Adaptive Agents for Sequential Vickrey Auctions with Homogeneous Goods

In this paper, we aim to study the behaviour of the autonomous adaptive agents in sequential Vickrey auctions with homogeneous goods. We describe how to revise these adaptive agents, including ZI-C, ZIP, GD, and AA. The performance of these agents is evaluated by two sets of experiments. It turns out that truth-telling is not a dominant strategy; when it is easy for all the buyer agents to buy their goods, one should not tell his true value; when it is difficult to buy, one should not submit a random value; the performance of adaptive agents, such as ZIP and GD, is very good but cannot be guaranteed to be the best in various market environments. Through another set of experiments, the results demonstrate that AA agents perform very well compared with the truth-telling agents and agents using other adaptive bidding strategies.