A copula function approach to infer correlation in prediction markets

Abstract

We propose the use of copula methods to recover the dependence structure between prediction markets. Copula methods are flexible tools to measure associations among probabilities because they encompass both linear and non linear relationship among variables. We apply the proposed methodology to three actual prediction markets, the Saddam Security, the market of oil spot prices and the Saddameter. We find that the Saddam Security is nearly independent of the oil market, while being highly correlated to the Saddameter. The results obtained appear to suggest that the Saddam Security prediction market may be noisy or overlooking some political factors which are instead considered by Saddameter and the oil market. © 2009 ICST Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering.