CREST: Cross-Reference to Exchange-based Stock Trend Prediction using Long Short-Term Memory

Abstract

Due to a large number of tradings in the stock market, it generally experiences fluctuations throughout the day. Such oscillations influence market capitals of the companies listed on a stock exchange. Hence, in order to take suitable trading steps, prediction of the future stock price as well as trend direction becomes a crucial task. National Stock Exchange (NSE) and Bombay Stock Exchange (BSE) cover the largest market capitalization in India. A company’s shares are traded according to the stock exchange it is listed on; it can also be listed on multiple exchanges. Various approaches have tried to predict stock markets in terms of future indices, price movement, and returns perspectives, however, analyzing stocks of a company, which is listed on multiple exchanges, has been limited to the financial perspectives. In this article, a cross-reference to exchange-based stock trend (CREST) prediction method is proposed using long short-term memory (LSTM). The daily stock prices of Wipro Limited (WIPRO) company, which is listed on NSE as well as BSE, have been collected and the stock price movement of WIPRO in one exchange has been analyzed for predicting the trend in the other exchange. To identify the applicability of our approach, CREST has also experimented with Infosys Limited and Larsen & Toubro Infotech Limited companies. The performance is evaluated using root-mean-square error and directional accuracy along with precision, recall, and F-measure for the results of all three companies.