ScholarWorks@숙명여자대학교

초록

A popular and lucrative area of research has al-ways been stock prediction. Stock prediction using traditional deep learning has been proven to provide better accuracy and returns. However, as artificial intelligence developed, the idea of reinforcement learning (RL) emerged. The rise of RL in the financial markets is fueled by a number of benefits that are specific to this area of artificial intelligence (AI). RL, in particular, enables the combination of the 'prediction' and 'portfolio construction' activities into a single integrated step, allowing machine learning challenges to be precisely customized to investors' objectives. Conveniently, significant limitations like transaction costs, market liquidity, and investor risk aversion can be considered simultaneously. Despite the fact that supervised learning techniques continue to receive the majority of attention, the RL research community has achieved great strides in the financial field during the previous several years. This paper introduces the overall concepts and applications of RL and stock prediction. Additionally, current technology trends are presented based on several application domains. In summary, this paper explores RL-based research trends in the field of stock prediction and makes suggestions for future research avenues.