Opinion Dynamics Explain Price Formation in Prediction Markets

Area of Science:

• Computational Social Science
• Economic Modeling
• Network Dynamics

Background:

• Prediction markets are recognized forecasting tools, yet existing models often oversimplify their complex price dynamics.
• Understanding the intricate mechanisms driving market behavior is crucial for improving forecasting accuracy.

Purpose of the Study:

• To develop and validate a novel prediction market model that integrates social network opinion dynamics.
• To explore how agent interactions and opinion formation influence market price behavior and empirical properties.

Main Methods:

• A computational model was developed where agents with opinions interact within a social network, updating beliefs via the Deffuant model.
• Agents' opinions inform their betting strategies in a simulated prediction market.
• Historical data from the PredictIt exchange platform was utilized for model validation.

Main Results:

• The proposed model successfully replicates key empirical properties of prediction market time series, such as volatility clustering and fat-tailed return distributions.
• Optimal market behavior was observed when agent opinions exhibited a specific level of variance.
• The study demonstrates a novel method for validating opinion dynamics models using real-world prediction market data.

Conclusions:

• Integrating opinion formation dynamics into prediction market models enhances their ability to capture real-world complexities.
• The Deffuant model of opinion dynamics, when applied within a social network context, provides a robust framework for understanding market behavior.
• This research offers a new approach for validating social influence models using empirical financial data from prediction exchanges.