Paper denies effects of mental state in player performance

Discussion in 'General Chit-chat' started by Waterd103, Apr 16, 2012.

  1. Waterd103

    Waterd103 Well-Known Member

    https://docs.google.com/viewer?url=...e.Tversky-Hot-Hand-Basketball.pdf&chrome=true

    A Paper that basically try to see if players belief in hot streaks is true or false.
    The paper shows that ability of "regular" people as well as players to on average predict the incremental chance a player has to score based on previous immediate results.

    The paper succeds, yet I quote.
    "However, the outcomes of both field goal and free throw attempts were largely independent of the outcome of the previous attempt"
    Which is to my knowledge crazy claim or one that defies too much current knowledge.
    He later emphazise this idea when he says "The independence between successive shots...."

    He later says says"A major difference between the two process [The process of coin toss and trying to score in basketball] is that it is hard to think of a credible mechanism that would create a correlation between successive coin tosses, but there are many factors (e.g., Confidence, fatigue) that could produce positive dependence in basketball. The availability of plausible explanations may contribute to the erroneous belief that the probability of a hit is greater following a hit than following a miss."

    It's tables and experiments are based on small samples that do not show this lack of correlation or independence. There is no way that seeing those tables (where some players show from 13% increased score rate to -13% decrease in score rate) one can conclude there is no correlation.

    The idea that mental state or previous immediate results doesn't affect future performance goes against a lot of what has been learned in poker, where programs exist to find these patterns over millions of hands.

    Not only the whole concept is proved outside this study that there is correlation, but the study itself fails to show there is no correlation.
    Can somebody argue otherwise?

    This is a very very important study because if mental state has no effect whatsoever in performance of players, I'm sure that will interest A LOT to some people and should have profound impacts on how sports and games should be approached in competition.
  2. Leartes

    Leartes Well-Known Member

    The paper is from 1985 is there no more recent research on the topic?

    EDIT:
    It should be fairly easy to do the math at home if free throw statistic from the current/last seasons are available to the public (don't know but basket ball has a lot of statistics freaks).
  3. CrystalChaos

    CrystalChaos Moderator Staff Member

    It is based on streaks within a game so I don't think it would be easy to do that at all, unless there is some database that has the order of made or missed shots.
  4. Waterd103

    Waterd103 Well-Known Member

    I don't know if there is recent research, maybe there is, I just posted it here because it seems some people on the chat believe this paper to be valid and it made me sad.
  5. Leartes

    Leartes Well-Known Member

    You'd need to check if it is more or less likely to hit/miss the second free-throw after the first. Or you could analyze how much the hit-rate varies between matches - not only for free throws but for all attempts. If you find a lot of variance in the hit % between matches it indicates streaks.
    Obviously some variance is expected since the opponent is better/worse but this could be normalized by calculating the average hit-rate vs any specific team and modifying the individual player rates by that.

    I think the paper is valid but it only scratches the surface and therefore the conclusion might be wrong.

    There might be players that have a natural tendency to do long or short streaks. e.g. you could get overconfident after starting a streak and therefore break your streak a lot. Other players build longer streaks. The paper only indicates that there are no obvious results for this sample size. Also keep in mind that there were outlier in almost all tables - not sure if that is to be expected ...
  6. Waterd103

    Waterd103 Well-Known Member

    This blog entry http://blog.axonpotential.com/there-are-no-such-thing-as-hot-streaks/ , that crystalchaos linked me talks exactly about this and show research that links to paper that proof that mental state does in fact affect performance (didn't read those paper yet).
    Sadly the conclusion of the blog entry is that there is a contradiction because the first paper shows no correlation. but if you look at the data it does show a correlation. So....
  7. SW

    SW Active Member

    http://xkcd.com/904/

    That said, I don't think mental state is that huge of a factor in high level play unless we are in "choke" territory. If you lose performance due to intermediate results affecting your mental state, you probably are not very good.
  8. Leartes

    Leartes Well-Known Member

    After spending some thought on it I believe neither the paper nor other experiments can easily prove or disprove the correlation.

    Just like SW said, if you lose confidence because of a short streak you're not really professional as it happens all the time. Also all experiments done in isolation shouldn't show any correlation because the pressure is just way lower than in any real game with thousands of fans screaming at you.

    Any data gathered from real games has a very high noise as teams play different strategies depending on their opponent.

    I propose future experiments should be done in a cleaner environment. e.g. testing accuracy and reaction time in first person shooters during a tournaments. It is far from ideal but data should be less noisy imo and the samplesize gets much larger as players take more shots.
  9. Logo

    Logo Well-Known Member

    The big thing to me is the stress/importance of the events. Missing a free throw in a game where game is meaningless (early on in season, already in/out of play-offs, game score already has a wide gap) is no big deal and is unlikely to matter in performance. But what about when the game is down to the last quarter, very close, a decisive game (say game 7 of finals), and a player is lining up to take 2 foul shots when their previous 4 foul shots were all misses. Is that going to impact the % of making the shots?

    Basically I tend to agree that short-run hot and cold streaks don't affect performance much on their own, but when the pressure or focus is on a single game (rather than a season) or in say a high stakes poker game it may change and the study doesn't appear to account for that as they take large data sets regardless of game situation/importance.

    My intuition would lean towards saying that the # of events required to cause a shift in overall confidence* is smaller when the pressure of the situation is higher. Since individual basketball games rarely matter all that much most players won't have streaks during the regular season. I'd love to see the data analyzed on just post-season games, or even more specifically games where the team is facing elimination.

    *To bring out the effects mentioned by the blog linked by Waterd where thinking you will do bad means you do worse.

    ----
    Also foul shots are pretty clean data wise since there is no defense and they're always taken from the same spot.
  10. Leartes

    Leartes Well-Known Member

    They are great!

    But as you said, the weight they carry differs wildly and I doubt the samplesize is large enough if you take importance into account. Therefore I was brainstorming how to get larger clean sets. I was thinking about fps because they are much "purer" in the way that everything is much more streamlined.
    Sample size increases greatly but the importance of every individual action decreases.
  11. Gon

    Gon Active Member

    Spectator booing and the home advantage: A study of social influence in the basketball arena.
    http://psycnet.apa.org/psycinfo/1984-13606-001

    I read this full document a few years back. I can't find access to the whole thing, but the abstract has some information. Basically they found team performances to be influenced by periods of extreme crowd noise/booing. I felt it relates because it shows how performance following an emotional event can be altered.
    Leartes likes this.
  12. Jobber

    Jobber Active Member

    Analytics is becoming a huge aspect of the pro sports industries. They even have their own popular annual convention. As you could imagine, then, that there's been a lot of research on the concept of the "hot streaks" sports, and a lot of growing material on "clutchness." There's a lot of newer and better material to read, since analytics has become such a central part of the business of pro sports these days, but some of the very best material is proprietary that teams don't like sharing.

    Since this is an older paper (don't have time to read it all), most of the emphasis is probably focused on the raw results. Looking purely at outcomes, it's hard to argue against the interpretations of the results. If there's no statistically significant findings, that's just how it is. Unfortunately, it doesn't really analyze all of the contributing factors that affect likelihood of success or failure in an activity like shooting. Intuitively, I believe if you could magically observe all those factors, then yeah, you could actually make predictions of future success or failure, but we're just not at that point, and who knows if we'll ever be.

    That said, I read a few recaps about the Sloan Conference, and apparently biometric data is the new frontier in analytics.

    I don't think it is a good conclusion to assume that mental state has absolutely no effect on performance. Rather, you can say that an effect just wasn't observed, for reasons to be discovered. You can't isolate enough variables just by looking at data points.

Share This Page