An Advanced Approach for Improving the Prediction Accuracy of Natural Gas Price

Abstract

As one of the most important commodity futures, the price forecasting of natural gas futures is of great significance for hedging and risk aversion. This paper mainly focuses on natural gas futures pricing which considers seasonality fluctuations. In order to study this issue, we propose a modified approach called six-factor model, in which the influence of seasonal fluctuations are eliminated in every random factor. Using Monte Carlo method, we first assess and comparative analyze the fitting ability of three-factor model and six-factor model for the out of sample data. It is found that six-factor model has better performance than three-factor model and natural gas futures prices is strongly influenced by winter effect. We then apply the proposed model to predict the price of natural gas futures in the year 2019. It is found that natural gas prices have a weak upward trend in the coming year and are relatively volatile in winter.