A Comparison of Bidding Strategies for Simultaneous Auctions

Bidding for multiple items or bundles on online auctions raise challenging problems. We assume that an agent has a valuation function that returns its valuation for an arbitrary bundle. In the real world all or most of the items of interest to an agent is not present in a single combinatorial auction. We focus on bidding for multiple items in a set of auctions, each of which sell only a single unit of a particular item. Hence an agent has to bid in multiple auctions to obtain item bundles. While an optimal bidding strategy is known when bidding in sequential auctions, only suboptimal strategies are available when bidding for items sold in auctions running simultaneously. We investigate a hill-climbing bidding strategy, which is optimal given an infinite number of restarts, to decide on an agent's bid for simultaneous auctions. We provide a comparison of this algorithm with existing ones, both in terms of utilities generated and computation time, along with a discussion of the strengths and weaknesses of these strategies.