Assessing the performance of different prediction market formats in forecasting tasks

Abstract

Prediction markets have recently gained favour with the academic and business communities. Prediction markets have evolved a long way from their basic beginnings as friendly wagers among friends to become large scale markets connecting traders from around the world. They have been adopted into many large and dynamic corporations that require up to the minute information that can keep up with their business. Organisations like Google, HP, Yahoo! and Best Buy have been experimenting with prediction markets for demand forecasting tasks. Governments have also been using markets, although not always as successfully. The U.S. government looked at PAM which became the terrorist futures market in the post 9/11 world. This did not appeal to the American populous and it has since been withdrawn. Through technological advancements the capabilities and availability of prediction markets has grown. With this the interest in how they work and what can be done to improve the accuracy of the markets. This research looked at the inclusion of a deliberative technique to the markets to improve that accuracy of the market. For this research, markets that made use of discussion boards were used. They were compared against traditional markets, which had no means of communication between traders.The research took the form of a quantitative comparison between the two market types. Data was acquired from the Iowa Electronics Market (IEM) and Inkling Public Markets. The findings from this research indicate that there was a significant difference with α=0.012 for the markets at close. This indicated that there was a significant between the traditional (control) and non-traditional (experimental groups) markets from descriptive statistics it was indicated that the traditional markets performed better in the prediction tasks. The conclusions of this research indicate that allowing traders to communicate and see the actions of others creates group biases which impacts on their independence when making trades and thus on the performance of the market.Copyright