This study examines how the determinants of the political risk factor affect the forecasting performance of the United Arab Emirates’ stock market during the COVID-19 pandemic. The empirical investigations of this goal are conducted through using new machine learning models including a linear regression, an artificial neural network, a random forest, an extreme gradient boosting (XGBoost), and a light gradient boosting (LightGBM). We also use a game theory-based model the SHapley Additive explanation (SHAP) interpretation framework to evaluate the most important features for predicting the UAE’s stock market prices. The experimental results show that the LightGBM and XGBoost outperform the traditional machine learning models such as the linear regression and produce a holistic probability distribution over the entire outcome space, which helps quantify the uncertainties related to the effect of the COVID-19 pandemic on predicting the UAE’s stock market. The novel SHAP algorithm also provides insights in interpreting the complex “black box” architecture of the machine learning models to help predict this country’s stock prices. The results provide important implications for the political risk management in periods akin to the COVID-19 pandemic.