Forecasting Model for Crude Oil Price Using Artificial Neural Networks and Commodity Futures Prices

TL;DR

This study develops a neural network-based model to predict short-term crude oil price movements, demonstrating that incorporating futures prices improves forecast accuracy and provides insights for risk management.

Contribution

The paper introduces an optimized neural network structure and a data preprocessing approach for short-term crude oil price forecasting using futures data.

Findings

Dynamic 13-lag model is optimal for prediction.

Forecast accuracy is 78% for one day ahead.

Futures prices contain valuable information for short-term forecasts.

Abstract

This paper presents a model based on multilayer feedforward neural network to forecast crude oil spot price direction in the short-term, up to three days ahead. A great deal of attention was paid on finding the optimal ANN model structure. In addition, several methods of data pre-processing were tested. Our approach is to create a benchmark based on lagged value of pre-processed spot price, then add pre-processed futures prices for 1, 2, 3,and four months to maturity, one by one and also altogether. The results on the benchmark suggest that a dynamic model of 13 lags is the optimal to forecast spot price direction for the short-term. Further, the forecast accuracy of the direction of the market was 78%, 66%, and 53% for one, two, and three days in future conclusively. For all the experiments, that include futures data as an input, the results show that on the short-term, futures prices do hold new information on the spot price direction. The results obtained will generate comprehensive understanding of the crude oil dynamic which help investors and individuals for risk managements.