Adventures in clutch time

[Nathan]

Hi everyone, I decided to make this thread because I feel that clutch time is quite poorly understood currently, and a small amount of research could potentially help us understand it much better.

Here's a basic topic to start things off:

Good teams play even better in the clutch (last 5 minutes, +/-5 margin), and worse teams play even worse. Based on a linear regression over the last 5 seasons, clutch time magnifies a team's plus/minus per 100 possessions by a factor of about 2.2. That is, a good team that outscores opponents by 10 points per 100 possessions overall could be expected to outscore opponents by 22 points per 100 possessions in clutch time.

This could be due to a number of factors, off the top of my head:

-Good teams are usually on the winning side of things, which helps them a lot in the final 30 seconds because they're shooting free throws while their opponents throw up desperate heaves. My sense is that while this plays a role, it probably does not fully explain the discrepancy. It would be good to filter out intentional foul situations and hopeless halfcourt heaves somehow.

-Refs increasingly favor stars and popular teams in the clutch, consciously or not. My sense, again, is that this is only a small piece of the puzzle, but it would be interesting to look at clutch time foul/free throw disparity.

-Players perform very differently in clutch time. The simplest explanation is that better players, on average, get better, and worse players get worse. But this isn't totally satisfying (why should worse players get worse?). Perhaps all players get better, but the best players improve the most? It would be interesting to see what good box score stats like BPM say about this (e.g. calculate clutch BPM for a few seasons, and compare players' overall BPM to their clutch BPM). To go further, we could attempt to make a clutch BPM stat, which is potentially a very rich topic. While sample sizes are usually too small to make definitive claims about individual players, it would be interesting to see how the value of box score stats like 3 pointers, steals, assists, etc. change in clutch time.

-At a team level offenses and defenses are quite different in clutch time. Maybe good defenses get better in clutch time, and good offenses also get better in clutch time. Or perhaps the "defense wins championships" mantra rings true, and most of the change happens on the defensive end. This is a question that could be answered fairly easily, though it would be good to filter intentional fouls, etc. as mentioned above.

I look forward to hearing your thoughts, and feel free to pose your own clutch-related questions/ideas!

[Mike G]

...(why should worse players get worse?). Perhaps all players get better, but the best players improve the most?

I'll go with this as the default to be disproven. Players can get outplayed even when they're doing their best.

[Kevin Pelton]

Interesting topic! Do you have a graphic for clutch net rating vs. overall net rating?

-Good teams are usually on the winning side of things, which helps them a lot in the final 30 seconds because they're shooting free throws while their opponents throw up desperate heaves. My sense is that while this plays a role, it probably does not fully explain the discrepancy. It would be good to filter out intentional foul situations and hopeless halfcourt heaves somehow.

I'd love to see that too. In the interim, I wonder how much we could tease this out by looking at the discrepancies in free throw rates by win percentage/net rating/measure of likelihood to be winning these games.

The simplest explanation is that better players, on average, get better, and worse players get worse.

One thing we do know is that better players tend to play larger roles offensively late in games. Usage rates in clutch situations go far beyond what we'd ever see over the course of a game. So it stands to reason that shot creation would differentiate more between teams in clutch situations than overall.

[rlee]

Another factor to consider: The best teams tend to have the best coaches. Whatever coaching impact discrepancy there is may be enhanced in "clutch" situations, e.g. strategic decisions, including ATOs.

[colts18]

I did a topic on this nearly 3 years ago

viewtopic.php?f=2&t=8575

Clutch play correlates well between two time periods.

[rlee]

Coincidentally, this was published today:

http://airalamo.com/2016/12/27/spurs-fo ... is-season/

[Nate]

...
This could be due to a number of factors, off the top of my head:

...

It could just be explained by teams going harder during clutch time than on average.

Goldman and Rao discussed clutch time in their Sloan paper...

http://www.sloansportsconference.com/wp ... %20NBA.pdf