Future of Work and Digital Management Journal 4 Serial Number 11 2026 01 01 The Impact of Geopolitical Factors on Oil Market Risk Prediction Using a Machine Learning Approach The Impact of Geopolitical Factors on Oil Market Risk Prediction Using a Machine Learning Approach 1 16 EN Journal Article 2025 07 10 <table> <tbody> <tr> <td> <p>The global oil market, as one of the key pillars of the international economy, is influenced by complex geopolitical factors that make risk prediction a critical challenge. This study investigates the impact of geopolitical factors on oil market risk and develops a machine learning–based model to improve prediction accuracy. Daily time-series data of West Texas Intermediate (WTI) crude oil prices and geopolitical indices—including the Geopolitical Risk Index (GPRD), geopolitical acts (GPRD_ACT), and threats (GPRD_THREAT)—from May 5, 2014, to April 26, 2024, were analyzed. First, multiple linear regression revealed that geopolitical acts have a positive and significant effect on oil price volatility; however, limitations such as residual autocorrelation and non-normality reduced the model’s efficiency. Subsequently, four machine learning models—Random Forest (RF), Support Vector Regression (SVR), Decision Tree (DT), and Artificial Neural Network (ANN)—were trained. Among them, RF exhibited superior performance, achieving the lowest error in the test set (MAE: 0.005011, RMSE: 0.006188). Using the RF model, the conditional standard deviation was estimated to calculate the Value at Risk (VaR) at a 95% confidence level, and backtesting with the Kupiec and Christoffersen tests confirmed its accuracy. A comparative analysis with the GARCH model demonstrated the superiority of RF, supported by a higher Lopez statistic (4080.745 vs. 4033.800). These findings highlight the critical role of real geopolitical events in oil market volatility and show the advantage of machine learning in modeling nonlinear market dynamics. This study presents a novel framework for analyzing oil market risk, which can help reduce uncertainty and enhance economic decision-making.</p> </td> </tr> </tbody> </table> https://journalfwdmj.com/index.php/fwdmj/article/download/150/168