On convergence of forecasts in prediction markets

Paper: arXiv 2402.16345

Authors: Nina Badulina, Dmitry Shatilovich, Mikhail Zhitlukhin

Abstract

We propose a dynamic model of a prediction market in which agents predict the values of a sequence of random vectors. The main result shows that if there are agents who make correct (or asymptotically correct) next-period forecasts, then the aggregated market forecasts converge to the next-period conditional expectations of the random vectors.

Complexity vs Empirical Score

• Math Complexity: 7.5/10
• Empirical Rigor: 2.5/10
• Quadrant: Lab Rats — theoretically deep, empirically untested

Why this score: The paper provides a rigorous mathematical framework and formal convergence proofs regarding aggregated market forecasts in dynamic prediction markets. However, it is primarily theoretical with little to no empirical validation or backtesting on real-world prediction market data.

Research Flowchart

  flowchart TD
    A["Research Goal: Do prediction market forecasts<br>converge to true conditional expectations?"] --> B["Input: Sequence of random vectors &<br>agents with (asymptotically) correct forecasts"]
    B --> C["Dynamic Market Model:<br>Agents trade derivatives over time"]
    C --> D["Stochastic Convergence Analysis:<br>Aggregate forecasts via market mechanism"]
    D --> E["Outcome: Market forecasts converge to<br>next-period true conditional expectations"]