Improving Traditional Stock Market Prediction Algorithms using Covid-19 Analysis

Jindal, R.; Bansal, N.; Chawla, N.; Singhal, S.; Ieee,

The stock market is an organized body where public companies offer their stocks through initial public offerings and traders buy/sell these stocks so as to obtain profits. It is dynamic and volatile in nature which makes the task of stock market trend prediction a complex problem. In recent times, the COVID-19 pandemic has made this task even harder. With the rising number of COVID-19 cases across the globe, the market has never been more volatile. This has resulted in the poor performance of various traditional trend prediction algorithms because these algorithms do not account for the impact of the pandemic on the stock market trends. The proposed work aims to enhance the stock market prediction ability of various common prediction models by taking into account the factors related to COVID-19. The forecasting techniques analysed are Decision Tree Regressor, Random Forest Regressor and Support Vector Regressor (SVR). Currently the most affected countries by COVID-19 are the United States of America, India and Russia. Therefore we have analysed the prediction performance of various approaches discussed in this paper on S&P 500 Index, Nifty50 Index and RTS Index using Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Results obtained showcase that all the techniques used perform better when the COVID-19 features were included.