Geopolitical risk (GPR) has emerged as a pivotal macro-financial factor with significant implications for asset pricing, especially in emerging markets characterized by institutional weaknesses and market inefficiencies. Recent advances in machine learning (ML) offer new opportunities to model complex, nonlinear relationships between macroeconomic risks and asset returns. This study examines whether GPR is a priced risk factor in Vietnam’s equity market and evaluates the efficacy of Support Vector Regression (SVR) in capturing related return anomalies. Using monthly data from 2009 to 2024, we construct a long-short portfolio strategy based on SVR forecasts of excess stock returns conditioned on GPR and Fama–French risk factors. The portfolio delivers statistically and economically significant returns, with the abnormal performance (P11) remaining robust across traditional models such as CAPM and FF3. Robustness checks confirm the persistence of the anomaly under alternative rolling windows, stock filters, and SVR specifications. The results challenge the Efficient Market Hypothesis in frontier markets and highlight the relevance of GPR as a priced factor. Methodologically, the findings underscore the advantage of ML-based nonlinear models in empirical asset pricing. Overall, this research contributes to the literature by integrating geopolitical risk and machine learning to uncover predictive patterns in emerging market returns.