A dynamic mechanism for time-constrained trading

The market mechanism design is important for both conventional and electronic commerce as it affects the fairness and efficiency of trading. In this paper, we propose a new market mechanism for time-constrained trading. Our mechanism mimics the traditional brokering system, where buyers and sellers meet together and negotiate through an electronic marketplace. At each time step, agents are paired up for negotiating a deal. We consider that buyer and seller agents are associated with a time constraint and a search cost, and thus must decide promptly and judiciously whether to accept an offer. One distinguishing feature of our mechanism is that the marketplace provides buyer and seller agents with statistical information about the goods. Specifically, the statistics include the probability distributions of obtaining and losing a particular offer. With this information, the agents' decision can be viewed as a Markov decision process, and the optimal (dominant) trading strategy can be computed. In other word s, buyer and seller agents can make trading decisions that maximize their expected utility without the need of speculating others' actions. We also explain how such a dominant strategy can be computed in an efficient manner. Another distinguishing feature of the proposed mechanism is that the statistics are updated continuously and hence the dominant trading strategy is adaptive. Experimental results verify that our mechanism facilitates a fair allocation to the trading agents under various demand and supply conditions.