Crude Oil Futures Price Prediction Using a Hybrid CNN-LSTM Neural Network

Abstract

Crude oil futures prices are the core benchmark for the global energy market and macroeconomy. Their fluctuations are highly nonlinear, non-stationary, and time-dependent. These fluctuations are driven by a complex interplay of factors, including crude oil spot prices, commercial inventories, the US dollar index, market sentiment, and geopolitical events. Traditional econometric models and standalone neural networks struggle to capture these dynamic patterns accurately. They consistently suffer from two key limitations: low prediction accuracy and poor robustness. To address these limitations, a hybrid CNN-LSTM forecasting model is proposed. The CNN component extracts local spatial features from price data. The LSTM component captures long-term temporal dependencies in the time series. The proposed model is validated using actual WTI crude oil futures data. The dataset spans from 2014 to 2026. It is divided into three distinct market scenarios: stable, volatile, and extreme. Comparative experiments are conducted against a standalone LSTM baseline model across all three scenarios. The results show that the proposed hybrid model outperforms the baseline in all market scenarios. It delivers an overall prediction accuracy improvement of approximately 4%. This study provides an effective approach to improve the accuracy and robustness of crude oil futures price forecasting.