An experiment on prediction markets in science.

Area of Science:

• Decision Analysis
• Behavioral Economics
• Information Aggregation

Background:

• Prediction markets aggregate dispersed information and incentivize knowledge acquisition.
• Integrating these markets into scientific publication practices requires investigation.
• Existing research highlights the potential of prediction markets in forecasting.

Purpose of the Study:

• To examine the compatibility of prediction markets with scientific publication workflows.
• To evaluate different information disclosure settings within prediction markets for scientific research.
• To assess the impact of subsidizing market makers on prediction market performance.

Main Methods:

• An experiment was conducted across three information disclosure settings: fully public, fully private, and initially private then public.
• An automated, subsidizing market maker was employed to enhance trading and liquidity.
• Market performance was measured and compared across the different settings.

Main Results:

• The setting where information was initially private and then publicly disclosed demonstrated superior performance.
• Market performance in the hybrid setting matched that of fully public information and surpassed fully private information.
• Participants could leverage information advantages in the hybrid setting, unlike in the fully public setting.

Conclusions:

• A hybrid information disclosure model (initially private, then public) optimizes prediction market functionality for scientific research.
• The use of subsidizing market makers is advantageous for integrating prediction markets into scientific practice.
• Aligning prediction markets with current scientific publication norms enhances their utility and effectiveness.