A Neural Network-Based Ensemble Prediction Using PMRS and ECM

Crude oil plays a significant role in the modern society and its price prediction attracts more and more attentions, not only for its importance to the modern industry, but also for its complex price movement. Based on PMRS, ECM and NN, this paper presents an integrated model to forecast crude oil prices. In the proposed model, PMRS is first used to model the trend of crude oil price, and then ECM is offered to establish to forecasting errors. Finally, NN is employed to integrate the results from the ones of PMRS and ECM to make the final forecasting values more accurate and desirable. The WTI spot prices and a set of financial indicators are utilized as inputs for the validation purpose. The empirical results show that the proposed integrated model can significantly improve the forecasting performance, compared with other four forecasting models, and it can be an alternative tool to predict crude oil prices.