Late Game Timeouts Part II: What Happens When Trailing

By John Ezekowitz

Two weeks ago, I published the first part of my late game timeout study from the 2009-2010 season. I found that when the score was tied, not calling a timeout was more effective than calling a timeout. In this post, I’m extending the same analysis to situations in which the team with the ball is trailing by one or two points. My results follow after the jump.

In my dataset, 296 teams trailed by one point and held the ball at the end of a period (second half or overtime). Of those, 155 teams called timeout, and 141 teams did not. Teams that called a timeout scored an average of 1.17 points per possession. Teams that did not call timeout scored an average of 1.11 points per possession. Using a two-sample Wilcoxon ranksum test, I tested the hypothesis that mean points per possession were the same whether a timeout was called or not. This hypothesis was confirmed (p=-0.565), indicating that there was no significant difference between calling a timeout or not calling a timeout. A further Chi^2 analysis of simply whether or not the team scored indicated the same finding. Thus when a team was down one point last year, there is no significant advantage to either calling or not calling a timeout.

I found similar results for teams that were down two points. Of these, 189 teams did not call timeout, while 165 teams did. The teams that called timeout averaged 1.05 points per possession while those that did not averaged 1.18 PPP. While this may seem like a sizable difference, the same Wilcoxon ranksum test showed that there was no significant difference between calling or not calling a timeout (p=0.113). Another Chi^2 analysis of simply scoring confirmed the no significant difference result. Thus for teams trailing by both one and two points, there is no advantage in the data between calling or not calling a timeout.

These results are very interesting, especially considering the finding that not calling a timeout was better when the score was tied. In the previous post on tied games, I posited that teams with the ball and the score tied held an advantage of asymmetric information over their opponents that was negated when a timeout was called. It may be that when a team is trailing, they do not hold such an advantage.

Some commenters posited (correctly, I think) that not calling a timeout left the defense in a state of disarray, allowing the offense to push the ball up the court for an easier shot attempt. If that were always the case, it would make sense that not calling a timeout would be a better option than calling a timeout in these two situations, as well. The data does not seem to support this, however. This could possibly be explained by arguing that teams start possessions at the end of games in a different way when they are tied compared to when they are trailing. The benefit from pushing the ball would seem to be greatest off of a missed shot. While this may warrant further study, I have a hard time believing that this would explain the different result.

Ultimately, what this analysis shows is that the decision to call a timeout when down one or two points is heavily based on situational factors unique to each game. A coach must consider the make-up of his team, the make-up of his opponent, the flow of the game, and his team’s ability to execute set plays (ones drawn up in timeouts). As opposed to the tied game situation, where the data clearly favored the no timeout option, these two situations demand a nuanced decision that is apparently not influenced by an overarching advantage one way or the other.

I hope you will come back for the final part of this series analyzing timeouts when a team is down three points. Again, I think the results are surprising and informative.

About the author


View all posts

1 Comment

  • If you have a team with a high basketball IQ then you don’t need to call timeout. Our team practices these situations whereas we won’t need a timeout. Depending on personnel on the floor etc… we may call one to change the lineup. Studying this has too many variables and probably not going to get accurate data to make a solid conclusion. This is why you’ll find that the two scenarios are always close and able to go either way. – Hope this helps with your study.
    PS. If you know Brianne Holland-Stergar let her know her uncle Jim says hello. She is does track there at Harvard

Leave a Reply

Your email address will not be published. Required fields are marked *