Managing the uncertainties in commodity trading: A Bayesian software implementation

Abstract

A decision support software tool is developed to aid commodity traders with the management of uncertainty. A methodology to control the risk of losses is examined and this is likely to be highly desirable for practical use in the financial markets. Bayesian techniques are used to forecast future рrices and this is combined with the optimization of a utility function to support a trader's buy and sell decisions. The difficulties of eliciting both a prior belief and a utility function are discussed. The effectiveness of the tool is explored using a practical application of the Design of Experiments (using simulated prices) and also by testing the software with traders. The trials of the tool with traders received positive feedback.