Usage When an NBA Star Sits

The news breaks an hour before tip: the team's best scorer is out. Within minutes the lines on his teammates move, and the obvious question gets asked everywhere at once. Who gets his shots? The honest answer is less satisfying than the one most people give, because a star's absence frees up two different resources, and they rarely go to the same place.

The NBA research framework already makes the point that usage does not spread evenly. This piece works through the mechanics: what the usage number actually measures, how to trace the minutes and the shots separately, and why the with and without data that looks like a direct answer is usually a very small sample dressed up as one.

What usage rate actually measures

The NBA.com stat glossary defines usage percentage as the percentage of team plays used by a player while he is on the floor. The Basketball Reference glossary shows how that is estimated: field goal attempts, plus 0.44 times free throw attempts, plus turnovers, compared with the same total for the team during the minutes the player was on the court. The free throw weight is there because a trip to the line does not map one to one onto a possession.

Three consequences follow from that definition, and each one matters when a star sits.

• Usage is a share, not a count. Five players on the floor divide one team's plays between them, so a star's share cannot be removed without the others' shares rising. The question is only whose.
• Usage is measured per minute on the floor. A bench scorer with a high usage rate in 14 minutes has shown a willingness to shoot, not an ability to carry that share across 30 minutes against starters.
• Usage counts plays that end with a shot, a trip to the line, or a turnover. It does not count passes that create a shot for someone else. A primary playmaker's absence removes creation that never appeared in his own usage number.

That last point is the one people miss. When a ball dominant guard sits, his teammates lose not only a player who took shots but the player who set up many of theirs. Some of them get more attempts and worse ones. A spot up shooter can see his volume hold steady while his looks get harder, because nobody is bending the defense for him anymore.

Minutes and shots flow to different players

Trace the absence as two separate flows. The first is minutes. The star played, say, 34 minutes, and those minutes go mostly to whoever backs up his position, plus some extra run for the starters around him. That flow is governed by the depth chart and the coach's lineup preferences, and it is usually the easier of the two to estimate.

The second flow is plays. The star used a large share of the offense while he was on the floor, and those plays go to whoever can create a shot in the lineups that remain. That flow is governed by skill, not position. The backup who inherits the minutes is often a low usage player, and the plays skip past him to a secondary creator who was already starting.

Not all of the star's plays get redistributed at all. Teams without their best creator often play a little differently: more possessions end in late clock shots, more end in turnovers, and some coaches slow the game down. Part of the vacated usage becomes worse possessions rather than extra shots for anyone in particular. Assume the full share lands on teammates and you will overstate almost everyone.

Different stats redistribute by different rules

Scoring follows creation. Assists follow whoever now brings the ball up and runs the actions, which can be a different player from the one who inherits the shots. Rebounds follow floor position and size, so a missing big man's rebounds go mostly to the players who occupy the same space, and a missing guard's absence barely touches the rebounding picture at all. Threes follow shooters who get more attempts, but their efficiency often suffers when the player who drew help defense is gone.

So research the absence stat by stat. A single rule, such as spreading the missing points by minutes, will produce errors that point in different directions for different markets on the same player. The guard who gains shot volume may gain little in assists. The big man who gains minutes may gain rebounds and very few shots.

On/off and with or without data are noisy

The tempting shortcut is to look up how teammates have performed in games the star missed, or in minutes when he sat on the bench. Both are worth looking at. Neither is as clean as it seems.

On/off numbers describe the team with the player on the court versus off it, and the NBA.com glossary defines net rating at the player level as the team's point differential per 100 possessions while he is on court. The trouble is who else is out there. A star's off court minutes tend to come against opposing benches, inside stretches the coach chose, and in the tail ends of blowouts. His teammates' numbers in those minutes describe a different game from one where he is absent for all 48.

Games he missed entirely are closer to the right question, but the sample is usually tiny. Four or five games is common. Those games came against specific opponents, on specific rest schedules, sometimes with other players also out. Sample size in player prop analysis covers why a handful of observations cannot separate a real role from ordinary variation, and here the problem is worse, because the games were not chosen at random. A player who scored heavily in four games without the star may have met four weak defenses.

Extreme results in small samples tend to be followed by less extreme ones. That is regression toward the mean, and it applies with full force to the teammate who went off the last time the star sat. Read the with or without log for opportunity, not for results.

What to pull out of those games

1. Minutes for each rotation player, compared with their minutes when the star plays. This is the most stable signal in the log.
2. Field goal attempts plus free throw attempts per minute, a rough usage proxy you can compute from any box score. It tells you who took on shots, independent of whether they fell.
3. Who started and who closed. Closing lineups tell you who the coach trusts when it matters.
4. Anything else that was different: other absences, back to backs, the opponent's quality. If two starters were out, the log answers a different question from tonight's.

Timing, confirmation, and what the line already knows

Absence news reaches the market fast, and the most obvious beneficiary is the first line to move. By the time a star is confirmed out, the flow of minutes to his direct backup is usually reflected in the numbers. The flows that are slower to be priced are the second order ones: the secondary creator's shot volume, the rebounding change when a big sits, or the drop in efficiency for a shooter who loses his best passer.

Confirm the status before doing any of this work, and confirm it again close to tip. Reading an injury report explains why a designation is a prompt to investigate rather than an answer, and why the effect on teammates is often more researchable than the effect on the injured player. Also check whether the absence changes who starts. A starting assignment can shift minutes and closing roles in ways a bench role never does, and role changes and props covers how to tell a durable change from a one night experiment.

Finally, keep the minutes estimate honest. The minutes and blowouts piece explains how a team's 240 player minutes act as a fixed budget. When a star sits, redo that budget from scratch. It will often show that the obvious beneficiary cannot take all the minutes people are assuming, because someone else also needs to play them.

What this research cannot tell you

Usage redistribution is one of the most researchable situations in the sport, and it is still an estimate built on judgment. Coaches experiment when a star is out. A player who has never carried a large share for a full game may take it on, or may defer, and the evidence that would tell you which is precisely what does not exist yet. The first game without a star is a real unknown, and a range that reflects that is more honest than a confident number.

The working order is short. Confirm the absence. Rebuild the minutes budget. Trace the plays to the players who can create, not the ones who inherit the position. Adjust each stat by its own rule. Treat the with or without log as evidence about opportunity, and leave its results where they belong, in the small sample column.