The market for climate forecasts suffers from three major challenges: the fragmentation of relevant expertise, the information asymmetry inherent in long-range prediction when forecaster track records are not available, and the circularity of the interdependence between physical climate predictions and future greenhouse gas concentrations. These three issues can be simultaneously and elegantly addressed using prediction markets. Prediction markets are similar to financial futures markets, or even sports betting, but instead of being to transfer risks, or for entertainment, they are specifically designed to elicit and aggregate disparate information. Prediction markets have been used to predict things such as the outcomes of elections and the reliability of psychological research but attempts to use them for predicting climate have met with mixed results due to the low liquidity typically encountered by markets for specialized topics. The liquidity problem can be tackled using automated market making algorithms and this approach will be illustrated with examples of successful climate-related prediction markets. How such markets can be used for long-range climate prediction, addressing the problem of circularity, will also be explained. An upcoming demonstration market for predicting Atlantic hurricane activity, in which university teams are invited to take part, will be introduced.