The COVID-19 pandemic has caused turbulence in the world’s financial markets on an unprecedented scale, testing the validity of conventional models that forecast the stock price. Having unrivaled capabilities for processing big volumes of data and complicated relationships, machine learning algorithms have emerged as one of the powerful tools available for forecasting stock prices in the COVID-19 period. The paper aims to provide a review and discuss various models developed using machine learning, specifically Linear Regression, Support Vector Machine (SVM), Decision Trees, Random Forests, Long Short-Term Memory (LSTM), and Reinforcement Learning, and their individual effectiveness regarding stock price prediction in the face of pandemic uncertainty. The performance of the models has ranged between good performance to successful handling of market fluctuations, yet issues with interpretable results and integration of external factors like news and policies have persisted. Expert systems, explainable tools like SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME), Automated Machine Learning (AutoML), and transfer learning can be considered for overcoming the challenges mentioned above and improving the models’ performance in a dynamically changing environment.
