CAREER: Foundataions of Markets as Information Aggregation Mechanisms

职业生涯：作为信息聚合机制的市场基础预测市场是为信息聚合而设计的市场。为了实现其目标，预测市场提供了一种合同，其未来收益与感兴趣事件的结果挂钩。市场参与者通过交易合约来表达他们对事件的信息。因此，市场价格潜在地包含了所有参与者的信息，并大致代表了对该事件的共识预测。预测市场在实践中已显示出作为高效信息聚合工具的巨大潜力。然而，它们可能会受到操纵。参与者有动机谎报自己的信息，以便在市场上获取更多利润或在市场外获得回报。通过操纵，信息聚合可能会失败，市场价格的可信度也会受到质疑。另一个挑战是运营组合预测市场的计算问题。当机制变得更具表现力，参与者可以披露他们关于结果组合的信息时，市场运行过程可能会变得难以计算，这使得这种组合预测市场不切实际。这个项目的目标是对由于参与者的自身利益和表现力的需要而在预测市场设计中出现的问题进行基础性研究。以计算机科学理论和博弈论为主要研究方法，陈博士将为预测市场建立战略和计算基础，从而设计出理论上合理且实际适用于复杂现实世界环境中的信息聚合的市场机制。该项目围绕两个主题：抗操纵预测市场和计算高效的组合预测市场。对抗操纵预测市场的研究考察了操纵何时发生，操纵如何影响信息聚集，以及如何设计预测市场机制来明确控制操纵的影响。它将为最根本的问题提供答案，即人们是否可以信任预测市场传达的信息，并为信息聚合设计强大的市场机制。对计算有效的组合预测市场的研究考察了组合预测市场的表达能力和计算复杂性之间的权衡，设计了易于处理的组合预测市场，并为难以处理的组合预测市场开发了高效的逼近算法。它将有助于将组合预测市场从理论上的人工制品转移到实际设备上，以聚合更丰富的信息。此项目不依赖于任何特定的应用程序域。因此，本项目中设计的机制可以应用于许多环境中的信息聚合，包括企业、政府和社会。该项目的结果还将有助于为可能对预测市场进行监管的政策制定提供信息。

CAREER: Foundations of Markets as Information Aggregation MechanismsPrediction markets are markets designed for information aggregation. To achieve its goal, a prediction market offers a contract whose future payoff is tied to outcomes of an event of interest. Participants of the market express their information about the event through trading the contract. The market price hence potentially incorporates the information of all participants and approximately represents a consensus forecast for the event. Prediction markets have shown great potential as highly effective information aggregation tools in practice. However, they may be subject to manipulation. Participants have incentives to lie about their information in order to seize more profit in the market or be rewarded outside the market. With manipulation, information aggregation may fail and the credibility of market prices is put into question. Another challenge is the computational problem of operating combinatorial prediction markets. When the mechanism becomes more expressive and participants can reveal their information on combinations of outcomes, the process of operating a market may become computationally intractable, which makes such combinatorial prediction markets impractical to use. The goal of this project is to perform a foundational study on issues that arise in prediction market designbecause of the self-interest of participants and the need of expressiveness. Using computer science theory and game theory as the main research approaches, Dr. Chen will establish the strategic and computational foundations for prediction markets so as to design market mechanisms that are not only theoretically sound but also practically applicable for information aggregation in complex real-world settings.This project is centered around two themes: manipulation-resistant prediction markets and computationally efficient combinatorial prediction markets. The research on manipulation-resistant prediction markets investigates when manipulation arises, how manipulation affects information aggregation, and how to design prediction market mechanisms to explicitly control the impact of manipulation. It will provide answers to the most fundamental question of whether people can trust theinformation conveyed by prediction markets and design robust market mechanisms for information aggregation. The research on computationally efficient combinatorial prediction markets examines the tradeoff between expressiveness and computational complexity of combinatorial prediction markets, designs tractable combinatorial prediction markets, and develops efficient approximation algorithms for intractable combinatorial prediction markets. It will help to transfer combinatorial prediction markets from theoretical artifacts to practical devices for aggregating richer information. This project does not depend on any specific application domain. Hence, mechanisms designed in this project can be applied to aggregate information in many settings, including business, government, and society. Results of this project will also help to inform policy making on the possible regulation of prediction markets.