Call Timeout? Not So Fast, NBA Coaches
With 11 seconds remaining in Game 6 of the NBA Finals, Kawhi Leonard of the San Antonio Spurs captured a defensive rebound with the Spurs trailing by one point. Rather than take a de rigueur timeout, the Spurs eschewed conventional wisdom and pushed the ball up the floor. It didn’t work. Manu Ginobili took the outlet and easily penetrated the lane in semi-transition, but was then stripped/fouled* by Ray Allen.
*I thought it was a pretty good strip by Allen, as the replays show he had already dislodged the ball before making relatively minimal contact with Ginobili’s arms.
For one of the few times in his career, Gregg Popovich was heavily criticized for failing to call timeout to set up a play for the most important possession of the game. especially with Tony Parker on the bench. But look at the play:
Despite the fact that Dwyane Wade took a predictable isolation jumper after running the shot clock down, the Heat are in relative disarray. Allen goes after the offensive rebound and misses, while Wade for some reason goes for a steal in the backcourt and then trails the play. Given the actions of Wade and Mario Chalmers, it’s possible some of the Heat don’t even know who they are guarding. Ginobili has a nice lane to penetrate and has Danny Green or Tim Duncan open behind him had he chosen to pass. It takes a tremendous strip by Allen to stop him. Ginobili loose in the lane is about the best look any drawn up play could hope to get.
But this must be an aberration; the conventional wisdom to call timeout after a late-game defensive rebound must exist for a reason, right?
It Is Very Difficult to Score Out of a Timeout
One of the key unanswered questions in basketball is how performance varies by context. As Portland Trail Blazer’s coach Terry Stotts has noted*, basketball is perhaps the most context dependent of the major American sports due to the high frequency of scoring and the quickness with which teams change ends.
*Said Stotts: “One thing I think is very difficult for basketball in comparison to football and baseball is those two sports have static events. There’s an at-bat and something happens on that at-bat and you can quantify each at-bat and same thing with football with, even though there’s 11 players on the field, you have a down and yardage each time and it’s a very static event. Basketball is a free-flowing, with different matchups and different set of circumstances, perhaps every time down the floor.”
By utilizing play-by-play data from each game of the 2012-13 NBA season (as well as computer skills far beyond my capabilities), it is possible to determine what effect “prior events” have on a team’s subsequent shooting percentage. The excellent site NBAwowy.com has this data for each individual team. Aggregating all of this data allows one to draw some very interesting conclusions league-wide, based on a sample of over 200,000 possessions throughout the season. At first blush, one conclusion immediately leaped off the spreadsheet before any other: Coaches should call timeout much less often after a change of possession.
Equivalent field goal percentage (eFG%) measures a team’s field goal percentage by accounting for the additional point conferred by a three pointer. League-wide, teams had an eFG% of 50.5% after a miss by the opposing team. After a timeout, that number falls over 5% to 45.3%. In other words, by calling timeout after a defensive rebound a coach reduces the average team’s shooting by the nearly the difference between Oklahoma City’s third-ranked and Chicago’s 29th-ranked overall eFG%. The effect was so extreme that none of the league’s 30 teams managed to shoot better after a timeout than after a miss.*
*Indiana, Washington, Memphis and (ironically) San Antonio came the closest, but aside from these four every other team shot at least 1.9 eFG% worse out of timeouts than off misses. Golden State had by far the largest differential at 11.4 %.
An examination of shooting after other prior events reveals that the average team can expect to score less points coming out of a timeout than in any other situation. One would expect that shots following steals (which often lead to fast breaks) or offensive rebounds (because they tend to be closer to the basket) would be much more efficient, and indeed teams shot 61.5 eFG% off steals and 53.6 eFG% off the offensive glass compared to the league average eFG% of 49.6. While these results may have been expected, timeouts proved inefficient even in comparison to other situations where the offensive team has to take the ball out of bounds. Off made baskets and free throws, (47%), non-shooting fouls (48.3%) or dead ball turnovers (46.9%), shooting eclipsed that after a timeout by a minimum of 1.6 eFG%.
Although it defies conventional late-game wisdom, better shooting off makes compared to timeouts perhaps can be explained by the fact that a make allows for a few transition opportunities and the chance to attack full court. But what about compared to other true dead ball situations? Do opposing coaches’ defensive substitutions in timeouts explain the difference? Does the break afforded by a timeout allow tired players to recover and exert maximum effort on the following defensive possession? Or, are coaches simply better at coaching defense than offense in timeouts? Whatever the reason, this is powerful evidence that timeouts confer an advantage on the defense, although caveats abound (more on that later).
If it is true that calling timeouts hurts the offense, how should coaches adapt? Most importantly, calling timeout immediately after a defensive rebound should generally be avoided except in special circumstances.* In fact, given the apparent negative effects of timeouts on offense, I posit that teams should try to save their timeouts for when they have practical uses, such as saving a possession that is about to turn into a jump ball or advancing the ball to half court late in the game when trailing.** Or, coaches could save them for the last few seconds of possessions that have gone completely awry out near halfcourt, allowing a reset and a chance at a decent shot.
Failing that, taking a timeout in a dead ball situation where the other team has the ball is optimal, especially when the team will soon have to use one of its “automatic” TV timeouts.
*These might include situations such as advancing the ball into the frontcourt when leading in the last few seconds of the game to avoid a potential turnover under under your own basket.
**Part of what made Popovich’s decision not to call timeout in Game 6 so good was that he saved his last timeout in case the Spurs were stopped. The Heat would make two free throws after the Ginobili turnover, but the Spurs were able to at least advance the ball into the frontcourt to have a reasonable shot at a tying Danny Green three, although it was ultimately blocked by Chris Bosh.
The Weaknesses of the Study
A few potential quibbles regarding these conclusions immediately come to mind. The first, as many have probably realized by now, is using eFG% to measure offensive effectiveness rather than points per possession or at least True Shooting Percentage. The issue with the available data is that a “possession” is delineated by when the ball changes hands via a made shot, missed shot and defensive rebound, or turnover. So, multiple “events” such as a non-shooting foul or an offensive rebound can occur on every possession. A related issue is that eFG% only measures what happens on possessions that end in a field goal attempt, rather than a turnover or free throw attempt, so some of the highest- and lowest-value plays that can make or break an offense are not captured in the analysis. Still, field goal shooting does explain a high percentage of the variance in offense, and no plausible explanation comes to mind for why free throw attempts or turnovers would be particularly higher or lower, respectively, out of timeouts to counteract the clearly lower eFG%. While these uncaptured effects are certainly worrisome, the differences (particularly between after misses and timeouts) and large sample size give a fair degree of confidence that there is a real negative effect on offense from calling timeouts.
Another potential landmine is the hypothesis that a number of timeouts are taken in the fourth quarter or overtime when a trailing team is in desperation mode, which could serve to reduce post-timeout performance. I have not had a chance to run the full numbers yet, but a cursory check of a few teams’ post-timeout performance in the first three quarters (when desperation mode would not apply) did not appear to show a noticeable difference from the overall numbers.
But despite these issues, the fact that shooting so much worse coming out of timeouts than in any other situation indicates that coaches would be better served letting go of the reins and letting their players play. This is not to say that these teams would not still benefit from coaches’ input on late-game strategy. On the contrary, abandoning late-game timeouts would put a premium on the more difficult task of training players to follow the optimum strategy in crunch time on the fly.
A somewhat expected result was the fact that the league-wide eFG% was higher off misses than makes, and higher still off offensive rebounds and steals. This has obvious implications for measuring the effectiveness of offenses and defenses. An offense like the Warriors, who shot after opponents’ misses a league-high 30.14% of the time, will have an easier time of it than Sacramento, whose shooters had the benefit of a preceding miss only 25.45% of the time.
In a similar vein, the Clippers shot a league-leading 8.15% of their shots off steals, converting those shots at an amazing 68.7 eFG%, second behind Denver’s 69.2 eFG% and 2% better than third place Oklahoma City. By comparison, the average team took 6.82% of its shots off steals. The Clippers’ amazing performance off steals buoyed their offense, perhaps indicating that their halfcourt troubles under Vinny Del Negro’s much-derided offensive system were more severe than their fourth place offensive rating might indicate.
Brooklyn, Oklahoma City, Golden State and Dallas were the only teams to shoot worse off their own offensive rebounds than they did off defensive rebounds.
- Dallas led the league with a whopping 6.65% of their shots taken out of timeouts, well above the league average of 5.02%. Aside from Dallas, the top 10 teams in shooting out of timeouts were all below .500–except for Miami and Indiana.