Consensus and Differences of Opinion in Electronic Prediction Markets

Abstract

Businesses are exploring the use of prediction markets to assist with forecasting. In this study, we focus on the ability of prediction markets to reflect the consensus of trader forecasts as well as the dispersion of these forecasts. Using a real?money, computerized, anonymous double?auction market mechanism, we examine a series of markets forecasting a real?life outcome, i.e., movie box office performance. This continuous outcome was segmented into a small number of mutually exclusive ranges, each associated with a winner?take?all contract. This payoff structure allowed us to determine whether the contract prices reflect the dispersion of the traders' individual forecasts, which were submitted before trading began. We find that these markets do an excellent job revealing the consensus forecast. The market prices for all contracts are consistent with the entire distribution of trader forecasts for most of the markets. In addition, we find that markets with a wider pool of traders tend to result in superior forecasts.Businesses are exploring the use of prediction markets to assist with forecasting. In this study, we focus on the ability of prediction markets to reflect the consensus of trader forecasts as well as the dispersion of these forecasts. Using a real?money, computerized, anonymous double?auction market mechanism, we examine a series of markets forecasting a real?life outcome, i.e., movie box office performance. This continuous outcome was segmented into a small number of mutually exclusive ranges, each associated with a winner?take?all contract. This payoff structure allowed us to determine whether the contract prices reflect the dispersion of the traders' individual forecasts, which were submitted before trading began. We find that these markets do an excellent job revealing the consensus forecast. The market prices for all contracts are consistent with the entire distribution of trader forecasts for most of the markets. In addition, we find that markets with a wider pool of traders tend to result in superior forecasts.