Evaluating the machine learning based momentum stock trading strategies with back-testing: An emerging market perspective

As the market participants increase, algorithmic trading has grown in prominence. Because of that, extra considerations are needed to develop a profitable trading strategy. Unlike the stock market of developed economies, emerging markets such as the stock market of Bangladesh is yet to reap the benefits of algorithmic trading using Machine Learning (ML)-based “Backtesting” mechanism to deploy trading strategies in real-time. Hence, by fulfilling the gap this study develops an ML-based trading strategy for Dhaka Stock Exchange (DSEX) which is further validated by the "Backtesting" mechanism. For this experimental study, past 05 years data from DSEX broad index and 49 enlisted companies have been pulled. The study has gone through a two-stage approach namely, the overall market scenario of DSEX and individual company scenario and investment opportunity by employing an ML-based “Backtesting” mechanism. The study finds that the “Backtesting” model performs adequately since the sharp ratio turns out to be 0.50 which represents a market-beating performance if achieved over the long term. In terms of practical implications, the trading applications could integrate the model to generate better strategic insights, while the users/traders would be more confident and forthcoming to trade with better buy-sell strategies. The study also enhances the understanding of the domains of ML-based “Backtesting”, predictive analytics, and investment strategy.