Forecasting crude palm oil price movement through dynamic sampling band based on historical price distribution

This study aims to address the problem where there is no theoretical support for the sample size of data used for forecasting price movement in existing models for forecasting of crude palm oil price movement. To achieve this, an innovative approach known as Dynamic Sampling Band has been proposed in this thesis. It uses the historical price distribution to determine the sample size used to identify price equilibrium for forecasting future price movement. The results of this study have shown that the proposed innovation can help to improve forecasting accuracy.