Stock Volatility Prediction Using Machine Learning during Covid-19

Abstract

Predicting the stock prices’ movement has been considered a tedious task due to the stock market's volatile behavior. This study is an attempt to predict stock prices based on 15 trading factors by applying deep learning algorithm support vector machine (SVM) on daily price data of three stocks of COVID-19, collected from secondary sources and processed through Python software. The finding of this study suggests that the linear kernel gives 53%, the polynomial kernel gives 46% and the radical basis function (RBF) kernel gives 62% validation accuracy, which shows the RBF kernel predictive ability is highest than others during high volatility. The current study contributes to the stock return literature-originated machine learning algorithms during the unprecedented market condition. The finding of this study is helpful for investors and traders who calculate stock return in their portfolio diversified decisions, regulators, and policymakers in the formulation of their regulations & implementation decisions while the market condition is unprecedented. Thus, the support vector machine (SVM) algorithm offers an accurate prediction of stock return to investors who invest during high volatility.