Despite almost 30 years of research having generally pointed to the contrary, new research by Stanford Graduate School of Business professor Jeffrey Zwiebel and UC Berkeley co-author Brett Green suggests that the common perception of “hot hand” streaks in sports may not just be an illusion.
Hot and cold streaks refer to the sports phenomenon of stretches of time during which a player is doing exceptionally well or badly, respectively. A classic study by Amos Tversky and Thomas Gilovich found that such streaks were not to be found when examining basketball. Zwiebel found that most research done since that study suggests that what people see as streaks is nothing more than random chance — something he and his co-researcher aimed to assess.
“If you randomly flip a coin, you’ll see some streaks, so they are arguing that basically this is a very striking example of a cognitive mistake that many people are making,” Zwiebel said.
Zwiebel said that he and Green were motivated to study the phenomenon for two reasons. As sports fans, denying the existence of these streaks seemed wrong. Zwiebel was also concerned about possible flaws in the work of behavioral economists investigating hot hands, especially in regards to tests without proper statistical power, and failing to look at endogenous responses.
An example of an endogenous response is when an opposing basketball team could alter defense in such a way as to prevent players on hot streaks from making an exceptional number of shots. Zwiebel cites Kobe Bryant as an example of this principle in action. Despite his high level of ability, his career shooting percentage is around the league average.
In order to address the dynamic defense concern, Zwiebel and Green focused on baseball, in which such mobilization of the defense is more limited. To address concerns about statistical power, Zwiebel and Green accumulated over a million data points from 12 seasons of baseball.
The results of the work were highly promising: According to Zwiebel, statistical and strategic significance for the hot hand effect is demonstrated in multiple categories. The study found that the difference in performance in players between hot and cold streaks is about a standard deviation.
“So it means if you’re an average player normally, the difference between a hot and cold player would be the difference between a 50th-percentile player and an 80th-percentile player,” Zweibel said.
Itamar Simonson, a professor of marketing at the GSB, teaches the original research done by Tversky and Gilovich in a doctoral-level class.
Simonson said that the phenomenon of hot hands is relevant to finance because it ties into the idea of confirmation bias.
“So confirmation bias is the simple idea that if you have a hypothesis, if you believe in something and then you see evidence that is kind of ambiguous, you tend to find support for your prior hypothesis,” Simonson said.
As an example, Simonson cites a study done by Tversky and Donald Redelmeier that showed patients with arthritis would endorse the idea that joint pain was associated with rain, even when day-by-day responses and weather reports showed no relation between rain and the pain they had experienced.
Given the new perspective of baseball, however, Zwiebel believes that his work may suggest a new theory with a greater scope.
“Take any activity where you have some people who are better than others, or some people have a pay scale and there’s good and bad participants, [and it] is likely to have both long term and short term components of ability,” Zwiebel said. “So that’s actually our hypothesis, where a hot hand is a short term component.”
Contact Skylar Cohen at skylarc ‘at’ stanford.edu.