Natural gas, as an important energy commodity, plays a significant role in both industrial and domestic applications and especially electrical generation because it is energy-efficient and has lower carbon dioxide emissions. Therefore, developing models for accurate natural gas price forecasting and direction of price changes is critical because these forecasts should be useful in policy-making decisions covering both supply and demand of natural. In this study, natural gas spot prices in Henry Hub from Jan 7, 1997 to Mar 20, 2012, have been applied. Here, the model that has been selected for forecasting natural gas prices by Box-Jenkins methodology, is ARIMA(1,2,2) which gives reasonable and acceptable forecasts. Its information criterion was 0.4675 and 0.4742 for Akaike and Schwarz criterion and quality forecasting measurements are 0.0928, 0.0727, 2.9156 and 0.0185 for RMSE, MAE, MAPE and Theil-U, respectively. These forecasting measurements can be used in comparison with the other models’ forecast performances in different researches in future. Our results show that most of the time, the actual values of prices are located in tne range of forecasting distance and the forecast series follow the actual series closely.

Keywords