Coal Price Prediction Using Financial Indices

The price movement prediction in the futures market is difficult due to fluctuating demands and supplies. This thesis addresses the problem of coal price movement prediction. The study compares two prediction models using two different datasets. The first dataset includes daily trading data, while the second dataset contains both daily trading data and computed financial indices. The data from Indonesia and Australia between 2010 and 2019 is used for the experiment. The experimental results show that the second model achieves higher accuracy. The market simulation also indicates that the second model enjoys a larger trade gain higher than 30% of the budget within a year.