Bruin Sports Analytics
Shooting For the Stars: Increasing Usage Rates in the NBA
By: Nadeev Alam and Nathan Wetmore
There is no doubt that today’s NBA is a stars’ league. Star players have unprecedented control over all levels of their basketball team. Today, the league’s superstars often give significant input on their team’s roster and coaching staff decisions. NBA fans commonly refer to LeBron James as “LeGeneral Manager,” citing his apparent and aggressive pushes for his teams to make high-profile trades and acquisitions. The Brooklyn Nets’ last four seasons are another prime example. Two-time Finals MVP Kevin Durant and three-time All-NBA Kyrie Irving put the league on notice before the 2019-2020 season when they both signed with the Nets. Just the following season, the team added another MVP in James Harden. There was a new dynasty, basketball was ruined, and Durant had his second superteam. Or at least, that was the expectation. Barely a year (and a second round exit) later, Harden forced his way to the Philadelphia 76ers. Another year later, Durant and Irving left for the Phoenix Suns and Dallas Mavericks, respectively. By leveraging their basketball talent, today’s NBA superstars drive league-altering decisions.
However, it is not just that stars have more say. In addition, there are more stars in today’s NBA. Historically overlooked small-market teams, such as the Memphis Grizzlies, Milwaukee Bucks, Minnesota Timberwolves, and Portland Trailblazers, all have a big-name player. At least, as a fan, it may feel like every team has a star. But maybe social media has just allowed us to become familiar with more players than fans of the past. Today, every highlight is immediately uploaded for the world to see. Before social media, watching games live was often the only way to become familiar with the league’s talent. So, what do the numbers say? Are there more stars in today’s NBA?
To quantify a “star,” we look at usage percentage. Usage percentage is an estimate of the percentage of team plays used by a player while he was on the floor. In other words, usage percentage is the portion of plays during which a player had an opportunity to score. We calculate the statistic using the following formula. Parts of the formula preceded by “Tm” indicate a team stat, while the rest refer to the individual player.
This equation can be understandably hard to look at and understand. However, it is not essential to understand the exact calculation as long as we can use the statistic to measure a “star player’s” impact relative to his teammates. The league average usage percentage is always 20%, as five players are on the floor for a team at any given time. We will use this statistic to analyze the change in the number of stars and the impact these players have on winning championships.
Star(s) Are Born
We have proposed that the NBA is more star-studded than in the past. But what do the numbers say? We took the top ten players in usage percentage for each season, first for the regular season:
There is a clear trend from the 1970s to today. The league’s stars have been utilized increasingly over time—the significance of the jump becomes apparent when examining the league leaders in the statistic. In the 1967-68 season, the first one where the NBA recorded the metric, Mel Daniels led the NBA with a usage percentage of only 26.08%. Russell Westbrook and James Harden led the way in recent years, with usage percentages over 40% in 2017 and 2019, respectively. And currently, Giannis Antetokounmpo has a usage percentage of 39.06% so far this season. So, NBA offenses are more centered around their best player. But this is not enough. A few high-usage players on poor teams could easily skew this trend. For example, Bradley Beal had a top-ten usage percentage in two of the previous three seasons. However, the Washington Wizards only won one playoff game between these two years. And if the trend does not matter for winning teams: who cares? So, we looked at the season-to-season leaders' usage percentages in the playoffs.
The trend continues in the playoffs. Star players have carried their teams at higher rates since the 1990s, even in the playoffs. We have established that there is an evolving roster construction model where the dominant player takes nearly double the average starter's role. Over time, we have seen the model become more popular and extreme. But does the model win championships?
So, let us examine the best player on the best team. The following shows the playoff usage percentage of Finals MVPs over the years.
While there may appear to be an upward trend since the late seventies, there is too much variance to generalize the relationship between a team’s best player’s usage percentage and winning a championship. Michael Jordan is in his own class, putting up a usage percentage over 35% during four of his championship runs. Meanwhile, some teams’ MVPs had a usage rate below 20%–the league average.
So, we still need to determine if a single-player-centric team is the new winning model. To better understand this model, we must examine the distribution of usage rates within each championship team. Looking at the skews on the following plot should show some trends among championship teams.
The boxes (in blue) contain the middle 50% of the usage percentages for that team. The horizontal line (in red) in the box represents the median. The whiskers extending from the box show the range of the data that is not considered an outlier. If the box’s upper whisker is longer, more players on that team had higher usage rates. If the lower whisker is longer, more players had lower usage rates.
Outliers are displayed as individual points on the plot. Outliers can indicate unusual or exceptional performance by a particular player. Outliers significantly higher than the rest of the data may indicate players who were critical to their team's success and greatly impacted the outcome of the championship. We can analyze these outliers to determine the factors that led to their success.
However, it is difficult to pin a trend in the plots. Moving from left to right, it is unclear if the box plots are widening or compressing one way or another. We can draw no accurate conclusion about the rise of the star-centric model in the NBA just by viewing the distribution of the champion’s usage percentages. This should have been expected. After all, countless factors influence an NBA team’s path to the finals. For example, coaching, team chemistry, and health are also critical parts of a team’s playoff run.
So, we must alter the approach. Instead of viewing every championship team, we only viewed teams that utilized the star-centric model but expanded the search to the entire playoffs.
This chart takes the team with the highest used player in the playoffs–for example, Michael Jordan’s 1993 Bulls–and plots their playoff wins against their leader’s usage percentage.
It is important to note the axes on this chart. The graph is time-independent, and the horizontal axis starts at 30%, which is still significantly high. The players represented by the dots on the left side of this graph do not have low usage rates; there were just a few players with higher ones. From this graph, what we were beginning to think is all but solidified. There is little evidence to support having a star with a high usage rate leads to an increase in playoff wins or odds of hanging a banner.
None Quite Like Mike
We cannot continue discussing stars and usage percent without addressing His Airness, Michael Jordan. Each graph shows a spike in outlandish usage percentages surrounding the Chicago Bulls’ first three-peat. Jordan is undoubtedly one of the greatest basketball players ever, and his impact on the NBA cannot be overstated. As the focal point of the Bulls' offense, he revolutionized the way teams approached basketball, and this approach soon spread throughout the league. Jordan's scoring prowess and unmatched skillset allowed him to dominate games, and his success helped pave the way for future generations of superstars to emerge.
Before Jordan, NBA teams tended to be more balanced, with several players contributing to the offense. However, after Jordan's success with the Bulls, teams realized that having a dominant star player might be the key to winning championships. This success led to more teams building around one or two superstar players, with the rest of the roster designed to complement their strengths. This trend has continued to this day, with many playoff teams having one or two players with usage percentages well above the league average.
However, deeper analysis shows that there may not be any correlation between any particular style of roster construction and winning championships. Michael Jordan is one of only three players to lead the league in playoff usage percentage and win the championship in the same year, and the only one to do it multiple (four) times. The other two were Gus Williams in 1979 with the Seattle Supersonics and Andrew Toney in 1983 with the Philadelphia 76ers; however, neither won Finals MVP. MJ is indeed one of one.
Other outliers appear in our charts that further prove the lack of a correlation between roster construction and playoff wins. For example, in the bottom left corner of the previous graph, you can see Russell Westbrook with a usage rate of a staggering 46.98%, the highest ever in the playoffs. Yet, the Oklahoma City Thunder only won one playoff game and were eliminated in the first round of the 2017 playoffs.
We found more extremities when examining Finals MVPs. In 1978, Wesley Unseld won the finals MVP with an abysmal 10.9% usage percentage, barely more than half the usage percentage of an average player. Commonly referred to as the worst Finals MVP ever, Unseld was outscored by 2 Washington Bullets teammates by over 10 points per game and was still awarded MVP. In the modern era, Kawhi Leonard won Finals MVP in 2014 with a usage rate of 18.9%, and Andre Iguodala won the award the following year with a percentage of 15.1%. Leonard’s 2014 San Antonio Spurs were commonly called a “starless” team and praised for their cohesive teamwork. In 2015, many believed Stephen Curry should have been Finals MVP, as Iguodola mainly won this award for his defensive efforts. These examples, especially those from the last decade, lead us to a conclusion. A single star is not necessary to win a championship, and it is not even a guarantee to make it past the first round of the playoffs.
Today’s NBA is undoubtedly run by its stars. The past five decades of usage percentage data show that basketball rosters are increasingly built around one or two players. However, it is unclear if this model is a winning one. Instead, we have seen teams win with one central player and teams with many contributing (but no dominating) players win the NBA finals. Currently, it looks like the trend will continue as every team engages in the arms race for star players. However, it is impossible to make championship predictions based on roster construction alone. Simply put, there is more to basketball than one player.