The recognition heuristic in sports betting predictions.
Stuart and Penny enjoyed to travel, on this occasion they took advantage of the excitement over the world cup as an excuse to go on holiday. They saw the sights, enjoyed the local delicacies and watched a few matches. Stuart decided to place a bet on one of the matches but had never gambled before. Penny noticed that many people when betting simply chose the team, or athlete, that they were familiar with as the winner. Since neither Stuart nor Penny had gambled before and they did not know a lot about the athletes they both agreed that choosing the familiar team as the winner was the best way to gamble.
When gambling on the outcome of an event such as a football match, tennis match, or even a golf game many people predict that the familiar athlete or team will win. Regardless of the type of event or scoring criterion choosing the familiar is often the strategy that is used by laypeople (i.e., non-experts). This ‘familiar strategy’ is called the ‘recognition heuristic’ (Gigerenzer & Goldstein, 1996). The recognition heuristic has been studied extensively in the decision-making literature with many different examples of predictions for sporting events.
Researchers from the University of Illinois investigated the role of the recognition heuristic when making predictions about the outcome of basketball matches (Jacobson et al., 2009). Jacobson and colleagues (2009) choose to use basketball from the National Collegiate Athletic Association (NCAA) as the medium to study the recognition heuristic. In 2007 alone, an estimated $2.25 billion US dollars was bet on the final four rounds of this basketball tournament (McCarthy, 2007). Of the teams to get into the final four rounds 70% of these teams were highly seeded. Gambling records show that the favourite teams were over-backed when betting. The researchers used data from matches between 1985 and 2009. They asked people to predict the outcome of basketball matches based on very little information – just the names of the teams. The results revealed that most of the participants used the recognition heuristic, predicting that the well-known basketball teams would win the match.
A second experiment that used basketball data to investigate the role of the recognition heuristic utilised data from the National Basketball Association (Hall et al., 2007). One hundred and sixteen undergraduate students from Princeton University took part in the experiments by Hall et al (2007). Firstly, participants were asked to list as many NBA teams as they could recall. The data from sixteen games involving these teams was then selected. In this experiment all participants received statistics (win record, halftime scores), and half the participants received the names of teams (e.g., NY Knicks vs NJ Nets). Participants attempted to predict the outcome of games based on the information they received. The Los Angeles Lakers are a very familiar team with many people but did not have the best statistics – if using the recognition heuristic participants would ignore the statistics and predict the LA Lakers as a winner (unless against a more familiar team). Contrary to the recognition heuristic most people would expect that the more information we are given the better we can make predictions. The results of this study found that participants overestimated the chances of winning for familiar teams – using the recognition heuristic.
The recognition heuristic has also been studied with football (soccer) data (Pachur & Biele, 2007). In a third study laypeople and experts were asked to make predictions about the outcome of football matches in the 2004 European Soccer Championship. Pachur et al (2007) investigated five mechanisms for making predictions. Although the experts made more correct forecasts than laypeople the recognition heuristic accounted for 90% of all predictions made – the recognition heuristic was found to be an effective way to make predictions based on little information.
In the case of tennis, the recognition heuristic has also been investigated (Serwe & Frings, 2007; Scheibehenne & Arndt, 2007). In one study using data from the 2003 Wimbledon tournament 90% of predictions made when a recognised player was against an unrecognised player relied on the recognition heuristic to make predictions (Serwe & Frings, 2007). The final study used data from the 2005 Wimbledon tournament (Scheibehenne & Arndt, 2007). Amateur tennis players and laypeople were asked about which tennis players they recognised. Laypeople recognised only 11 or the 128 players (9%) on average, whilst amateur tennis players recognised 49 players (39%) on average. Here the recognition heuristic accounted for 70% of all predictions made about the winner of a tennis match regardless of whether the person making the prediction was an amateur tennis player or layperson.
As we have seen the recognition heuristic accounts much of the predictions made when forecasting the outcome of sporting events. Like Stuart betting for the first time many of us use the recognition heuristic whether we know about the heuristic or not. Although at first thought making predictions based on very little information seems like a bad idea the recognition heuristic can often help in making the correct predictions, after all, familiar teams are normally the most successful teams. So, if you are placing a small bet on a sporting event and you are a layperson (not an expert in the sport) then it is likely that this heuristic will be very useful tool in the decision-making toolbox.