Academic Research

Yiling Protocol is built on peer-reviewed academic research from Harvard University.

Primary Paper

Self-Resolving Prediction Markets for Unverifiable Outcomes Siddarth Srinivasan, Ezra Karger, Yiling Chen Harvard University — Published at ACM Conference on Economics and Computation (EC 2025)

• arXiv (v2, Feb 2025)
• ACM Digital Library

Abstract

The paper addresses prediction markets for outcomes that cannot be directly verified. It proposes a mechanism that pays agents the negative cross-entropy between their prediction and that of a carefully chosen reference agent. Markets terminate probabilistically, and the final agent — who observes all prior forecasts — serves as a proxy for ground truth.

Key Results

Theorem 1 (Exponential Decay): When the reference agent observes k independent informational substitutes, the strategic adjustment term diminishes exponentially: |Δ| ≤ (1-δ)^k.

Theorem 2 (Strict Truthfulness): If k exceeds a threshold depending on prior beliefs, signal granularity (τ), and information quality parameters (δ, η), then truthful reporting strictly dominates any deviation.

Theorem 3 (ε-PBE): Without knowledge of τ, the mechanism achieves ε-Perfect Bayesian Equilibrium where the maximum gain from deviation is bounded by 𝒟_η(Δ, y), decreasing exponentially in k.

Core Assumptions

1. Common Knowledge of Rationality — agents are risk-neutral Bayesian reasoners
2. Common Prior — shared prior distribution over outcomes and signals
3. Stochastic Relevance — each distinct signal induces a unique posterior
4. Conditional Independence — agents' signals are independent given the outcome
5. (δ,η)-Informativeness — uniform bounds on signal quality

How Yiling Implements the Theory

Citation

@inproceedings{srinivasan2025self,
  title={Self-Resolving Prediction Markets for Unverifiable Outcomes},
  author={Srinivasan, Siddarth and Karger, Ezra and Chen, Yiling},
  booktitle={Proceedings of the 26th ACM Conference on Economics and Computation},
  year={2025},
  publisher={ACM}
}