The Warriors are a cresting wave

5-4-16 Continuity Cover Art
5-4-16 Continuity Cover Art /

After Tuesday’s come-from-behind victory, the Golden State Warriors are leading the Portland Trail Blazers two-games-to-none in the Western Conference Semifinals, with Game 3 to be played in the Rose Garden on Saturday evening. That the Warriors have been able to establish a 2-0 series lead without the services of their best player, Stephen Curry, is a testament to the quality, depth, and cohesion of the entire roster. In contrast, the Portland Trail Blazers roster is comprised of mostly young and unproven talents, with little experience playing together in the playoffs or otherwise. As such, this second round series between Golden State and Portland is providing a useful opportunity to observe the impact of roster continuity on team success.

Portland’s Low Tide

The Portland Trail Blazers had a rough offseason. The 2014-15 season had been another solid one; they finished with more than 50 wins for the second consecutive year and things were looking up. The squad was led by the young and talented quintet of Damian Lillard, LaMarcus Aldridge, Nicolas Batum, Wesley Matthews, and Robin Lopez. However, a first round playoff exit precipitated an exodus from Portland; first Aldridge and eventually four of the five Rip City starters left the team: Aldridge signed with San Antonio, Batum was traded to Charlotte, Matthews signed with Dallas, and Lopez signed with New York. Additionally, reserves Steve Blake, Aaron Afflalo, Joel Freeland, Dorell Wright, Thomas Robinson, Will Barton, Alonzo Gee, and Victor Claver left the team, too. All in all, the Blazers lost 12 players who had accumulated 12,962 minutes in 2014-15, which was 65% of the total team minutes played. Portland only retained one starter, Lillard, and five reserves, Chris Kaman, C.J. McCollum, Meyers Leonard, Allen Crabbe, and Tim Frazier. Franchises have ebbs and flows, and suddenly, the Blazers seemed like they were in a very low tide.

Minute distribution for the Golden State Warriors and the Portland Trail Blazers in 2014-15; gray players were traded or left via free agency the following summer.

5-4-16 Minutes Played GSW POR 2014-15
5-4-16 Minutes Played GSW POR 2014-15 /

In stark contrast to Portland, Golden State returned nearly their entire roster from the 2014-15 championship season. They released former All-Star David Lee in a money-saving move, Justin Holiday signed with Atlanta, and Ognjen Kuzmic went to play in Europe after the Warriors rescinded his qualifying offer. But, aside from these three players, the Warriors brought everybody back, including each of the top-nine rotation players: Stephen Curry, Draymond Green, Klay Thompson, Harrison Barnes, Andre Iguodala, Andrew Bogut, Shaun Livingston, Marreese Speights, and Leandro Barbosa. Overall, the Warriors retained 12 players and 18,097 minutes, 92% of the team’s total for 2014-15.

Basketball Reference has compiled their own measure of roster continuity, which they define as the proportion of a team’s regular season minutes that were filled by players from the previous season’s roster. Note, this is a different, but complementary way of calculating continuity from the one above. Using this historical data available from as early as 1952-53, the histogram below demonstrates that the Warriors and Blazers represent the two opposite ends of the continuity continuum.

Histogram of roster continuity (% of minutes given to returning players) for each NBA team in each regular season since 1952-53.

5-4-16 Continuity histogram
5-4-16 Continuity histogram /

By this measure, the 2015-16 Blazer team had 47 percent roster continuity, which is equivalent to the 11th percentile among all teams in NBA history since 1952-53, i.e., 89 percent of all teams in NBA history had more continuity than this Portland team. The Warriors had 95 percent roster continuity, which is equivalent to the 97th percentile, i.e., only 3 percent of all teams in NBA history had more continuity than this Golden State team. The most common level of continuity (the mode) was to have 79% of the minutes played by returning players; 43 different teams achieved this level of continuity. This season, the Warriors had the most roster continuity in the league and only the Los Angeles Lakers (46 percent) and the New York Knicks (42 percent) had less than the Blazers.

The Warriors’ Cresting Wave

In addition to the minutes-lost comparison, it’s helpful to examine the quality of the players lost by each franchise from last year’s teams. Here, I’ll use the Value Over Replacement Player (VORP) summary statistic to quantify a player’s quality. For all my VORP calculations, I’m focusing on players with positive VORP and setting aside minor contributors with negative VORP.

Value over replacement player (VORP) distribution for the Golden State Warriors and the Portland Trail Blazers in 2014-15; gray players were traded or left via free agency the following summer. Players with non-positive VORP are not shown.

5-4-16 Player VORP in 2014-15
5-4-16 Player VORP in 2014-15 /

One thing that is evident from the above plot is that most of the players that the Blazers lost during the summer of 2015 were not significant contributors to the team in the 2014-15 season. It was really just the starters who had positive VORP and it would be those key players who would be missed most dearly. The same can be said for the players who stayed; Lillard was the only one among the six returning players who made a meaningful contribution to the Blazers last year. Taken together, the Blazers lost most of the quality on their roster; 9.1 of the team’s total 15.6 positive VORP (58 percent) did not return in the 2015-16 season.

While the Blazers brought back only one proven player in Lillard, the Warriors kept all six of their most important contributors. In all, this season’s incumbent Warriors accounted for 21.2 of the team’s total 22.3 positive VORP in 2014-15 (95 percent).

The decision to keep the Warriors together during the summer of 2015 was not arbitrary; the Golden State front office has carefully curated this roster, buoying the team with helpful pieces and jettisoning superfluous ones to build a deep and versatile collection of players. When Joe Lacob and Peter Guber bought the team in 2010, they identified Curry, the team’s first round draft pick from the previous year, as their most crucial asset and franchise cornerstone. In April 2011, Bob Myers joined the Warriors front office as the Assistant General Manager and he started to help fill in pieces around Curry.

In the 2011 draft, the Warriors selected Thompson, Washington State’s 6-7 shooting guard, to pair with the smaller Curry in a backcourt that would eventually become the Splash Brothers. Myers was promoted to General Manager after one season and in the summer of 2012 — in his first draft at the helm — he added Barnes, Festus Ezeli, and Green. Not bad! During the subsequent 2012-13 season, Myers traded team leader, Monta Ellis, for injured big man, Andrew Bogut, in an unpopular (at the time) personnel decision. The Curry-led Warriors made their first playoff appearance in 2013, impressing many observers, most importantly, their first-round opponent, then-Nugget, Iguodala. Iguodala joined the Warriors in the summer of 2013, providing the final element of the Dubs’ Championship core.

We can visualize this roster building process by plotting the VORP contributed to the team by each of the current Warriors starting with the 2008-09 season, the year before Curry was drafted. The plot illustrates how smart negotiations in the trade and free agency markets, effective drafting, and player development have combined to create a cresting wave of quality on the Warrior’s roster, with the team VORP total swelling as high as 23.5 this season.

Value over replacement player (VORP) distribution for the Golden State Warriors and the Portland Trail Blazers from 2008-09 to 2015-16; gray players have been traded or left via free agency prior to the 2016 Playoffs. Players with non-positive VORP are not shown.

5-4-16 VORP GSW POR from 2008-2016
5-4-16 VORP GSW POR from 2008-2016 /

The Blazers, on the other hand, are in a much different phase of their team building process. The two players who have been with the Blazers the longest, Lillard and Leonard, joined the team as recently as the 2012 Draft. As discussed above, nearly all of the quality players that have sustained the Blazers success over the last few years left the team this summer, as evidenced by the substantial shadow of missing VORP looming over the Portland half of the plot. The consensus at the start of the season was that the Blazers would not be able to compete for a playoff spot in the vaunted Western Conference.

However, free agent additions, Ed Davis and Mason Plumlee, have supported Lillard with useful contributions and C.J. McCollum has had a breakout year, winning the league’s Most Improved Player Award. Consequently, the team has surprisingly found itself in the second round of the playoffs. Perhaps, then, rather than being at a perceived low point, maybe the Blazers are actually just a year or two behind the Warriors trajectory towards Championship success. After all, they do have the highest quality players under the age of 26 in the league. And they are essentially following the Warriors playbook of building their team around a young point guard who is a virtuoso on the offensive end of the court.

All that optimism is well and good, but, at least for now, the Blazers look to be too young to compete with the battle-tested Warriors. At the end of Game 2, the Warriors’ wave of talent came crashing down on the Blazers’ heads and the young squad drowned in the surf or, at least, well, they definitely choked. The Blazers led most of the game thanks to an incredible collective shooting performance in the first three quarters, but they managed a meager 12 points in the final period, converting on only 1-of-10 field goals from outside of the lane and tossing the ball out-of-bounds on multiple occasions. The Warriors finished the game on a 16-4 run and won handily in the end.

Golden State often win games like this, by stringing together stop after stop on the defensive end. It’s just so easy to be distracted by Curry’s ridiculous shot making, that sometimes we don’t notice. In some ways, it’s more straightforward to see the Warriors true nature with Curry temporarily sidelined; they are a defensive team! It is on this end of the floor that continuity has been most critical to their success; the Warriors are well-practiced at the choreography and communication that are required to employ their switch-heavy, undersized defensive schemes. After years of playing together, each individual knows exactly where his teammates will be at all times and their movements are all perfectly synchronized.

Of course, one advantage of youth is a certain brashness and the Blazers have already demonstrated that they can defy the odds, toppling the Clippers with four straight victories in Round 1 after trailing that series by the same margin, 2-0. They will hope to repeat this feat against Golden State starting on Saturday and only time will tell if they will be able to break free from the Warriors’ undertow.

Do Wins Flow from Continuity?

OK, so the Warriors and Blazers represent historically extreme levels of roster continuity — a discrepancy that seems to be having a demonstrative effect on their performances in the current playoff series — but, is continuity inherently predictive of winning? For one clue, we can plot the roster continuity of each NBA team versus their subsequent regular season winning percentage using the historical Basketball Reference data.

Regular season win percentage as a function of roster continuity (% of minutes given to returning players) for each NBA team in each regular season since 1952-53.

5-5-16 Continuity vs win percentage
5-5-16 Continuity vs win percentage /

Evidently, there is a moderately strong association between continuity and winning (correlation of r = 0.47), albeit with a wide spread in the data. What is less clear is the order of the relationship, i.e., what comes first: the continuity or the winning? Undoubtedly, there have been many teams, like the Warriors, who embraced continuity and player development and were subsequently rewarded with success. Alternatively, there may be other teams who happened upon a winning roster spontaneously and subsequently did their best to keep it together for as long as possible.

To untangle the direction of the relationship, in the bottom panel of the figure above, I repeated the same plot three times, stratifying for team record in the previous regular season. I used win percentage cut-offs of <40%, 40-60%, and ⋝60%, to divide the teams into three groups which roughly corresponded to the lowest quartile of the league, the highest quartile of the league, and everybody in between. As you might expect, continuity isn’t as helpful when you start with a bad team; the relationship between continuity and win percentage was weakest among the teams who won less than 40% of their games during the previous season. Continuity had more of an impact on team success when the team was good or, at least, not bad the previous season.

Both the 2015-16 Blazers and the 2015-16 Warriors were among the top quartile of teams, with 2014-15 winning percentages of 62.2 percent and 81.7 percent, respectively. In this rarified group, the Blazers’ tear down was even more unusual; only 12 other teams in NBA history who had won at least 60 percent of their games in the previous year subsequently experienced as much change as Portland did this year. Interestingly, history suggests that the Blazers did about as well as expected this season. Their 53.7 percent winning percentage was bang-on what would be expected based on their team’s quality the previous season and their level of continuity (as can be quickly surmised by their position on top of the trend line in the bottom right panel of the above figure); five of the other 12 teams with the similarly low continuity in this similarly good group of teams finished with better records than the 2015-16 Blazers’ mark and seven finished worse off.

Stepping in a River

The Warriors are clearly building something special right now. They’ve just backed up last year’s championship-winning effort with the best regular season ever. Obviously, the Warriors’ front office would like to sustain the team’s success for as long as possible. They’ve already embraced continuity by opting not to trade Klay Thompson for Kevin Love in the summer of 2014, but how far will they go to keep this team intact? How important will continuity be to this team in the near future? To gauge the impact of continuity on the type of sustained success that the Warriors covet, it’s instructive to use the historical Basketball Reference data to trace the path of the four most powerful dynasties in NBA history: Bill Russell’s Boston Celtics (1957-58 to 1968-69, ten Championships), Magic Johnson’s Los Angeles Lakers (1980-81 to 1989-90, four Championships), Michael Jordan’s Chicago Bulls (1990-91 to 1997-98, six Championships), and Tim Duncan’s San Antonio Spurs (1998-99 to 2015-16, five Championships).

Roster continuity (% of minutes given to returning players) for each NBA team in each regular season since 1952-53; four great NBA dynasties are highlighted.

5-4-16 Continuity vs time NBA dynasties
5-4-16 Continuity vs time NBA dynasties /

The Celtics, Lakers, and Bulls each maintained a remarkable level of continuity during their respective dynastic reigns, staying well above the league average for continuity year after year. In each of the 12 seasons highlighted, the Celtics gave at least 80 percent of the team’s minutes to returning players; likewise, the Lakers had a minimum continuity of 82 percent; and — with the exception of one substantial retooling of role players that corresponded with Jordan’s season in the minor leagues — the Bulls too dipped no lower than 80 percent continuity. These were the old days though, and a modern dynasty needs to have more flexibility.

Continuity is especially challenging in today’s NBA landscape, because players have ample opportunity to change teams via free agency and owners have (self-imposed) obstacles to paying their players what they are worth. I’ve written briefly about how the Spurs’ roster structure has continually pivoted around Tim Duncan over the years and the Spurs mini-rebuilds are evident in the plot above as well. There were three separate instances when San Antonio’s continuity dropped down around 60 percent: in 2001-02 (59 percent), 2003-04 (62 percent), and 2009-10 (61 percent).

The first shift, in 2001-02, saw Tony Parker and Steve Smith joining the team to replace the departing Derek Anderson, Sean Elliott, and Avery Johnson. Soon after, in 2003-04, the Spurs lost their long-time leader, David Robinson, along with three guys named Steve (Jackson, Kerr, and Smith), they gained valuable contributors, Rasho Nesterovic and Robert Horry, and they gave more minutes to Parker and Manu Ginobili. Finally, in 2009-10, Bruce Bowen retired and the Spurs added Richard Jefferson, Antonio McDyess, and Keith Bogans. Not coincidentally, the Spurs produced a championship soon after each of these big changes, in 2003, 2005, and 2014. The Spurs have rebooted again this year swapping out role players Tiago Splitter, Corey Joseph, Marco Belinelli, Aron Baynes, and Jeff Ayers for LaMarcus Aldridge and David West, resulting in a relatively low continuity score of 69 percent and another fantastic regular season record of 67-15.

Plato attributed his fellow Greek philosopher, Heraclitus of Ephesus, with the aphorism that “you could not step twice into the same river.” Heraclitus was pointing out that change is inevitable; his observation was like the ancient predecessor of “YOLO!”. The same has been true for San Antonio and it will be true for Golden State’s roster too. In theory, striving for continuity seems like a no-brainer for a team that just completed the winningest season of all time. But, in practice, the Golden State front office can expect the Warriors’ river to keep flowing; Iguodala, Bogut, Livingston, and Barbosa are all on the downstream side of 30 and each of these players are likely to give way to younger replacements by the end of the decade.

The Warriors commitment to continuity will be seriously tested as early as this summer as Harrison Barnes, Festus Ezeli, and many of the Warriors’ other role players will be free agents looking for raises. Despite their best intentions, it may be essentially impossible for the Warriors to keep this team together again this summer, at least not to the extent that they did last summer. But for now, the Warriors will continue to ride their cresting wave in pursuit of a second straight NBA championship.