The relevance of group expertise for the accuracy of a prediction market

This research has focused on the relevance of group expertise with regard to a prediction market. A prediction market is a futures market in which a large group of people can express their opinion about the out coming of a certain event by buying shares from the answer that – according to the participant – is most likely to be correct. Prediction markets are increasingly implemented in enterprise environments. The main goal of this research has been to advise companies that have implemented or are considering implementing a prediction market about which type of participants – with regard to expertise in the specific field – to include in the prediction market environment to get the most accurate results. An experiment has been performed during the Wimbledon competitions of 2009. Two prediction market environments have been set up to be able to compare the accuracy of the predictions of experts versus non-experts. Also, a survey among the (possible) participants of the Wimbledon experiment has been performed, mainly to get a better insight in the decision making process of the two groups. In addition, two data sets in a non-prediction market environment have been analyzed. The first one is a European Championship pool of soccer in 2008 in which the group results of experts were compared with non-experts. The second data set that has been analyzed is regarding the predictions of the AEX-index in the next month and six months. Again, the accuracy of experts was compared with the accuracy of the non-experts. Our main conclusions have been that experts do not significantly make more accurate predictions than non-experts. This implies that all types of participants should be included in an enterprise prediction market. However, it is more difficult to trigger non-experts to participate then it is to trigger the experts. To that respect, it has been suggested to further research how to give non-experts incentives that will lead to a higher overall response rate. Another conclusion has been that there is a difference between the strategy of experts and non-experts while predicting. The effect of the strategic differences between experts and non-experts on the prediction market is another interesting topic that should be further researched.