A Revenue Maximizing Strategy Based on Bayesian Analysis of Demand Dynamics

For firms in an oligopoly service network, demand learning based dynamics pricing is an efficient way to maximize their revenues. This paper introduces a Bayesian method to learn demand behavior from the perspective of game-theoretic dynamics, where non-parametric techniques for nonlinear time series are incorporated, such that stringent parametric assumptions are removed. We determine the unknown quantities in our demand learning model from historical market data through a Markov chain Monte Carlo (MCMC) algorithm. Based on the calibrated model, how future demands respond to prices can be predicted, and the optimal pricing policy for the next planning period is obtained by a heuristic optimization algorithm. Simulated examples show that our new method is efficient for integrating demand learning and dynamic pricing, and capable of discovering a wide range of demand dynamics.