’ small-ball scheme.  Although less heralded, HB’s strength in the defensive post, his work on the defensive boards, and his ability to switch defensive assignments on any opponent’s screen are likewise critical to creating the havoc on which the Warriors small-ball unit thrives. If Draymond is the Boss Key, then Barnes is the Small Key.  Kerr needs to couple both keys to unlock the full potential of the small-ball lineup.  The plot below shows the productivity of the three categories of small-ball lineups (note: this is essentially a zoomed in version of the blue box from the plot above).  

Obviously, Kerr recognized the power of the Green/Barnes combo right from the start. His desire to keep this duo together explains why, at the start of last season, the Lineup Wizard asked Iguodala (a former All-Star) to come off the bench as a reserve. This year, Kerr has featured Green and Barnes together in 65% of his small-ball lineups. These lineups are the most effective of any that the Warriors have, producing 0.71 net points per minute. Due to a mid-season ankle sprain for Barnes, Coach Kerr also employed several smaller-ball lineups with Brandon Rush, Andre Iguodala, and others playing the nominal power forward position (29% of the small-ball minutes), but these lineups were less effective (0.24 net points per minute). Likewise, during the rare moments that Barnes was asked to fill the center role (5% of the small-ball minutes), the team did not flourish (-0.07 net points per minute).

To put the success of the Warriors small-ball units in proper historical context, I used the lineup finder from Basketball-Reference. Of the 2,404 five-man combinations that played at least 100 minutes for any NBA team since the 2000-2001 season (first available data) — about five qualified lineups per team per season — the Warriors favorite small-ball lineup was the second-most productive overall (0.96 net points per minute).

Saving Steph’s Extra Life for the Final Boss Fight
Last week, Seth Partnow showed us how the Golden State coaching staff has been conserving the effort of their star player, Stephen Curry, by providing frequent fourth quarter rest.  Here’s another way to visualize Steph’s level of effort, which emphasizes the intelligent influence of Coach Kerr.  The boxplot below shows the number of minutes Steph played in each of three types of games: close games (<10 point margin), cruise-control games (10-19 point margin), and blowout games (20+ point margin) under three different Warriors head coaches: Mark Jackson (light blue, during his final two seasons: 2012-13 and 2013-14), Luke Walton (bright blue, during the first 43 games of the 2015-16 season), and Steve Kerr (dark blue, during the 2014-15 season and the remainder of the 2015-16 season).     

The plot shows that the expected minute-vs-margin trend — more minutes for Steph in closer games — was evident during the tenure of all three head coaches.  However, it’s interesting to note that, in closely contested games (i.e., <10 point margin), Kerr typically showed much more restraint with Curry’s effort than his predecessor, by about five minutes per close game (medians: 40.4 vs. 35.4 for Kerr and Jackson, respectively).  Compared to Kerr, Walton was also more reluctant to sub out Curry during close games (median of 37.7 minutes).  Walton got the Warriors off to 24-0 start and led them to 39 wins in his 43 coaching appearances, but he used Curry as a security blanket in close games.  In contrast, Kerr is unflappable and cold-blooded with Steph sitting on the bench.  He consistently trusts his bench units in close games and he’s been playing the long game with Curry’s minutes, sacrificing short-term point production for long-term health and stamina.  Kerr’s doing his best to save all of his one-ups for the big bosses down the road, including the San Antonio Spurs’ two-headed point guard monster of Tony Parker and Patty Mills.  

the final two years of Jackson’s tenure as head coach (2012-13 and 2013-14), Curry’s heavy second-half minute load led to diminished effectiveness in the fourth quarter. Combined in those two seasons, Curry’s effective field goal percentage was 55.6% during the first three quarters and 54.7% during the fourth quarter plus overtime (-0.9%). In contrast, this year, Kerr is asking Curry to do less in the second half, typically resting him for at least the first six minutes of the fourth quarter, longer in blowouts. As a result, Curry’s fourth quarter effective field goal percentage has skyrocketed to 65.8%, better than his (also improved) 63.0% effective field goal percentage for quarters one through three (+2.8%) Whereas, Steph used to be visibly worn down by the end of a close contest, he now has the energy to continue hoisting shots from 30 feet and scooping leftie layups until the buzzer. No doubt Kerr is hoping that Curry’s improved fourth quarter efficiency will be a microcosm for his productivity in the fourth quarter of the NBA season. I’ll be watching to see if the Lineup Wizard can post a new high score during this last week of the regular season.