Risk-Based Evolutionary Bidding Strategy for Online Multiple Auctions

Many empirical studies have proofed that the agents which without programmed to adjust their bidding behaviors adaptively perform poorly in complex e-market circumstance. The paper proposed a risk-based evolutionary (RBE) bidding strategy for agents participating in multiple auctions, which is flexible and adaptive to the changing environment of online auctions. Two models were developed to determine the optimize values for different parameters of the strategy. The first one is a risk-based bidding strategy model constructed by combining two tactic functions, in which some of behavior-relative parameters adjusted adaptively according to the change of agent's risk attitude. The second one is an evolutionary model to searching for the optimal values for other parameters of the strategy. A real-valued coding genetic algorithm was proposed which shows effective searching path and rapid convergence rate. Contrasted to the other bidding strategies proposed in previous works, the (RBE) bidding strategy can adjust bidding behaviors adaptively and rapidly according to the change of the market circumstance, and perform effectively in the dynamically changing environment of multiple online auctions.