Adaptive Strategies for Multicriteria Auctions: An Empirical Study

Abstract

In this paper, we propose bidding strategies for conducting automated reverse auctions based on a non-compensatory multicriteria model. We conduct an empirical multiagent study in order to appreciate the relevance of the proposed strategies. In this type of auction and in order to ensure ascending process evolution, the design of such automated systems often uses a fixed bid increment representing the minimum amount by which a bidder must improve on the current best bid. This article suggests adjusting the bid increment as the auction process goes on. To this end, we propose buyer counterproposal which ensures an acceptable solution at any given time. To this end, we refer to the auction context based on the number of remaining suppliers or the remaining time at each process step. We also present and demonstrate some interesting properties of the proposed algorithm. Finally, we provide an empirical study that compares a fixed-increment strategy to our proposed strategies on the basis of a variation of different auction settings.