Stock Market Prediction of Bangladesh Using Multivariate LSTM with Sentiment Identification

Abstract

Stock market prediction is always challenging due to its volatile and dynamic movement. Apart from the technical factors, many external factors make it more difficult to predict the stock market of a developing country like Bangladesh. Therefore, it is not possible to accurately predict the stock market of Bangladesh by taking only the technical factors into consideration. Various studies have shown that some external factors like news sentiment, inflation, Gross Domestic Product (GDP), exchange rate, interest rate, and current balance of the country can affect the stock market trend, which is also applicable to Bangladesh. The main objective of this paper is to predict the trend of Dhaka Stock Exchange (DSEX), the largest stock market in Bangladesh by taking into account the technical stock market data as well as those appropriate external factors. This paper also compared the difference between the trend prediction with and without using news sentiment. All the technical and external stock market data from 2014 to 2021 is collected from verified sources. A multivariate Long Short-Term Memory (LSTM) neural network is used to predict the stock market trend. The experimental results indicate that news sentiment provides better performance in LSTM stock market trend prediction.