You can even miss while you're hot (in basketball, probably because you feel hot, and take harder shots)
Nuh uh. If you miss, you lose your "He's on fire" streak.
Their method is charming: Take a real-life sequence of hits and misses and rearrange it completely at random. If there's no hot hand, the resulting sequence should be no more or less streaky than the original. Miller and Sanjurjo find the opposite: Real-life data consistently becomes less streaky when rearranged, suggesting that shooters really do run hot and cold.
This is both (I) pretty compelling and (II) difficult to be no one has ever done before, if no one has in fact ever done it before. It seems like the most obvious and straightforward way to test the hypothesis.
It seems like a round-about way of proving that shitty defense is highly correlated within a given game.
3 --I thought that's why they were testing free throws?
2: I think the obvious way to test it was what the original authors did: seeing how often people score after they have already scored several times in a row. I'm surprised there's something wrong with this, though I'm too lazy to read the explanation why.
The boy/girl family sideshow in the slate article is a little confusing but I thought I sort of followed it, but the discussion of the same point in the separate linked article on the supposed Google interview question totally lost me.
I didn't read the article so I didn't know it was free throws.
That J/rdan E//enberg pops up everywhere! I swear there is no person who pops up more times as a friend-of-friends in my FB feed, and general interest articles and books, who has no idea who I am.
If he's popping up on your FOF lists, then you're probably popping up on his, too. So he might know who you are, at least as much as you know who he is.
I'm on a phone conference, so I can't open Slate in case they have ad that makes noise.
Well, I'm not doing and saying particularly interesting things. I'd ignore me.
I either don't have the math to get entirely clear on what the arguments are doing or don't have the necessary motive to figure it out, and the latter means I have no idea which it is. But I do have to admit that "scientists and math-y people say X, but recently some economists argued Y" seems to me to have a pretty reliable rule of thumb associated with it.
Here is more on the hot hand issue:
http://andrewgelman.com/2015/07/09/hey-guess-what-there-really-is-a-hot-hand/
http://andrewgelman.com/2015/09/30/hot-hand-explanation-again/
I still don't understand it
12 is funny.
To the OP, it really is funny how emphatic hot hand denialists are that the experience that all athletes describe*, that anybody who's achieved even a tiny bit of athletic success has felt, is imaginary. I mean, it goes with the general sabermetric approach, which says that outsiders looking exclusively at numbers can have insights that are obscure to those in the game, but that doesn't mean people in the game know nothing. I think that's becoming more and more understood/appreciated, but I think veterans of the original sabermetric wars aren't willing to give up on their shibboleths, and hot hand denial is one of them.
*one which correlates pretty well with objectively-measured success! It's not as if an athlete in a slump says, "It's weird, the pitch looks like it's moving so slowly, I can read the spin, yet I keep rolling over to second base." The denialist claim is, presumably, that athletes think they're seeing a giant basket because the shots keep dropping, which is stupid, but I don't know how else they shrug it off.
I guess I never thought about this in detail before, but I find it pretty obvious that success in a given shot correlates with success in another shot to a degree that is inversely related to the amount of time between them.
What is it with the insistence that a number divided by zero is infinity? Presumably it is in some mathematical senses and purposes, but in this particular context I'd think the better answer is "indeterminate / not valid".
Andrew Gelman gets it right. The original studies didn't have a large enough sample to determine that there was no hot hand, but it's still very true that people see a hot hand much more often than it actually exists. That is, the original study was wrong, but it's not wrong for the reasons that 14 is saying. The subjective experience athletes are having is likely at least as far from reality as assuming "no hot hand."
The denialist claim is, presumably, that athletes think they're seeing a giant basket because the shots keep dropping, which is stupid, but I don't know how else they shrug it off.
There are all sorts of contexts where people's sincerely held beliefs about what they're seeing right in front of them are just wrong, and looking at stats demonstrates it. The contexts where I've been convinced of it are usually race/gender sorts of things, but also things like belief in imaginary food intolerances and so on.
I haven't gone over the hot hand math, but if it were true that it couldn't be found in the stats, all the first-hand reports of the basket looking huge wouldn't convince me that something was going on.
The really important point of all of this, is that you need to pay much closer attention to the size of the effect that you're measuring. If the effect size is small, then you need a really big sample to detect it. Furthermore, if you expect the effect size to be small and you nonetheless manage to get a statistically significant result with a small sample, that means that almost certainly that answer is wrong even though it's "statistically significant."
I've always thought, on the basis of remote visual analysis* that some players probably are streak shooters while others probably shoot very close to whatever their average is.
*Watching games on tv.
Is 17.last arguing that athletes think they're hot more often than they really are? That when David Wright went 3-for-24 against the Dodgers, he would have told you that the pitches looked big and slow?
And to 18, the point is that the athlete knows, immediately and with clear, objective confirmation, whether he's having success. This isn't a hazy, after-the-event self-assessment (which is what the race/sex things you're using as comparison tend to be). A guy who thinks he had a great week because he hit a couple home runs, when in fact he was otherwise useless, isn't what we're talking about.
I'd add here that, most of the time, athletes playing well don't claim to be in that sort of zone. I mean, you'll get, "I'm seeing the ball well right now," but in a cliche, "one game at a time" sense. You can hear it in their tone, the enthusiasm, because being in that kind of zone feels amazing.
I think the idea is that athletes who are having 'streaks' that aren't statistically surprising at least sometimes interpret that experience as a subjective feeling of 'can't miss'. Not that the belief that "I'm hot now" is unrelated to their success, but that it's caused by success, rather than being independent information.
I think the idea is that athletes who are having 'streaks' that aren't statistically surprising at least sometimes interpret that experience as a subjective feeling of 'can't miss'.
Right. The way that I think of the "no hot hand" position is that it doesn't argue that streaks don't occur and, importantly, it doesn't make any argument about the subjective feeling of a streak, it just argues that streaks don't occur more frequently that you would expect from independent random events.
21 is definitely the case, and there's perpetual debate about whether streaky players are any more or less valuable than consistent ones with the same overall numbers. It probably varies by sport (in basketball, one player really can carry a team in a way that a [non-pitcher] baseball player simply can't).
I should add that even consistent players are streakier than they seem; on any scale less than (at least) a few games, outcome distribution is always lumpy.
They computed each player's shooting percentage in two different contexts: after three straight hits, and after three straight misses. If players' shooting tended to run hot and cold, as the average fan believes, you'd expect a shooter to be more likely to hit a free throw after three straight hits than after three straight misses.
Here's a dumb question: I don't get why this is the initial comparison. Why wouldn't you just compare P(hitting a free throw) to P(hitting a free throw, given three in a row)?
23: That just doesn't align with what that experience feels like.
You bike. Are there not days when it feels like you're dancing in the pedals, just effortlessly flying? You don't look at the clock when you arrive in order to identify that feeling, do you?
Actually, specifically on basketball, there's one point I'm unclear on: feeling in the zone is generally fleeting, where you hit a bunch of shots in a row. But guys also have runs of hot games, and what I'm unclear on is whether they describe it as feeling in the zone the whole time, or whether it just feels like they're hitting that zone for a stretch in each game.
Surely this is discussed, but I've never followed the game that closely.
23: That just doesn't align with what that experience feels like.
Did 24 make sense? In my mind the "no hot hand" theory doesn't have anything to say about what the experience feels like -- that isn't what the theory is about.
(I say this having had the subjective experience of days when it feels like I can't miss and days when it feels like I can't hit anything.)
Speaking of wheels and streaks, if you use a wheel chair in Pittsburgh, you're probably starting to feel like it's not your day.
Arguably the Mets player who has hit home runs in 6 consecutive playoff games, an all-time postseason record, would be an example of being hot. But people fail to note that for much of baseball history the playoffs were often less than 6 games in total (prior to 1969, pennant winners, based on record, went straight to the best-of-7 World Series. Coincidentally 1969 was the first year the Mets won.)
Anyway, last time we had this argument wasn't it about fair coins having hot streaks? Sure, real humans, psychology, whatever, but are people still arguing that streaks occur in events uninfluenced by humans?
I don't know the percentage of the city that uses a wheel chair, but let's say 1%. And there's maybe one homicide a week in the city. And the number of pedestrians killed by a car is probably below one a week. So the odds that a single day would see one guy in a wheel chair get shoot and another get hit by a car and pushed into a bus would have to be pretty small.
Similar to the argument debunking Jeter as some magical October player because there are all these stats about him having e.g. the most playoff hits ever. Well yeah, he played in an era of 3-round playoffs, so when his teams won the WS he had over twice as many ABs in which to get hits.
Between that, the cyclist last week, and the guy running the ten mile race this weekend, I'm thinking of sitting in my living room for the rest of the week.
Wait, they used free throws? If so, I'm not even going to discuss this anymore, and while the punishment might seem disproportionate to the offense, I hope they (and their children) burn in hell.
I applaud your vigor, but spell out the reaction some?
Just that you don't do a bunch of free throws in a row, so 'hot hand' doesn't make sense?
Just because I don't read links you post doesn't mean you can get away with it.
29: I understand the arguments of 24. Aside from the fact that the underlying math behind your final sentence may turn out to be wrong, I'm not sure I buy it, but it's plausible.
I guess where I struggle is, if streaks are pure random variance, then why do they feel different (as you yourself say they do)? I mean, I know the LB argument, but I'm rejecting it because, again, not every athlete who's playing well experiences it as being in the zone. If it were pure confirmation bias, then every hot player would say (and feel) that they can't miss right now. They don't.
There are individual differences between how susceptible to confirmation bias different athletes are? Doesn't seem impossible.
There are also days when you just subjectively feel good. If one of those overlaps with a statistically unremarkable 'streak', it'd be natural, but wrong, to connect them.
I once went through a streak of being very susceptible to confirmation bias.
31, 33: Sure, but Murphy's entire batting line is amazing; it's not just the homers.
Also, we've had 20 years in which 2 teams play a minimum of 11 games, and I'm sure the average is well over 14. And the two additional teams that lose in the second round of the playoffs play a minimum of 7 games and, again, more often 9+. That's a lot of potential 6 game streaks. Looking only at the best 5 hitters on each team, you're looking at more than 2600 potential 6 game streaks. It's true that Murphy's not really competing against guys in the first 65 World Series, but there've nonetheless been a lot of guys who had the chance to get there first.
More importantly, the fact that the regular season record is only 8 tells you how remarkable 6 is.
Holy cow. Only 17 guys have homered in 6 straight games in all of ML history*. Now 18.
*well, you can't look at the deadfall era for that. So since !920. And, indeed, it was achieved in '22 and '24. Gehrig did it, but never the Babe.
Let's talk about confirmation bias: this apparently incorrect study has been embraced for 30 years by... people who want to believe that there's no hot hand. Hmm.
spell out the reaction some?
A hot hand in basketball is about rhythm, turning off some part of your conscious mind, and usually also feeling super-energized. Free throws are antithetical to all of that. You have to stop, calm down, think about what you're doing, and throw up a gentle, precision shot. In fact, one way to disrupt the rhythm of a hot player is to foul him. Send him to the line, take him out of his groove.
Also, a genuine hot hand is pretty rare, and different (obviously) from being very good, but also different from playing well relative to your own mean. We're talking about a five, maybe ten minute period that happens a few times a year for any given player.
I have no grasp on any details of any of this, but focusing on the feeling as necessarily meaning something seems delusional. We know we shoehorn our experience of the world into story form, as with the sugar/hyperactivity link.
Maybe I'm misreading this but there are a ton of guys (including the usual big names) who have destroyed the ball in the postseason. Murphy is at .421, 7HR, OPS 1.462. Except for home runs that doesn't even get him onto any of these leaderboards. Lloyd McClendon (who?) had an OBP of .750 in 16 PAs the 1992 NCLS for the Pirates.
22: I think what you'd have to test is how many times they were "streaky" without having had that feeling. That's why you'd need the numbers; it wouldn't be obvious to the athlete without the feeling. If streaks happen reasonably often at random, they're only going to need to line up now and then with hot hand feeling to induce the belief that they always go together.
Only 17 guys have homered in 6 straight games
Homered in on 6 games. Or maybe you mean honered.
Honus Wagner was probably one of them, yes.
48 -- the thing about Murphy is that he isn't usually a power hitter at all, so it has a special "deal with Satan" quality to it.
On the OP, I've always assumed that people can indeed get hot and feel in the zone, but that it's basically unpredictable when such streaks will end. It's still a human activity generated by skill, however, so it's conceptually odd to think of the hot streak as "random." Other than that, my math is at an approximately 5th grade level so who knows. Value-added commenting!
Or maybe you mean honered.
I actually typo'd that and chuckled to myself.
48: His career SLG is #2 all time, with a few more PAs than the #1 guy (who's just ahead). Because of the way they're calculating, he doesn't qualify, but he will after his first at bat tomorrow night.
Looking at OPS doesn't tell you much, though, without context. For instance, Manny Ramirez, the only guy on that list who's got close to as many PAs in a single postseason, has an OPS 285 points higher. but his wRC+ (similar to OPS+, but better IMO) was 331 to Murphy's 305. So 25 percentage points better than Murphy, and in a smaller sample. 25 points, as it happens, was the difference between Trout and Harper this year. 28 points below that gets you Goldschmidt, so we're not talking big differences at that end of the scale.
Without being able to sort for a minimum number of plate appearances, you're really not looking apples to apples. The Legendary Lloyd, for instance, had that crazy OBP in just 16 PAs, far fewer than half of Murphy's.
53.3 - what? Do you mean career SLG in the post-season?
Except for home runs that doesn't even get him onto any of these leaderboards.
Some of those stats are skewed by the number of plate appearances. Murphy has 39 appearances so far, and is slugging 1.026. That doesn't quite get him into the leader-board, but he has more plate appearances than any of the guys in the top 10. And maybe 1.026 that doesn't compare with Gehrig hitting over 1.700 in the 1928 World Series, but Gehrig did that in a much smaller sample sized of four games with 17 plate appearances.
See above for fifth grade math level, but aren't post-season stats just totally pointless? The sample size is so small and the postseasons now are so different, I don't see what you get from numbers that you don't get from "Daniel Murphy has gotten insanely hot and has hit a shit-ton of home runs this post-season."
1: Was not expecting to be reminded of NBA Jam today.
Free throws? That's got to be a joke. No one talks about a hot hand in free throws or even a streak unless it's like 10 or more in a row, which for many players would take more than one game.
30/32/34: Oh jeez. I'm home sick and put me on team never cross a street again.
59: This is mostly true, but there are exceptions -- if DeAndre Jordan hit 10 free throws in a row...
57: Stats have two types of uses, narrative and predictive/analytical. The smallness of the sample means that they have no predictive value, and tell you little of the underlying performance*, but they still tell you what's been happening, and let you compare with historic performances. And, indeed, they tell you that Soler was roughly as good as Murphy, but just didn't hit as many homers.
Point being, they do let you fact-check the emotional narrative. In an alternate universe where Murphy had hit those homers but done little else, a look at the stats would tell you that the homers were neat, but his overall value was middling.
*for the season, Murphy's power really did increase, and that's a big enough sample to say that it was a real-ish thing, but the current power surge tells us nothing going forward, because in 40 PAs (almost) anyone can get that hot
I'm not going to deal with that because tired.
What's the evidence against sugar and hyperactivity? Just that people are bad at judging what constitutes hyperactivity or that there isn't much link between sugar and energy levels?
People can't distinguish between the behavior of kids given sugary soda and the control group given diet soda, IIRC.
66: Correct. ISTR that it was Israeli research that pretty definitively debunked the link.
What about kids being given one sugary soda and kids given five sugary sodas?
There's probably a point at which the IRB won't let you give the kids that much sugar.
So, the IT people who are asking me why I want software installed on my laptop instead of using a remote connection probably would not like it if they knew I used nothing but gmail for work and that my only back up is copying my files to a thumb drive that I carry home.
70: Correct. Research from Bulgaria, Nevis and Cambodia concurred on this subject.
You clearly haven't caught the Hamilton fever that's sweeping the nation.
19: Or the other way to interpret a negative result is as a bound on the possible effect size. You can't really prove that there is no possible effect; you can establish that if there is an effect, it is so small that you shouldn't worry about it in practical terms. Like the baseball clutch hitting studies that show that any clutch hitting effect, if it exists, is probably limited to one or two extra hits per season:
So it is legitimate to say, "Benito Santiago hit .267 last year, but he's been one of the top clutch hitters in baseball, which makes him as valuable as a .269 hitter." The data backs you up here.
But replace .269 by .271 and you are depending on Santiago's clutch data being reliable, rather than due to luck. Replace it by .272 and you have gone beyond any prediction which can be justified by the statistics alone. Replace it by .276 and you have made a statement which is inconsistent with the data on clutch hitting.
73: The last paragraph should also be part of the blockquote.
. Like the baseball clutch hitting studies that show that any clutch hitting effect, if it exists, is probably limited to one or two extra hits per season
One or two extra hits, maybe, but they were clutch hits.
59: Yes, free throws. Which have the obvious advantage that they are susceptible to statistical analysis without all the obvious confounders of field goal shooting (player changing shot selection in response to success, defense playing differently, etc.). It's the same shot, from the same position on the floor, without the defense being a factor (other than how they put him there).
While studying free throws has a little bit of "the light is better over here" to it, I definitely heard people talk about streaky free throw shooting prior to all the studies about it coming out.
a little bit of "the light is better over here" to it
That's all it has. Three free throws in a row is in no way aa meaningful streak. You hit three free throws and miss the next one regularly and you're a mediocre free throw shooter at 75%.
What I've never understood about hot hand research is that, as a default, it seems to treat a shot as a random event - a coin flip, as some of the linked material discusses. But even coin flips aren't necessarily "coin flips" - with a little practice you can cheat the toss to influence the outcome.
Many more people practice the mechanics of shooting a basketball than the mechanics of cheating a coin toss.
A basketball player isn't a 40% shooter because the God of Statistical Probabilities made it so; he's a 40% shooter because he has, overall, better control over his mechanics than a 30% shooter. And judging from my experience as a mediocre adolescent basketball player, that control isn't equally distributed over time. For one thing, if it were, how could players become better shooters? Literally, what do anti-hot-handers think is involved in that process? It's been 10 years since I touched a basketball, but I still know what the two habitual mechanical errors in my shot were. When I committed those errors, my shots had a bad chance of going in. When I didn't, they had a much better chance of going in. And I was fully aware of when I was and wasn't committing those errors, and to a certain extent I could Jedi mind trick myself into better success. Sometimes. And then I felt "hot". (In a nutshell, "just keep your eye on the ball" turns out to be pretty profound and tricky advice.)
On the subjective feeling point, people feel like they're on a hot streak in games of pure chance (think gambling) and it's frequently, but they're just wrong. The question is to try to tease out how much of the feeling you have a hot hand on the basketball court is the same phenomenon as feeling like you're on a hit streak in craps.
OT: When the Amish go to McDonald's for breakfast, they go big.
You hit three free throws and miss the next one regularly and you're a mediocre free throw shooter at 75%.
But you get to say, "In you face!" to Shaq.
||
Here's a story just for Halford: Britich and Irish cows are closer to aurochs than European cows in general.
|>
83 is interesting but not surprising really since modern cattle are only domesticated aurochs. Presumably they would interbreed whenever they had the opportunity, so all you need is a region with a thriving wild population, a small number of introduced domesticates and fairly crude methods of controlling the latter (ranching, not farming). I believe the last aurochs lived in Poland; they should sample some bones there.
Was there a record for "most inside-the-park home runs on the first pitch of the World Series"? I guess there is now.
There's a report that the KC pitcher's father died (age 63, heart disease) right before the game and it's not clear if he's been told yet or whether the team will tell him during the game. Fox has said they will not mention it during the broadcast.
11: not true heebie you are awesome.
87: Oddly, I heard Fox mention it on the broadcast...
Here's the story. It seems that Fox mentioned it after his wife told him (after he'd been pulled). Amazing story, all the moreso because he's the third Royal to lose a parent in the last 3 months.
Never happens where it might be appreciated.
BTW, to the Murphy discussion, I heard somewhere that, among guys with 39 or more PAs, his was the third-highest OPS of all time.
Of course, that's before he dropped it 138 points, all the way down to a paltry 1.324. That's in 45 PAs, and I'd be surprised if the larger sample size doesn't keep him very high.
The Streak is dead, long live the Streak.