Abstract/Summary

Forecasting election outcomes is a hugely popular activity, and not without reason: outcomes can have significant economic impacts, for example on stock prices. As such, it is economically important, as well as of academic interest, to determine the forecasting methods that have historically performed best. However, forecasts are often incompatible, as some are in terms of vote shares, and others are probabilistic outcome forecasts. In this paper we set out an empirical method for transforming opinion poll vote shares into probabilistic forecasts, and then evaluate the performance of prediction markets and opinion polls. We compare along two dimensions: bias and precision. We find that converted opinion polls perform well in terms of bias, and prediction markets on precision.