A Study of Crude Oil Price Forecasting Based on Long Short-Term Memory Model

In industrial production, crude oil prices play a pivotal role, influencing economic stability due to their propensity to induce fluctuations. Predicting these prices accurately is thus a crucial task in economics. This study addresses this challenge by employing a Long Short-Term Memory (LSTM) model, trained on data spanning January 2005 to January 2019, to forecast crude oil prices. Compared against expectations from economists, financial markets, and policymakers, the LSTM model demonstrates robust fitting and reliable predictive capability across various time frames. Notably, it outperforms alternative models in terms of accuracy when forecasting crude oil prices over extended periods. However, the model's accuracy diminishes with shorter forecasting intervals, suggesting the need for supplementary variables in shorter-term predictions. Therefore, while the LSTM model proves effective for longer-term forecasts, enhancing its precision for shorter intervals may necessitate integrating additional predictive factors. This research underscores the LSTM model's potential as a valuable tool in navigating the complexities of crude oil price forecasting, contributing insights crucial for informed economic decision-making.