Calling timeouts in the NBA
How much of a season disappears into timeouts? I dug into 2015-16 play-by-play to map when coaches stop the game, who stops it most, and whether a timeout actually halts a run.
How much of the 2015-16 season disappeared into timeouts? I counted them from play-by-play data.
Across the regular season (1,230 games) there were 16,508 timeouts. Setting aside the league's automatic TV timeouts, teams called slightly more on the road (7,210) than at home (7,041).
How much does each team stop the game?
Teams vary a lot. Popovich's Spurs called a league-low 391; Brett Brown's rebuilding 76ers called the most, 608. Timeout usage is mostly a matter of coaching personality.
When do timeouts happen?
For the first three quarters the pattern is strikingly regular: two clean peaks a quarter, lining up with the league's mandatory TV timeouts. Those aren't charged to a team, which is why the league-wide Total line spikes far above the Home and Away traces. The fourth quarter breaks the pattern: each team is limited to two timeouts after the three-minute mark, so coaches spend early, and the end of a close game is timeout galore.
What's the score when coaches intervene?
The distribution leans negative. The median (−2, red bar) means half of all timeouts came when the calling team was losing by more than two. But coaches also call them when losing advantage: a +20 lead cut to +10 still reads as +10 here, so that cause stays hidden. To see it, I tracked the score in a tight window around each timeout.
Do timeouts actually stop a run?
Below is the calling team's net point swing in the 90 seconds around a timeout, zeroed at the start of the window, so a line sliding down means the team is being outscored. Use the dropdown to swap the league average for any single team.
In the 90 seconds before a timeout the calling team bleeds about two points of margin. The moment the whistle blows the slide stops, flattens, and ticks back up. Stop-the-bleeding timeouts are real.
Which teams let it ride?
How far does a team let a run go before calling a timeout? The Nuggets had the slowest trigger, letting a run reach nearly three points; the Magic were the twitchiest, reaching for one after barely a point.
...and whose timeouts actually work?
Once it's called, does the run actually stop? This compares the swing in the 90 seconds after a timeout to the moment it was called. Almost everyone lands positive, so a timeout really does tend to stop the bleeding. The Warriors were best at +0.9 points; the Suns were the only team that, on average, kept sliding. Good teams don't just call timeouts, they answer them.