Forecasting Stock Prices via Deep Learning During COVID-19: A Case Study from an Emerging Economy

Abstract

In this study we apply a Deep Learning Technique to predict stock prices for the 30 stocks that compose the BIST30, Turkish Stock Market Index before and after the onset of Covid-19 crises. Specifically, we utilize the Bi-Directional Long-Short Term Memory (BiLSTM) model which is a variation of the Long-Short-Term Memory (LSTM) model to predict stock prices for the BIST30 stocks. We compare the performance of the model to other commonly used machine learning models like decision tree, bagging, random forest, adaptive boosting (Adaboost), gradient boosting, and eXtreme gradient boosting (XGBoost), artificial neural networks (ANN), and other deep Leaning models like recurrent neural network (RNN), and the Long-Short-Term Memory (LSTM) model. The BiLSTM model seems to have better performance compared to conventional models used for predicting stock prices and continues to have superior performance in the Covid19 period. The LSTM model seems to have a good overall performance and is the next best model.