search for


Modular reinforcement learning for dynamic portfolio optimization in the KOSPI market
Journal of the Korean Data & Information Science Society 2021;32:213-26
Published online January 31, 2021;
© 2021 Korean Data and Information Science Society.

Taeyoon Kim1 · Bonggyun Ko2

12Department of Mathematics and Statistics, Chonnam national University
Correspondence to: 1Master course, Department of Mathematics and Statistics, 77, Yongbong-ro, Buk-gu, Gwangju 61186, Korea.
2Corresponding author: Professor, Department of Mathematics and Statistics, 77, Yongbong-ro, Bukgu, Gwangju 61186, Korea. E-mail:

This achievement was supported by National Research Foundation of Korea, funded by the government (Ministry of Science and ICT) (No. 2019R1G1A110070412).
Received December 1, 2020; Revised January 9, 2021; Accepted January 15, 2021.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
In stock investment and asset management, portfolio distribution and optimization are essential parts to manage risk and maximize returns, and have been traditional problems to be solved in the financial sector. Meanwhile, a lot of research have been conducted on deep learning in recent years, and reinforcement learning is also making great progress. Accordingly, attempts have been made to apply the reinforcement learning methodology to portfolio management in recent years, but most of the research is limited to cryptocurrencies with large transactions. In this paper, we implemented a neural network that composes a portfolio through two types of an Evaluation Stock module (ESM) that selects stocks for investment and an Asset Allocation module (AAM) that allocates the selected stocks. The constituent stocks of the KOSPI 200 were considered for investment.
Keywords : Deep learning, KOSPI, portfolio, reinforcement learning, time series.