A Tractable Combinatorial Market Maker Using Constraint Generation

We present a new automated market maker for providing liquidity across multiple logically interrelated securities. Our approach lies somewhere between the industry standard—treating related securities as independent and thus not transmitting any information from one security to another—and a full combinatorial market maker for which pricing is computationally intractable. Our market maker, based on convex optimization and constraint generation, is tractable like independent securities yet propagates some information among related securities like a combinatorial market maker, resulting in more complete information aggregation. We prove several favorable properties of our scheme and evaluate its information aggregation performance on survey data involving hundreds of thousands of complex predictions about the 2008 U.S. presidential election.

DudikLaPe12.pdf
PDF file

In  ACM Conference on Electronic Commerce

Details

TypeInproceedings
> Publications > A Tractable Combinatorial Market Maker Using Constraint Generation