A synchronous deep reinforcement learning model for automated multi-stock trading

A Black Swan event-based hybrid model for Indian stock markets' trends prediction

Among all the application areas of the time-series prediction, stock market prediction is the most challenging task due to its dynamic nature, and dependency on many volatile factors. The unpredictable fatal events called Black Swan events also highly influence the stock market. If the successful stock trends prediction is achieved, then the investors can adopt a more appropriate trading strategy, and that can significantly reduce the risk of investment. In this work, a time-efficient hybrid stock trends prediction framework(HSTPF) is proposed to successfully predict the future trends of the stock market even during the periods of Black Swan events. Here, to improve the prediction accuracy of HSTPF, the Black Swan events analysis and features selection operations are performed, and also the performance of various machine learning classifiers are analyzed. A vast number of experiments are conducted on the two real-world stock market datasets S&P BSE SENSEX and Nifty 50, to analyze the performance of the proposed framework. The framework is applied for the single-step and multi-step ahead predictions. The experimental results show that the proposed framework produces over 86% of accuracy, and during the Black Swan events, its accuracy is almost 80% for single-step ahead predictions. For the multi-step ahead of predictions, the HSTPF is produced satisfactory results. The framework also outperforms other existing similar works even during the Black Swan events in terms of prediction accuracy, and its computational time is also very low.