OPEC Crude Oil Price Prediction Based on Chaos Theory and GMDH-GA

The price of crude oil is exposed to various factors that cause random, sudden, and chaotic price fluctuations. Accurate forecasting of oil price has a central impact on the macro economy. The aim of this study is to predict the fluctuations of Organization of Petroleum Exporting Countries (OPEC) crude oil in the long-term using the chaos theory and the GMDH-GA algorithm. First, the daily oil price time series is decomposed by wavelet transformation. Then, chaos is tested using the embedding dimension, the Lyapunov power, and GA tests. Finally, time series noises are reduced by reconstructing the wavelet phase space. Three nonlinear models, namely the GMDH-GA model, the GMDH-GA wavelet model, and the GMDH-GA extended model, were used to forecast time series. Although the results showed that all three models were favorable in terms of the root mean square error (RMSE) and the correlation coefficient, the developed GMDH-GA neural network model with a low RMSE and high correlation coefficient was most effective in predicting the daily price of OPEC crude oil.