Observing Winning Streaks: Statistics and Psychology
Are winning streaks the result of superior talent? Or are they more random than we might suspect? Today on the blog Alex Titkov takes a look at how we as humans can often be deceived by recent runs of good form.
Whether we are having fun with our friends predicting who will win the next match, playing fantasy football, or placing a bet on this weekend’s matches, we try to look for information and patterns that will give us potential insight or an edge for us to reap personal or even financial reward. To get a better idea of these things this time around we’ll take a look at some statistical and psychological perspectives into the concepts of winning streaks and psychological momentum.
The first article I came across when looking into these phenomena was Andy Murray’s Olympic victory over Roger Federer last year. I spoke before about the idea of home field advantage and how typically there is an advantage towards a player playing on home ground due to familiarity of the stadium, lack of travel fatigue, crowd support and their influence on a referee. What was interesting though was that Murray had played Federer just four weeks prior at the exact same venue but had lost at the hands of Federer.
So what had changed?
In this article, In this article. Murray talks specifically about watching and being inspired by the recent gold medal success of fellow British athletes Jessica Enis, Mo Farah, and Greg Rutherford. This seems to have played an important part in his success. At the same time, the article gives insights from sport psychologists reflecting that crowd support also can play a part in the spurring on an athlete but given the fact that the venue was a controlling factor, it’s hard to measure if the crowd was any different four weeks prior.
Murray won the three sets decisively and given the pattern of those three sub-victories it’s easy to see why we would term this as a set of winning streaks. But is there truly such a thing as a winning or losing streak and a way to predict it?
The Randomness of Winning Streaks
One contrary perspective comes from Leonard Mlodinow who is a teacher of randomness at Caltech who published an article - The Triumph of the Random - and why baseball player Joe DiMaggio’s record 56 game hitting streak may just have as much attributed to chance as it did to his ability.
Mlodinow discusses the psychological disadvantage humans have when they attempt to predict outcomes, especially at the absence of a true pattern. He gives the example of a series of green and red flashes presented to an individual that alternate randomly but with the red flash occurring twice as often as the green flash. It’s then posited—what’s the strategy to predict the next flash? Mlodinow gives two answers:
- A non-human animal would more often than not guess red
- A human will more likely than not attempt to notice a past but false pattern
This is an evolutionary aspect of human psychology as we tend to notice patterns and interpreting them for our survival i.e. seeing a garden hose move and just briefly being frightened that it may be a snake. Though this has been beneficial for our survival, this can lead us astray when trying to predict future outcomes.
Mlodinow further adds this interesting gem to looking at randomness. If we took a graph of occurrences; say a series of wins, losses or draws having the idea of randomness in mind, we’d probably think the graph looks scattered without any real sense of stability or consistency.
But! An example of 100 coin tosses will yield a 75% chance of seeing a streak of 6 or more heads/tails and a 10% chance of seeing a streak of 10 or more.
While given these stats, Mlodinow is quick not to easily dismiss the role of talent in the case of DiMaggio for example. Factoring in his lifetime batting average of .325, this gives DiMaggio’s side of the “coin” an increased weight of 75% in his favor. Moreover he adds the human tendency for the need of control and deeply associating with progress and triumphing over the odds.
He showed this with an experiment in coin flipping among a group of Yale students who understood the randomness of the action but when asked if distractions or practice would influence the outcome, quite a few responded that it indeed would. If anything, it makes me think twice if an Ivy League education is all it’s really cracked out to be.
His concluding example adds in the subtleties of season-to-season performances. Using a computer analysis, they compiled randomly generated seasons from 1871 to 2005 from actual player statistics of each year and repeated the process 10,000 times (or 10,000 alternate universe baseball histories). They found 42% of the simulated alternate histories matched or exceeded DiMaggio’s streak with the longest being 109 games.
Other studies looking into the concept of psychological momentum have also been inconsistent and mostly have come from the perception of those watching the action. It’s been attempted to use the physics definition of momentum of velocity multiplied by mass wherein velocity is described as a “big play” and mass as “game importance” and “emotional investment”. Even while showing short clips of a game to a number of “fans” do indeed lend to the support of this model, the overall end result did not always follow suit.
There are limitations to these studies especially when measuring “big impact” moments as opposed to say a collective of “small impact” moments and ambiguity with the terms of psychological momentum and flow/zone.
Regardless of these things, it’s quite apparent that understanding and predicting momentum from game to game is indeed complex but it most definitely should be kept in mind the strong influence of chance.
If randomness has anything to teach us, it’s to be weary in certainty when comparing two patterns (WLWW or WWLL) and attributing false meanings to which one will indeed come out the victor.
Follow Alex on Twitter: @alextitkov
And read more of his work at the EMSEP Sport and Exercise Psychology Blog
- Tag: Betting Theory