MULTIVARIATE METHODS IN ASSESSING THE ACCURACY OF PREDICTION MARKETS EX ANTE BASED ON OHE HIGHEST-PRICE CRITERION

This study successfully establishes the principal component analysis with discriminant analysis (PCA-DA) model to assess the accuracy of contracts in the prediction markets ex ante based on the highest-price criterion. Trained by the xFuture data (7,274 contracts of future events) from 2006-2011, the PCA-DA model shows learning effects and provides 97.72% confidence to predict the outcome of any contract discriminated to the correct prediction group in the Exchange of Future Events. However, we need to greatly improve the low confidence of 19.58% for the PCA-DA model to predict the result of any contract discriminated to the incorrect prediction group.