A Proposal for the Use of Prediction Markets to Forecast Influenza Activity

We propose that prediction markets may be useful for predicting influenza activity by aggregating the expert opinion of healthcare professionals quickly, accurately, and inexpensively. Experimental markets have been successfully used by the Iowa Electronic Markets (IEM) to predict other outcomes ranging from election results to movie box office receipts. Since 1988, the IEM has developed an impressive prediction record, substantially superior to alternative mechanisms such as polls and surveys. Prediction markets aggregate disparate information from multiple participants who indirectly reveal their personal knowledge when they trade. Available evidence from over 15 years of running experimental markets suggests that prediction markets can outperform both refined statistical forecasting methods and well-designed surveys. In addition, prediction markets could be used for tracking and forecasting other emerging infectious diseases from SARS to avian influenza.