JCAT: a platform for the TAC market design competition

Auctions, when well designed, result in desirable economic outcomes and have been widely used in solving real-world resource allocation problems, and in structuring stock or futures exchanges. The field of auction mechanism design has drawn much attention in recent years from economists, mathematicians, and computer scientists. In traditional auction theory, auctions are viewed as games of incomplete information and traditional analytic methods from game theory have been successfully applied to some simple types of auctions. However, the assumption of prior common knowledge in the incomplete information approach may not hold in some auctions, and computing analytic solutions may be infeasible in other auctions. Both of these problems hold in the case of continuous double auctions . As a result of these problems, researchers often use computer simulation of auctions in which traders are software agents. Such agents, armed with various learning algorithms and optimization techniques, have been shown to produce outcomes similar to those observed in auctions with human subjects [7]. Indeed, software traders are capable of outperforming human traders [3]. Along with the automation of traders, computer scientists have started to take evolutionary and adaptive approaches to automatically creating auction mechanisms [1, 9, 10]. Although this work has produced promising results, it has one common theme — the only comparisons that are made are indirect. The results from one lone market are compared with those of another lone market. In contrast, not only do traders in an auction compete against each other,