Abstract
Information aggregation mechanisms are designed explicitly for collecting and aggregating dispersed information. Prediction markets represent one of the best examples of how this kind of "wisdom of the crowds" can be used. We use a Twitter-based prediction market to suggest that carefully designed market mechanisms can bring to light trends in dispersed information that improves the accuracy of our predictions. The information system we are developing combines the power of prediction markets with the popularity of Twitter. Simulation results show that our network-embedded prediction market can produce better predictions using information exchange in social networks and can outperform other prediction markets that do not use social networks. We also demonstrate that as cost decreases and more and more agents acquire information, the prediction market prices fully incorporate all available information, and the forecasting performance of the network-embedded prediction market is better.
Recommended Citation
Qiu, Liangfei; Rui, Huaxia; and Whinston, Andrew, "A Twitter-Based Prediction Market: Social Network Approach" (2011). ICIS 2011 Proceedings. 5.
https://aisel.aisnet.org/icis2011/proceedings/economicvalueIS/5
A Twitter-Based Prediction Market: Social Network Approach
Information aggregation mechanisms are designed explicitly for collecting and aggregating dispersed information. Prediction markets represent one of the best examples of how this kind of "wisdom of the crowds" can be used. We use a Twitter-based prediction market to suggest that carefully designed market mechanisms can bring to light trends in dispersed information that improves the accuracy of our predictions. The information system we are developing combines the power of prediction markets with the popularity of Twitter. Simulation results show that our network-embedded prediction market can produce better predictions using information exchange in social networks and can outperform other prediction markets that do not use social networks. We also demonstrate that as cost decreases and more and more agents acquire information, the prediction market prices fully incorporate all available information, and the forecasting performance of the network-embedded prediction market is better.