Hot Hand or Hot Head (Sandy Weil, 2009)
Posted: Fri Apr 22, 2011 2:17 am
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Tue Mar 10, 2009 10:45 am Post subject: Hot Hand or Hot Head Reply with quote
For those of you who attended the Sloan conference, I've posted the slide deck on my website.
http://www.sportsmetricians.com follow link to hot hand.
I hope to post a draft of the paper (very dense) a bit later: for those who are interested in some of the off-slide results, the additional robustness checks, and the summaries of the findings for those who did not get to hear the talk.
---------
A quick summary for those who did not hear the talk: we find that these players we studied shoot worse after a make than after a miss; stronger effect after a made jumper than after a made non-jumper. After a made basket, they shoot more jumpers and fewer non-jumpers. And they shoot sooner. There is a summary slide at the end of the slide deck.
The draft of the paper contains much more explanation.
---------
We welcome any feedback you might have.
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Tue Mar 10, 2009 11:21 am Post subject: Reply with quote
Sandy, welcome to the interrogation room.
I can't tell from the slides: After a made shot, is a player more likely to shoot a 3ptr? Do the extra points from made 3's offset the .035 lower FG%
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
schtevie
Joined: 18 Apr 2005
Posts: 182
PostPosted: Wed Mar 11, 2009 9:13 am Post subject: Reply with quote
Sandy,
Thanks for the slides. Waiting for the draft of the paper, I was wondering if you could answer a couple of question about the interpretation of the results. In your summary, you write that "if everyone on the team behaves this way, it costs the average team 4.5 wins per season".
Do you have an estimate of how much a team with the "hottest hand" would actually be hurt by its misperceptions, given that all other teams, presumably, are making similar offsetting mistakes?
Also, does the 4.5 wins per season include the opportunity cost of shooting early? That is to say that the cost is not just the lower shooting percentage on the "hot" shot but also the forgone opportunity of finding a better one.
Thanks.
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Wed Mar 11, 2009 4:09 pm Post subject: Reply with quote
Mike G wrote:
I can't tell from the slides: After a made shot, is a player more likely to shoot a 3ptr? Do the extra points from made 3's offset the .035 lower FG%
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
Thanks, Mike.
On your first question, yes. On average, a player is more likely to shoot a 3fga after a made basket than he is after a missed basket. To see those results, check out the far right column on slide 9 (determinants of current shot type). On average, their 3fga share goes up 5 percentage points after a made jumper.
On your second Q: I'll have to check and get back to you.
On your third Q: I'm not sure what you are asking that we could do with eFG% or TS%. Please clarify.
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Wed Mar 11, 2009 4:20 pm Post subject: Reply with quote
schtevie wrote:
Do you have an estimate of how much a team with the "hottest hand" would actually be hurt by its misperceptions, given that all other teams, presumably, are making similar offsetting mistakes?
Not yet, no. This 4.5 was a very rough cut at the cost to teams.
That hottest hand is an interesting way to look at the cost to the team because, as you say, all teams are probably hurt by this perception. But, since it is already probably a pipe dream to think that a team could change its players' behavior/perception, I'm not sure that it is any more meaningful to consider the cost of changing from the max to the mean (as your calc would imply) than to think about changing it from the mean to zero (which is the 4.5). I guess if you could replace your players with average-affected players, then your figure would be the one to consider.
schtevie wrote:
Also, does the 4.5 wins per season include the opportunity cost of shooting early? That is to say that the cost is not just the lower shooting percentage on the "hot" shot but also the forgone opportunity of finding a better one.
It does not include an opportunity cost. That might be an offshoot that we could get from modeling the player's team points on the next chance as a function of the player having made/missed his previous shot (which I started thinking about while responding to Mike G and which your comment has further encouraged me to consider). Thanks.
Back to top
View user's profile Send private message
timmys24
Joined: 19 Feb 2009
Posts: 3
PostPosted: Thu Mar 12, 2009 11:28 am Post subject: Reply with quote
Thanks for sharing this info.
I apologize if I'm off point as I did not hear the talk and am going solely by the slides and this thread.
Is shot selection after a made basket the primary reason the hot hand doesn't seem to exist?
I wonder if there is a way to measure players who do not act as if they have the hot hand, and continue their shot selections in the flow of the offense to see how effective they are. (eliminate shots taken sooner, and the increase in team's portion of shots.)
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Thu Mar 12, 2009 1:26 pm Post subject: Reply with quote
I've posted a very rough and VERY DENSE version of my draft of the paper to my site. It has 25 pages of text and 11 pages of tables: you've been warned.
http://www.sportsmetricians.com
link to hot hand.
Back to top
View user's profile Send private message
Mountain
Joined: 13 Mar 2007
Posts: 1492
PostPosted: Fri Mar 13, 2009 4:33 am Post subject: Reply with quote
I've only skimmed the article but I wonder if the focus on the most prolific shooters was the way to go or if a follow-up will help further. Looking at the list I'd call a lot of them steadily "big-headed". A short hot streak and their feeling of being hot probably does make them even more "hot-headed" and perhaps less efficient than for other players who aren't as big-headed normally and might feel a bigger difference of when they are 'hot" and might be more right?
Is it possible to see how players perform on their next five shots after becoming minimally hot? Are players more right over the course of the next five shots than over the next one?
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Fri Mar 13, 2009 8:46 am Post subject: Reply with quote
Sandy Weil wrote:
Quote:
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
I'm not sure what you are asking that we could do with eFG% or TS%. Please clarify.
From the pdf file:
Quote:
... our subjects average 46.7% field goal shooting after any missed shot but it drops to 43.2% after a made jump shot. ..
.. assume that all the players on a team experience similar drops ... the average team gives up about 1.8 points per game ...
The reaction of a team to an opponent (who looks like he's heating up) is to send a double team, or at least a stronger defender. The counter-reaction of the hot shooter is to shoot before the double arrives. Sometimes this is a 3-pt shot, which when made is 50% more points.
Thus, you may compare effective FG% before and after a make/miss, rather than undifferentiated FG% (as if all FG were the same).
Another common event is for the hot shooter to be fouled. This may be more a defensive decision, or provoked by the hot shooter himself. If his FT/FTA are not counted in the before/after miss/make matrix (as they are in TS%), you again may lose a part of the player's effectiveness.
Yet further, the 'hot' shooter may get double-teamed and find open teammates for easy baskets. You might then wish to look at the team eFG% and TS% after a given player has entered the 'hot zone'.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Mountain
Joined: 13 Mar 2007
Posts: 1492
PostPosted: Fri Mar 13, 2009 1:34 pm Post subject: Reply with quote
I share Mike G's perspective / interest in also having study results by eFG% or TS% or you could even go to offensive rating.
I also share timmys24 interest in evaluating and knowing the names of the "cool hand Lukes" who don't react in the typical "hot hand" / "hot head" manner.
Back to top
View user's profile Send private message
Italian Stallion
Joined: 04 Mar 2009
Posts: 3
PostPosted: Fri Mar 20, 2009 2:19 pm Post subject: Reply with quote
Are we arguing that "hot hand" does not even exist?
Are we arguing that players that believe they might have a hot hand because of a made shot overcompensate by potentially taking worse shots?
Are we arguing that teams adjust effectively defensively when a player has a hot hand?
I think this is very interesting issue.
Many people think of the eFG% of a player as being kind of fxed, but IMO shooting is not like flipping coins where the percentage IS fixed. IMO a player's eFG% is variable from night to night within a range but his results also vary out of randomness.
It's variable depending on whether he feels extra energetic and focused or tired and distracted, whether he feels confident or insecure, whether his form is slightly off or is extra sharp because of specific practice drills etc...
I never played serious basketball, but I played semi pro billiards. I would often adjust my playing decisions for how I was playing on "that day".
I might be 90% to make a certain shot over the course of a year, but I was only 80% on some days and close to 100% other days. I also had good and bad days that I felt were just random.
We all see NBA players that are typically good shooters have terrible days and vice versa. It's almost impossible for a fan to know whether that's a random statistical event or the result of something unique going on with the player that day. It's probably even tough for the player himself to know (though I could tell pretty easily on the pool table) .
I've always felt that the really good players should be good at determing whether their bad/good streaks are random or they are having an especially good/bad day and adjust at the edges like I did at pool effectively but this sort of suggests they are over compensating (at least when they think they are hot)
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Fri Mar 20, 2009 2:36 pm Post subject: Reply with quote
Every person, animal, and piece of machinery has its good days and its bad days. Basketball is one activity in which a defense reacts to suppress the hot scorer. This effect dwarfs all others.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Harold Almonte
Joined: 04 Aug 2006
Posts: 576
PostPosted: Fri Mar 20, 2009 10:57 pm Post subject: Reply with quote
Kevin Martin appears to be a hot free throw attemptor every night, and defense (and refs) appears not to react to that.
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Tue Mar 10, 2009 10:45 am Post subject: Hot Hand or Hot Head Reply with quote
For those of you who attended the Sloan conference, I've posted the slide deck on my website.
http://www.sportsmetricians.com follow link to hot hand.
I hope to post a draft of the paper (very dense) a bit later: for those who are interested in some of the off-slide results, the additional robustness checks, and the summaries of the findings for those who did not get to hear the talk.
---------
A quick summary for those who did not hear the talk: we find that these players we studied shoot worse after a make than after a miss; stronger effect after a made jumper than after a made non-jumper. After a made basket, they shoot more jumpers and fewer non-jumpers. And they shoot sooner. There is a summary slide at the end of the slide deck.
The draft of the paper contains much more explanation.
---------
We welcome any feedback you might have.
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Tue Mar 10, 2009 11:21 am Post subject: Reply with quote
Sandy, welcome to the interrogation room.
I can't tell from the slides: After a made shot, is a player more likely to shoot a 3ptr? Do the extra points from made 3's offset the .035 lower FG%
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
schtevie
Joined: 18 Apr 2005
Posts: 182
PostPosted: Wed Mar 11, 2009 9:13 am Post subject: Reply with quote
Sandy,
Thanks for the slides. Waiting for the draft of the paper, I was wondering if you could answer a couple of question about the interpretation of the results. In your summary, you write that "if everyone on the team behaves this way, it costs the average team 4.5 wins per season".
Do you have an estimate of how much a team with the "hottest hand" would actually be hurt by its misperceptions, given that all other teams, presumably, are making similar offsetting mistakes?
Also, does the 4.5 wins per season include the opportunity cost of shooting early? That is to say that the cost is not just the lower shooting percentage on the "hot" shot but also the forgone opportunity of finding a better one.
Thanks.
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Wed Mar 11, 2009 4:09 pm Post subject: Reply with quote
Mike G wrote:
I can't tell from the slides: After a made shot, is a player more likely to shoot a 3ptr? Do the extra points from made 3's offset the .035 lower FG%
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
Thanks, Mike.
On your first question, yes. On average, a player is more likely to shoot a 3fga after a made basket than he is after a missed basket. To see those results, check out the far right column on slide 9 (determinants of current shot type). On average, their 3fga share goes up 5 percentage points after a made jumper.
On your second Q: I'll have to check and get back to you.
On your third Q: I'm not sure what you are asking that we could do with eFG% or TS%. Please clarify.
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Wed Mar 11, 2009 4:20 pm Post subject: Reply with quote
schtevie wrote:
Do you have an estimate of how much a team with the "hottest hand" would actually be hurt by its misperceptions, given that all other teams, presumably, are making similar offsetting mistakes?
Not yet, no. This 4.5 was a very rough cut at the cost to teams.
That hottest hand is an interesting way to look at the cost to the team because, as you say, all teams are probably hurt by this perception. But, since it is already probably a pipe dream to think that a team could change its players' behavior/perception, I'm not sure that it is any more meaningful to consider the cost of changing from the max to the mean (as your calc would imply) than to think about changing it from the mean to zero (which is the 4.5). I guess if you could replace your players with average-affected players, then your figure would be the one to consider.
schtevie wrote:
Also, does the 4.5 wins per season include the opportunity cost of shooting early? That is to say that the cost is not just the lower shooting percentage on the "hot" shot but also the forgone opportunity of finding a better one.
It does not include an opportunity cost. That might be an offshoot that we could get from modeling the player's team points on the next chance as a function of the player having made/missed his previous shot (which I started thinking about while responding to Mike G and which your comment has further encouraged me to consider). Thanks.
Back to top
View user's profile Send private message
timmys24
Joined: 19 Feb 2009
Posts: 3
PostPosted: Thu Mar 12, 2009 11:28 am Post subject: Reply with quote
Thanks for sharing this info.
I apologize if I'm off point as I did not hear the talk and am going solely by the slides and this thread.
Is shot selection after a made basket the primary reason the hot hand doesn't seem to exist?
I wonder if there is a way to measure players who do not act as if they have the hot hand, and continue their shot selections in the flow of the offense to see how effective they are. (eliminate shots taken sooner, and the increase in team's portion of shots.)
Back to top
View user's profile Send private message
Sandy Weil
Joined: 19 Jun 2008
Posts: 9
Location: Boulder, CO
PostPosted: Thu Mar 12, 2009 1:26 pm Post subject: Reply with quote
I've posted a very rough and VERY DENSE version of my draft of the paper to my site. It has 25 pages of text and 11 pages of tables: you've been warned.
http://www.sportsmetricians.com
link to hot hand.
Back to top
View user's profile Send private message
Mountain
Joined: 13 Mar 2007
Posts: 1492
PostPosted: Fri Mar 13, 2009 4:33 am Post subject: Reply with quote
I've only skimmed the article but I wonder if the focus on the most prolific shooters was the way to go or if a follow-up will help further. Looking at the list I'd call a lot of them steadily "big-headed". A short hot streak and their feeling of being hot probably does make them even more "hot-headed" and perhaps less efficient than for other players who aren't as big-headed normally and might feel a bigger difference of when they are 'hot" and might be more right?
Is it possible to see how players perform on their next five shots after becoming minimally hot? Are players more right over the course of the next five shots than over the next one?
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Fri Mar 13, 2009 8:46 am Post subject: Reply with quote
Sandy Weil wrote:
Quote:
Is it still possible to make comparisons relative to eFG% [Pts/FGA/2] and/or to TS% [Pts/(FGA+.44*FTA)/2] ?
I'm not sure what you are asking that we could do with eFG% or TS%. Please clarify.
From the pdf file:
Quote:
... our subjects average 46.7% field goal shooting after any missed shot but it drops to 43.2% after a made jump shot. ..
.. assume that all the players on a team experience similar drops ... the average team gives up about 1.8 points per game ...
The reaction of a team to an opponent (who looks like he's heating up) is to send a double team, or at least a stronger defender. The counter-reaction of the hot shooter is to shoot before the double arrives. Sometimes this is a 3-pt shot, which when made is 50% more points.
Thus, you may compare effective FG% before and after a make/miss, rather than undifferentiated FG% (as if all FG were the same).
Another common event is for the hot shooter to be fouled. This may be more a defensive decision, or provoked by the hot shooter himself. If his FT/FTA are not counted in the before/after miss/make matrix (as they are in TS%), you again may lose a part of the player's effectiveness.
Yet further, the 'hot' shooter may get double-teamed and find open teammates for easy baskets. You might then wish to look at the team eFG% and TS% after a given player has entered the 'hot zone'.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Mountain
Joined: 13 Mar 2007
Posts: 1492
PostPosted: Fri Mar 13, 2009 1:34 pm Post subject: Reply with quote
I share Mike G's perspective / interest in also having study results by eFG% or TS% or you could even go to offensive rating.
I also share timmys24 interest in evaluating and knowing the names of the "cool hand Lukes" who don't react in the typical "hot hand" / "hot head" manner.
Back to top
View user's profile Send private message
Italian Stallion
Joined: 04 Mar 2009
Posts: 3
PostPosted: Fri Mar 20, 2009 2:19 pm Post subject: Reply with quote
Are we arguing that "hot hand" does not even exist?
Are we arguing that players that believe they might have a hot hand because of a made shot overcompensate by potentially taking worse shots?
Are we arguing that teams adjust effectively defensively when a player has a hot hand?
I think this is very interesting issue.
Many people think of the eFG% of a player as being kind of fxed, but IMO shooting is not like flipping coins where the percentage IS fixed. IMO a player's eFG% is variable from night to night within a range but his results also vary out of randomness.
It's variable depending on whether he feels extra energetic and focused or tired and distracted, whether he feels confident or insecure, whether his form is slightly off or is extra sharp because of specific practice drills etc...
I never played serious basketball, but I played semi pro billiards. I would often adjust my playing decisions for how I was playing on "that day".
I might be 90% to make a certain shot over the course of a year, but I was only 80% on some days and close to 100% other days. I also had good and bad days that I felt were just random.
We all see NBA players that are typically good shooters have terrible days and vice versa. It's almost impossible for a fan to know whether that's a random statistical event or the result of something unique going on with the player that day. It's probably even tough for the player himself to know (though I could tell pretty easily on the pool table) .
I've always felt that the really good players should be good at determing whether their bad/good streaks are random or they are having an especially good/bad day and adjust at the edges like I did at pool effectively but this sort of suggests they are over compensating (at least when they think they are hot)
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 2363
Location: Delphi, Indiana
PostPosted: Fri Mar 20, 2009 2:36 pm Post subject: Reply with quote
Every person, animal, and piece of machinery has its good days and its bad days. Basketball is one activity in which a defense reacts to suppress the hot scorer. This effect dwarfs all others.
_________________
`
There's no I in analysys.
Back to top
View user's profile Send private message Send e-mail
Harold Almonte
Joined: 04 Aug 2006
Posts: 576
PostPosted: Fri Mar 20, 2009 10:57 pm Post subject: Reply with quote
Kevin Martin appears to be a hot free throw attemptor every night, and defense (and refs) appears not to react to that.