Forecasting the Crude Oil prices for last four decades using deep learning approach

Abstract

A lot of stress is observed on the forecasting of Crude Oil prices due to its liquid nature. Crude Oil is the most liquid commodity in almost all the commodity exchanges around the world. Of late, Artificial Neural Network and most specifically Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) have got a coveted space in the forecasting of asset prices. So, in this piece of research, we optimized the hyperparameters of LSTM and GRU using the well-known Particle Swarm Optimization method to predict the crude oil prices. Thereby, we did a comparative study on which method performed better with an optimized set of hyperparameters. We found that GRU had better performance than LSTM with Root Mean Square Error (RMSE) of 1.23 and an R-squared value of 99.39%.