By Ty Aderhold
When one looks at the NBA, it is easy to see that it is a superstar-driven league. Most experts will tell you it is almost impossible to win the championship without at least one superstar.[1] At the same time, many of these experts also like to praise the importance of role players. Marginal bench players who were afterthoughts in the regular season suddenly become one of the critical topics of the postseason.[2] While a role player can clearly swing a single game and perhaps even a series with some hot shooting or great defense,[3] will those contributions really matter over the course of an NBA season? Are the role players on a team like the Spurs really any better than the role players on the Trail Blazers? I decided to look for differences across multiple stat categories for role players on playoff teams and non-playoff teams to determine if winning teams really do have role players that are better in certain facets of the game.
Using data from the 2006 season on, I defined a role player as any player who averaged between 15 and 30 minutes a game. Obviously, the selection of the 30-minute mark could pose a few issues. But, while there are starters in the NBA that average less than 30 minutes a game, these starters are not the “superstars” of their team, so they could still be considered role players. With the upper limit of 30 minutes, there were two players most would consider “superstars” that made the list: Tim Duncan and Kevin Garnett. These two superstars are part of a data set of 1127 players, so their inclusion will have almost zero effect on the overall results. Another thing to remember is that this study in no way shows causation, only correlation.
I split the role players out by position, so as to compare point guards to point guards, shooting guards to shooting guards, etc. The categories I measured were true shooting percentage, the Hollinger assist ratio, rebounds per minute, usage rate, three point percentage, and defensive regularized adjusted plus minus (defensive RAPM). I also took z-scores for all of these statistics and totaled them, creating a total z-score stat for each player that is based on all of the aforementioned stats. A z-score tells us how far away from the mean the statistic in question is based on a bell-curve.[4] By converting each stat to a z-score, I am able to use the same scale for all of the stats. Therefore, each stat is given equal weight in the total z-score category, and a higher z-score would represent a “better” player for each statistic. Finally, I conducted individual group t-tests on each of these categories to check for statistical significance.
First, let’s look at the results for role players at point guard. Point guards on playoff teams played significantly better defense (p-value of .0005) and had a significantly lower usage rate (.0267) than their counterparts on non-playoff teams. There was no significant difference found between the two groups for true shooting percentage, assist rate, rebounding rate, or three-point percentage. There also was no significant difference in the total z-score, meaning that when these stats are combined and given equal weight, there is no difference between role playing point guards for playoff and non-playoff teams.[5] The chart below shows this data, and the categories with significant differences are in bold.
Now let’s move on to the role players at shooting guard, quite possibly the weakest position in the NBA. [5] Shooting guards on playoff teams had a significantly higher true shooting percentage (.0002), a significantly higher three point percentage (.0498), and played significantly better defense (.0439). There was no difference in rebounding rate, assist rate, or usage rate[6] between shooting guards on playoff teams and non-playoff teams. The total z-score of a playoff-bound shooting guard was significantly higher (.0092) than that of a non-playoff shooting guard. This finding seems to suggest that role playing shooting guards on playoff teams are better overall than their non-playoff peers. Furthermore, this difference is based on better shooting and defense, two skills commonly sought in role-playing shooting guards.[7]
Small forwards, along with shooting guards, fulfill the role of the “swingman” role player. Role players at small forward on teams that made the playoffs had a significantly higher true shooting percentage (.0007) and three point percentage (.0454). They also had a significantly lower usage rate (.0433) than small forwards on teams that did not make the playoffs. There was no significant difference in assist rate, rebounding rate, or defense RAPM between the two groups. The total z-score was also significantly higher (.0126) for playoff bound small forwards, again suggesting they are better overall than non-playoff small forwards. Perhaps the most interesting result, however, is that there was no significant difference in the level of defense between the two groups of small forwards. Perhaps the defense of small forwards is overemphasized compared to that of shooting guards and point guards.[8]
 Power forward has been a position in flux lately, losing the likes of Tim Duncan and Kevin Garnett to center, while simultaneously adding plenty of undersized forwards such as LeBron James[9] and Carmelo Anthony. Role players at power forward on teams that made the playoffs had a significantly higher true shooting percentage (p-value of .0026) and played significantly better defense (.0408) than their counterparts on teams that did not make the playoffs. There was no significant difference in three-point percentage, assist rate, usage rate, or rebounding rate. There was a significant difference in total z-score (.0213) for power forwards. In summation, role-playing power forwards on playoff teams are significantly better than those on non-playoff teams based on better shooting and better defense.
The center position seems to be a dying breed in the NBA. They even took the position out of the All-Star game,[10] so obviously a role-playing center is not at the top of the list of priorities for NBA teams looking to retool their roster. However, there are still significant differences between the centers on playoff and non-playoff teams. Playoff-bound centers that are role players play significantly better defense (p-value of .000), have a significantly higher true shooting percentage (.0001), and have a significantly lower usage rate (.0463). There is no significant difference for assist rate, three point percentage, or rebounding rate.[11]  The difference in total z-score for centers was significant (.0200), meaning that role playing centers on playoff teams were statistically better overall.
[1] Though Larry Brown and the 2004 Pistons have something to say about that.
[2] Looking at you Shane Battier.
[3] Still looking at you Shane Battier.
[4] A z-score of zero means that the statistic is at the mean, while a z-score of one means that the statistic is one standard deviation from the mean.
[5] Sorry, Derek Fisher.
[6] Mostly because there are only three superstar shooting guards: James Harden, Dwayne Wade, Kobe Bryant. Need further proof it is the weak position of the NBA: Joe Johnson is a six-time All-Star (cut to the five current Hawks fans in existence sadly nodding along).
[7] No one mention this to J.R. Smith or Jamaal Crawford please, for fear they might start taking every single shot.
[8] More evidence for why NBA nerds everywhere cringe when Thabo Sefolosha or Tony Allen miss yet another three.
[9] I will never become accustomed tocalling LeBron “undersized.” It just doesn’t seem right.
[10] It doesn’t help that the best center in the league for the past five years seemingly left his athleticism along with his dignity in Orlando.
[11] Good news for the entire Lopez family! Well mainly for Robin, Brook is clearly past the point of being a role player.
[12] Dikembe Mutombo wags his finger in agreement.
Great article. One thing to keep in mind is that there are 2 different types of role players. On the one hand there is a guy like Avery Bradley who plays with the starters and on the other hand there is a guy like JR Smith who generally plays with the second unit of all role players. When it comes to this distinction in shooting guards there are significant differentiations. For Bradley, it is important for him to be able to play good defense and then hit the open 3, but not necessarily to create for himself. For a guy like JR, it is important for him to create offense. While it is also important for a good defender to be on the second unit (moreso for other positions), it is not important for a gunner who can create his own low percentage shots and does not really do anything else to play with the first team. So sometimes these players only add value depending on the context of the role. Although then again, this is about correlation not causation.
Cheers