Back in 2015, Seth Partnow, now a member of the Milwaukee Bucks analytics team, did a great job analyzing how each team deployed lineups featuring each possible number of starters, from 0 to 5. Inspired by his approach, I gathered data on starting and bench units for the current 2018-19 regular season, through December 30th. I was particularly interested in the question in the title of this article: How important is a good bench to NBA success?
How Often Do Teams Play Bench Majority Lineups?
For the purpose of analysis, I call lineups which have 3 or more starters starter majority lineups and those with 2 or less starters bench majority lineups. The bench majority lineups are not a perfect proxy for what we might consider true “bench” minutes; after all, Curry and Durant plus 3 reserves is technically considered a bench lineup by my definition. But still, I think it’s a logical place to start an analysis.
To orient ourselves, here are some basic facts: Starter majority lineups have played about 63% of the minutes across all teams, and thus bench majority lineups account for the remaining 37% of minutes. The bench majority lineups play a minority of the minutes, but it’s not a tiny minority. True “bench mob” lineups, those with 0 or 1 starter, have played about 22% of all minutes. Of course, some of these minutes are coming at the end of lopsided games and are thus less important.
The statistics cited in the previous paragraph are totals across all teams; some teams deviate from these base rates in interesting ways. The Sixers, for example, have staggered their substitution patterns so that 2 or more starters are on the court for 92% of their minutes. Their most common lineup composition has 3 starters. The Raptors, on the other hand, like playing all 5 of their starters together. At 38% of their total minutes, the Raptors are second only to the Mavericks in terms of proportion of minutes playing all 5 starters.
How Effective are Starter and Bench Majority Lineups?
Of course, the teams not only vary in how often they deploy starter and bench heavy units, but also how effective these units actually are. To measure this, I simply calculated the net points (team points minus opponent points) per game for each team, at each number of starters from 0 to 5. Then I aggregated this information to determine which teams have the best and worst starter majority and bench majority lineups in terms of net points per game.
In the graph above, you can see that something is going right for the Raptors and Bucks when they play 5 starters and 3 starters, respectively. The Raptors (the orange line) have outscored opponents by a staggering 7.5 points per game when they deploy all 5 of their starters, almost 3 points per game higher than second place Oklahoma City. The Bucks, meanwhile, have been leaps and bounds better than every other team when playing 3 starters, at about +5 net points per game. That spike in the green line at 3 starters jumps off the page!
You can also clearly see where the Spurs excel. For each lineup composition from 0 starters to 3 starters, the Spurs are solidly above average. When the Spurs are playing 4 or 5 starters, however, the story shifts. They are in fact the 3rd worst team in the entire NBA in terms of 5 starter lineup net points per game, at almost -2 points per game.
To give you an idea of where the best and worst teams stand in terms of starter majority and bench majority net points per game, here are the top 5 and bottom 5 teams in both statistics. You can see the full table of teams here.
How Important Are Bench Majority Lineups?
Scanning the tables above, we can see that the best starter majority lineups have higher net points per game than the best bench majority lineups. The Bucks and Raptors are outscoring teams by more than 8 points a game when they play starter dominated lineups, while the Pacers and Spurs can’t crack +3.0 with bench majority units.
In fact, the marginal value, in terms of points per game, between above average and below average starter play is greater than the marginal value between above average and below average bench play. To see this, look at the graph below which plots each team’s starter and bench net points per game. The gap between the 7th best starter majority team (the Pacers) and the 7th worst starter majority team (the Nets) is roughly 6 points per game. For the bench majority units, that same gap between 7th best and 7th worst is about 4 points per game. Over the full 82 game season, I have found that 1 point of point differential is worth about 2.5 wins on average, so the difference between 6 points per game and 4 points per game could be 5 wins.
This gap between 7th best and 7th worst (roughly the interquartile range) is actually understating just how much the top 4 starter teams have separated themselves from the pack. The Bucks, Raptors, Thunder, and Celtics have all outscored their opponents by at least 6 points per game when playing 3 or more of their starters. To be really upper echelon, at least this season, you need to dominate the starter minutes.
Though having dominant starter majority lineups is the path to the top of the standings, benches can make a difference too. Two teams stand out as prime examples: the Pacers and the Magic. The Indiana Pacers, leading the league at about +3.0 bench majority net points per game, would be expected to win about 7.5 less games out of 82 if they merely had a +0.0 bench. The Orlando Magic, owners of the worst bench in the league at -5.1, would actually have a positive point differential if they could just survive those bench majority minutes without losing so much ground. They are middle of the pack at +1.7 starter majority net points per game.
The best teams have differentiated themselves from the field with their stellar starter majority lineup play, which is unsurprising given that these lineups play roughly 63% of the minutes. But the difference between a solidly above average and solidly below average bench could plausibly be 3 or 4 points per game, which is still quite substantial.