by Ken Pomeroy on Wednesday, December 19, 2012
Sometime in the last week or two, and I seriously don’t remember when, I was watching a broadcast and it was mentioned that one of the participating teams tracks how well it shoots free throws in the last five minutes. I don’t think I heard an explanation of why they were doing it. It could have been for something as innocuous as a blurb in the media notes or something like that. But if it was for anything more serious then it was a waste of time. To illustrate that, let’s consider the three possible things that would be learned from such efforts.
Scenario 1: Your team shoots better in clutch situations than in other situations. What do you do with this information? I guess you can tell your team, “Hey, guys, we really rise to the occasion when the game’s on the line”. There might be some utility in positive reinforcement.
Scenario 2: Your team shoots the same in clutch situations as in other situations. I suppose the reaction here would depend on your expectations. Do you think your team should be better than normal with the game on the line? If so, the response could be like the one outlined in Scenario 3.
Scenario 3: Your team shoots worse in clutch situations that other situations. Hmmm, maybe you could tell your team to concentrate harder. This is the most common prescription of media types. You know, when a player misses a free throw or two late in a game, and we get the “you have to make your free throws in this situation” and “it’s all about concentration”. As if the player at the line is unaware that it would have been better to make the free throws than miss them. So you tell your team to concentrate and then check back in two weeks and the results are no different, then what do you do? Tell them to concentrate harder?
As you might imagine, I’m highly skeptical there is any reason to think that data from the last five minutes has any instructional value over a team or player’s overall ability to make shots. I would just offer this advice to those who do – track stats for every five-minute segment during a game and you’ll get an idea of how much free-throw percentage (or any stat) can change over time for seemingly no reason at all. In this post, I’ll look at this idea on the player level. But there’s something to be learned from team level data as well, and I plan to get to that in a separate post.
For instance, here’s Tony Wroten’s breakdown from last season for each five-minute period of the game. (Note: To make each player’s sample big enough, I’m using data from the most-frequent free-throw shooters from last season.) Wroten was not a very good shooter, making 58.3% of his attempts overall.
1st H FTM-FTA FT% 15-20 14-28 .500 10-15 14-29 .483 5-10 15-29 .517 0-5 21-33 .636 2nd H 15-20 15-25 .600 10-15 23-30 .767 5-10 25-39 .641 0-5 27-51 .529 cl0-5 8-15 .533
[cl0-5 includes cases in the last five minutes only where the score was within 4 points.]
Wroten was not very good late whether the game was close or not. He also was not good early in the game when there was less on the line. Should Lorenzo Romar have told him to harness that feeling he has between the 10 and 15 minute mark? I guess you could concoct a theory where Wroten has a pressure threshold. He’s good when the game is sort-of on the line, but when it’s really crunch time, he cracks. But he also has a problem on the other end when there’s not much on the line. So fix that, too. It really starts to sound silly when you view it in this way.
Let’s look at another high-volume free-throw shooter, but one who was really good – C.J. McCollum. He made 81.1% of his attempts last season.
1st H FTM-FTA FT% 15-20 8-11 .727 10-15 15-19 .789 5-10 18-23 .783 0-5 35-41 .854 2nd H 15-20 18-22 .818 10-15 25-31 .806 5-10 16-21 .762 0-5 61-73 .836 cl0-5 18-19 .947
McCollum will max out all of the character categories come draft time so it’s good to see he rises to the occasion in close-and-late situations. But why is he not so good with five to ten minutes left, yet really good in the last five minutes of the first half?
Here’s Jared Cunningham (73.7%), who really ran hot-and-cold depending on the time left.
1st H FTM-FTA FT% 15-20 6-7 .857 10-15 14-19 .737 5-10 18-20 .900 0-5 16-23 .696 2nd H 15-20 9-17 .529 10-15 38-48 .792 5-10 22-37 .595 0-5 70-92 .761 cl0-5 30-45 .750
Cunningham finished strong, although that was following a pretty bad effort with five-to-ten minutes to go. But he was concentrating pretty hard with between five and ten minutes left in the first half. If only Jared had just kept doing that the rest of the game, he might have been a really good shooter overall.
You’re getting sleepy, so I’ll end with Nate Wolters. Wolters went to the line 240 times and made an admirable 78.3% of his attempts.
1st H FTM-FTA FT% 15-20 7-9 .778 10-15 15-21 .714 5-10 17-18 .944 0-5 24-31 .774 2nd H 15-20 16-18 .889 10-15 35-42 .833 5-10 41-46 .891 0-5 45-66 .682 cl0-5 8-17 .471
Whoa, Nate, what’s up? You appear to be mentally weak, unable to make free throws with the game on the line, and probably unable to carry out basic day-to-day tasks as well. Then again, what explains the fact that Wolters is so bad in the last five minutes but nearly automatic in the previous five minutes? He could carry the gene for Wroten Pressure Threshold Disease, or it could be that clutchness is crap. I vote for the latter*.
The inspiration for this piece was the intro to this post by Phil Birnbaum (h/t Brian Burke), specifically his point about an 80% free-throw shooter needing to be lucky to make a free throw. Indeed, it’s true. It’s also true that a 90% (or even 99%) free-thrower needs luck, too. Just less luck than the 80% shooter. Want to be a better free-throw shooter in clutch situations? Be a better free-throw shooter, period. Better shooters in general tend to be better shooters during crunch time.
Some might say this view takes the fun out of the game, which those same people might say is the entire purpose of analytics. But let’s look at it from a different perspective. If you can just make free throws in important situations by CONCENTRATING, there’s less incentive to work on becoming a better shooter. Hey, Tony Wroten, stop daydreaming when the game’s on the line! Just CONCENTRATE and you, too, can come through in the clutch like C.J. McCollum. It’s that easy. Whereas, in the non-clutch world, your ability, achieved through a combination of natural talent and hard work, relates to your likelihood of late-game success. It’s no guarantee of success though, as Wolters’ ledger shows. But it requires much less good fortune for great players to succeed in those situations than it does for good players.
I like the latter theory better, but I also think any sensible interpretation of the data shown above would lead to that determination. Sure, some guys perform better than normal in close-and-late situations. Some guys also perform better than normal in other segments of the game. Is there any reason to think Nate Wolters (89%) had special powers with between 15 and 20 minutes left in the game and Jared Cunnigham (53%) didn’t? Then why do we need a reason to explain similar variation among players in the last five minutes?
*To be clear, I do believe that players generally perform better when the outcome of the game is in doubt than when it’s not. And if you want to call that clutchness, I’m fine with it. However, it’s a small factor in a sea of dozens of other factors that influences a player’s performance in any given possession. I don’t believe that a handful of special players have super-duper clutch powers.