Predicting Malaysian stock market with relative strength index and moving average convergence-divergence indicators using long short-term memory

Abstract

Some Malaysians hold stocks or mutual funds, and majority of them are new investors. Bursa Malaysia’s CEO stated that based on stock exchange data up to 2022, there are 80.8 million units of odd lot shares in Central Depository System (CDS) accounts, which has been expanding over the years. However, many people still need to learn how to use technical indicators for investing. Even though many researchers have used the Relative Strength Index (RSI) and Moving Average Convergence-Divergence (MACD) on other stock markets throughout the globe, there are limited studies that have been utilizing both indicators in the Malaysian stock market. This study mainly focuses on developing predictive modeling for individual investors, specifically RSI and MACD indicators on conventional Malaysian stock. Our research hypothesis is that the machine learning (ML) approach, the Long Short-Term Memory (LSTM) technique, can predict the Malaysian stock market. By combining the RSI and MACD indicators with LSTM, this study explores the possibilities of timing (buy and sell signals) in Malaysian stock market. This study used daily data gathered from the KLCI Index from 2011 until 2021. Performance metrics, including mean return, standard deviation risk ratio, and Sharpe ratio, were calculated, and hypothesis testing was conducted using t-tests on mean return and F-tests for risk comparison. The empirical data from the Malaysian market demonstrated that the RSI and MACD indicators combined with ML provide a considerable excess return. This strategy is applicable to Malaysia's conventional stock market since it generates an excess profit. The results show that MACD-LSTM provides more profitable trading than RSI-LSTM.