Bidding Strategy based on Market Dynamics

Abstract

The advent of e-commerce has brought about a radical change in the process of auctions. This has given bidders flexibility to try and get best possible deals by participating in multiple auctions which are happening for the same item in overlapping times. Such scenario will change the bidding strategies being used by humans or agents. One of the most important points for deciding the strategy of bidding in multiple auctions is to find out the expected end price of the auctions that is based on the market dynamics which changes on various dynamic parameters such as number of bidders, current price, and time left for closure of auction etc. In this paper, learning of market dynamics is proposed by using genetic algorithm. A function is evolved based on environmental parameters which will take these parameters along with current bid value as input and returns the expected end price. The performance analysis of the expected end price is done for the solution proposed. The required data is generated by simulating an electronic marketplace environment which consists of multiple overlapping auctions of the same item with multiple bidders participating in them. The total number of bidders are not constant but increase in the environment based on defined bidder rate. The performance of the proposed prediction method is measured by comparing predicted end price with the actual end price. This measurement takes place at predefined time intervals and at different behaviors of dynamic parameters. The results captured during the experiment are empirically evaluated through statistical models like multivariate ANOVA (Analysis of Variance) and Independent Samples t Test.