In the NBA, usage rate is a statistic meant to estimate how often a player contributes to the outcome of a play. In a sport with only one ball but five players on the court, it is inevitable that some players will be used more than others. A player’s usage rating is meant to measure this. The goal is to get an idea of how much importance an individual player has on their team’s on court performance.

Usage rating is not unlike target share or snaps in the NFL. Sometimes just being on the court or on the field doesn’t tell you how much a player contributes to the outcome of specific plays. In this article, we’ll talk at length about usage rate. Why should we care about it, how it is calculated, and how it can add valuable perspective to players’ performances.

Understanding usage rate is a good introduction to advanced sports analytics. Going beyond the basic box score to contextualize statistics is a key tool in analyzing the games we love. I’ve found this book to be a great example of how to incorporate sports analytics into your own work be it coaching, playing, or analytics. So, let’s dig in.

**Why do we care about Usage Rating?**

Usage rate isn’t a stat that will tell you anything about how good or how bad a player is. Rather, the true value of this stat is that it provides valuable *context* with which to interpret performance. Players with higher usage rates have more opportunities to impact a game.

Let’s compare two players. Steph Curry and Mikal Bridges had essentially the same three point shooting percentage in the 2020-2021 season. Does this mean they are the same quality of three point shooters? Our instinct, past performance, and the media all tell us that no, Steph Curry is the best three point shooter in the NBA.

Looking at Steph Curry’s usage rate can help us contextualize these statistics and let us see that Steph is actually the better three point shooter. Steph had over twice as many attempts from 3 as Mikal Bridges did. Because usage percentage essentially measures how often a player contributes to the outcome of a play, three point attempts is a component of it. So, Steph Curry will have a higher usage rate than Mikal Bridges.

Having a higher usage rate means that often you have to shoot in *suboptimal* circ*mstances rather than picking from the cream of the crop. Steph Curry is going to have to take – on average – harder shots from three than Bridges will. So, while they shot the same percentage from three point range, Steph Curry’s should be more impressive because his shots were harder. Thus, combining usage rate with three point percentage can help us properly contextualize three point percentage. This idea will be explored more in the section talking about efficiency.

A second fun way to use usage rating is when ‘constructing hypothetical teams’. In 2019-2020, James Harden’s usage rate was 36%. In 2018-2019, Durant and Kyrie had a usage rate about 30% each. So, if I told you each of these three guys were going to be on a team together, you should suspect that one or more of these guys would have to change their play style. If everything stayed the same, these guys would account for almost everything that happened to the nets. Using usage percentage is a statistical incarnation of the old phrase *there is only one ball* that people like to use to discredit super teams.

Usage rate can show up in many other scenarios like this where it provides valuable insight into just *how* a team plays. In the next section, we’ll look at how usage rating is calculated. It turns out, with a few observations, that the formula is extremely simple.

**How is Usage Rating Calculated?**

The official formula for usage rate is towards the end of this section. However, there is a *much* easier interpretation of usage rate that I will present first. First, a player is considered *used* in a play if their action ends the possession. There are three ways this can happen:

- They shoot,
- They get fouled and head to the free throw line, or
- They turn the ball over.

With this in mind, usage rate is given by the formula:

\text{Usage Rate} = \frac{\text{Possessions Ended by Player}}{Total Team Possessions with Player on Court} \times 100\%

That is really it. Usage rate of a player is the percentage of their on-floor possessions in which they dictated how the possession ended. If I take most of my teams shots or draw a lot of fouls, my usage rate will reflect this and be relatively large. If I am a role player and mostly set picks, play good defense, and sit in the corner waiting for open shots, my usage percentage will be lower.

There is some difficulty in actually measuring possessions, though. If you remember reading my previous article about pace and efficiency, actually measuring what constitutes a possession is fairly difficult. The NBA computes possessions from the box score by adding shots, turnovers, and free throw line trips (estimated by 0.44*FTA) and subtracts off offensive rebounds. The 0.44 times free throw attempts is meant to average out free throw line trips where one attempt is taken, where two are taken, and where three are taken to try to estimate total number of free throw line trips.

Thus, the actual formula for usage rate is:

100\% \times \frac{FGA+0.44\cdot FTA + TOV}{Team FTA + 0.44\cdot Team FTA + Team TOV} \div \text{\% of Game Played}

where FGA, FTA, and TOV are the player’s field goal attempts, free throw attempts, and turnovers while the denominator is the team totals for each of those statistics.

Now, there is one subtlety here. The statistic as it is computed does not really compute the percentage of possessions ended by a specific players’ actions while they are on the court. The numerator is correct; it actually measures the number of possessions that a specific player ended.

However the term

\frac{1}{Team FTA + 0.44\cdot Team FTA + Team TOV} \div \text{\% of Game Played}

estimates the number of possessions that occurred while that player was on the floor by taking the total number of possessions and multiplying by the percentage of the game the player was on the floor. This only accurately measures number of possessions a player was on the floor for if the pace of play is constant throughout the game. We know that isn’t true, but it is a good approximation. More on that later.

For example, if a team had 100 possessions in a game and a player played 20 of 48 minutes, then, this formula would estimate that he was on the floor for 41 possessions (100*20/48). In reality, a player may have been on the floor for more or less than these 41 possessions due to natural fluctuations in the pace of the game. Curiously, because of this fact, it is possible for a player to have over a 100% usage rate if he had an extremely high usage during a particularly fast paced portion of the game.

## Usage Rate and Efficiency

As I said before, usage rate is helpful to contextualize other statistics. My favorite such application is related to player efficiency. Players with lower usage rates get to be pickier with their shot selection. Players with lower usage ratings get to in general take an easier set of shots than players with higher usage ratings. There is a fairly strong observed negative relationship between usage rate and efficiency.

This is perhaps contradictory to what one might expect otherwise. If you didn’t know better, you might guess that players with higher usage rates are better players who, therefore, are more efficient. But, we actually see the exact opposite effect simply because higher usage rates means being used in more difficult scenarios. A player with high efficiency and high usage rate is quite rare indeed.

**NBA.COM USG% is NOT Usage Rate**

Interestingly, the way the NBA computes usage percentage is* not* by the formula I presented above. I first noticed this after sorting the players so far this season by USG%. This is what I found:

I thought it was strange that Vernon Carey Jr. (who?) had a usage rating of 100%. In fact, I did some digging and at the time of writing this – November 11th 2021 – Vernon Carey had only made one appearance. He entered the Nets Hornets game with 47 seconds remaining. The Hornets then had one offensive possession and Vernon Carey Jr. shot the ball on that possession. Then the game ended.

While Vernon Carey Jr. was on the court, his team had one possession and he was responsible for the end of that possession. Therefore, his usage rate should be 100% because his action ended 100% of the possessions that he was on the floor for. If you calculate his usage rate using the formula we included above, you get a usage rate of 41.8%. This is the number reported on basketball reference.

The reason for this discrepancy is that the NBA stats actually count possessions a player is on the court for rather than estimating this number by multiplying total possession by the proportion of a game a player is on the court for. The NBA stat is more accurate.

**Conclusions**

Usage rate is a valuable statistic, but certainly some improvements could be made. Most notably to me, is that there are other ways a player can contribute to the game other than those factors going into this version of usage rate. Players with high assist rates have their fingerprints all over the game. However, this will not be appreciated in the statistic in its current form.

As is, though, usage rate is a fairly valuable statistic that can contribute a lot of perspective when it comes to understanding other statistics.

If you liked this article explaining existing advanced metrics, see some of our other articles tackling similar subjects:

- RE24 and the Run Expectancy Matrix
- Real Plus Minus

To subscribe to email notifications when new posts are available, pleas use the form below.