Comparing trading behaviour and profit composition in prediction markets

Prediction markets have established itself as forecasting technique, especially within the IT industry. While the majority of existing studies focuses either on the output of such markets or its design settings, the traders who actually produce the forecasts got only little attention yet. Within this work, we develop a classification scheme for traders of a prediction market that is grounded on both, financial and prediction market literature. Over a period of three years, 127 prediction markets have been observed and its 4.329 traders are separated into seven subgroups (beginners, noise traders, average traders, experts, donkey traders, market makers and superior traders), based on their knowledge, experience and selectivity. We find empirical evidence for the existence of these subgroups and thus for the heterogeneity among the traders. For each of these subgroups, we analyze the trading behaviour and the profit composition.