Autonomous Agents and Multi-Agent Systems © 2005 Springer Science+Business Media, Inc. Manufactured in The Netherlands. DOI: 10.1007/s10458-005-4948-2 Autonomous Adaptive Agents for Single Seller Sealed 1 Bid Auctions 2 ANTHONY BAGNALL AND IAIN TOFT ajb,it@cmp.uea.ac.uk 3 School of Computing Sciences, University of East Anglia, Norwich, UK 4 Published online: XXX 5 Abstract. In developing open, heterogeneous and distributed multi-agent systems researchers often face a 6 problem of facilitating negotiation and bargaining amongst agents. It is increasingly common to use auction 7 mechanisms for negotiation in multi-agent systems. The choice of auction mechanism and the bidding strat- 8 egy of an agent are of central importance to the success of the agent model. Our aim is to determine the best 9 agent learning algorithm for bidding in a variety of single seller auction structures in both static environments 10 where a known optimal strategy exists and in complex environments where the optimal strategy may be con- 11 stantly changing. In this paper we present a model of single seller auctions and describe three adaptive agent 12 algorithms to learn strategies through repeated competition. We experiment in a range of auction environments 13 of increasing complexity to determine how well each agent performs, in relation to an optimal strategy in cases 14 where one can be deduced, or in relation to each other in other cases. We find that, with a uniform value distri- 15 bution, a purely reactive agent based on Cliff’s ZIP algorithm for continuous double auctions (CDA) performs 16 well, although is outperformed in some cases by a memory based agent based on the Gjerstad Dickhaut agent 17 for CDA. 18 Keywords: adaptive agents, auctions, zero intelligence plus. 19 1. Introduction 20 The dramatic increase in the quantity of goods and services sold via auctions has fuelled 21 greater interest in the study of protocols for auction structure and strategies for agent bid- 22 ding. The potential for software agents in e-commerce, and in auctions in particular, is 23 enormous [20, 52]. Agent systems can be employed as a practical mechanism by which 24 individuals and companies may more usefully engage in online commercial activity. The 25 applications for autonomous, adaptive agents that can compete and learn effectively in 26 real time online auctions are numerous. For example, Market Based Control (MBC) [8] 27 systems have been applied to a variety of applications, including air conditioning [24], 28 network bandwidth, telecommunications [12] and Advanced Life Support Systems (ALS) 29 [34]. Agent simulations may also serve as a theoretical economic testbed to study the effect 30 of alternative market mechanisms on competitor behaviour, as demonstrated by Grossklags 31 et al. [16], Farmer et al. [11] and Brewer et al. [6]. 32 The majority of adaptive agent research in simulated auctions has focused on algo- 33 rithms for bidding in double auctions, i.e. auctions with multiple buyers and sellers 34 [9, 13, 20, 46]. Double auctions and particularly continuous double auctions (CDA) 35 are an economic mechanism known to be very efficient at allocating resources [42] 36 and are widely used in online and offline markets. Agents for simulated CDA can 37 Journal: AGNT MS.: PIPS: NO00004948 TYPESET  DISK LE  CP Disp.: 8/10/2005 Pages: 34