Abstrakt
Over the last decades, there has been a marked increase in the interest in prediction and betting markets. This interest was driven by their demonstrated capacity to aggregate information and to improve on the more traditional forecasting methods.
In the first chapter, we analyze key differences between prediction and betting markets and make conjectures about the effect of these different characteristics on markets' performance. In the second chapter, we replicate the betting market experiment, cited in Plott, Wit & Yang (2003), to analyze the process of information aggregation.
Based on observed market odds, Plott, Wit & Yang (2003) show that information was aggregated on their market. On the other hand, model based only on the use of private information fits traders' behavior best.
In contrast to aggregate level data analysis employed in Plott, Wit & Yang (2003), we analyze individual level data and explain these paradoxical results. Finally, we conducted a CERGE-EI internal prediction market to take a well- known and experimentally tested information aggregation mechanism and implement it in a real institution for internal use to examine if it could work in this particular setting.
In the third chapter, we report on the results of this market and we show that for the performance of the market, design and implementation details matter greatly, especially in the case of internal small scale prediction markets.