AI Seminar

A Strategic Model for Information Markets

Rahul Sami, Assistant Professor, UM School of Information
SHARE:

Abstract:

Information markets, which are markets especially designed to aggregate traders' information, are becoming increasingly popular as a means for forecasting future events. Recently, researchers have developed two new market forms specifically for efficient information
aggregation: market scoring rules and dynamic parimutuel markets. In this talk, I will describe a new abstract game, the segment game, that enables strategic analysis of both of these market forms. The segment game is tractable to analyze, and has a simple geometric visualization that makes the strategic moves and interactions more transparent. We prove that the segment game can serve as a strategic model of both dynamic parimutuel markets and market scoring rules. This shows a surprising connection between these two market forms, and allows us to characterize optimal strategies in dynamic parimutuel markets; it is also a useful tool in designing new market forms.

This talk covers joint work with Evdokia Nikolova (MIT).

Rahul Sami is an Assistant Professor at the University of Michigan School of Information. He received his Ph.D. in Computer Science from Yale University in 2003, and spent two years as a postdoctoral assistant at the MIT Computer Science and Artificial Intelligence Laboratory. His research is on designing and analyzing incentive mechanisms, markets and reputation systems to enable self-interested parties to cooperatively solve common goals, especially in Internet protocols and applications.

Sponsored by

TOYOTA AI SEMINAR SERIES