Evidencing the Forecasting Performance of Predication Markets: An Empirical Comparative Study

Abstract

Accurately forecasting uncertain outcomes to inform planning processes and aid decision making is a perennial organisational challenge, and the focus of a substantial body of research in management science, information systems and related disciplines. Academic research suggests that prediction markets may be of significant benefit to organisations in meeting this challenge. However most of the empirical studies assessing prediction market performance are laboratory based and suffer from limits to their generalizability. Recent literature has called for research which analyses the performance of prediction markets in ecologically valid settings in order to evidence their effectiveness to potential organisational users. This paper answers these calls by designing a prediction market to forecast an uncertain real world event. The study then compares the forecasting performance of the prediction market with a number of more traditional forecasting approaches regularly used by organisations. The study is contextually situated in a low information heterogeneity problem space, where relevant information is freely available. The results suggest that in this context prediction markets outperform the other forecasting methods studied.