Development of a Stock Market Data Simulation and a Machine Learning Trading System Using GANs 


Vol. 49,  No. 9, pp. 1264-1273, Sep.  2024
10.7840/kics.2024.49.9.1264


PDF
  Abstract

One of the essential steps in designing a trading system is to build a training model using historical stock data. However, due to the nature of the financial markets, obtaining large amounts of stock price data is a time-consuming and expensive endeavor. The lack of data for model building can lead to various problems such as low generalization ability and poor prediction ability. Therefore, this paper proposes a new methodology for analyzing financial markets using adversarial generative neural network-based virtual data generation simulation. In order to overcome the heavy-tail phenomenon, which is one of the limitations of the geometric Brownian motion model in modern financial theory, we approximate the actual stock return distribution by inputting random numbers extracted from the GAN to simulate future stock price fluctuations that reflect the actual market behavior. Afterward, we trained the machine learning model on the two datasets, the generated data and the real data, and developed a trading strategy to compare the trading performance. The experimental results showed that utilizing the fictitious stock price data generated by the proposed methodology improved overall performance on trading metrics, and these results suggest a new solution to overcome the limitations of financial data. The results confirm the potential of data simulation and machine learning-based trading systems utilizing GANs to improve trading strategies and risk management.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Related Articles
  Cite this article

[IEEE Style]

S. Yoo, J. Jang, J. Kim, "Development of a Stock Market Data Simulation and a Machine Learning Trading System Using GANs," The Journal of Korean Institute of Communications and Information Sciences, vol. 49, no. 9, pp. 1264-1273, 2024. DOI: 10.7840/kics.2024.49.9.1264.

[ACM Style]

Sungju Yoo, Juhyeon Jang, and Jaeyun Kim. 2024. Development of a Stock Market Data Simulation and a Machine Learning Trading System Using GANs. The Journal of Korean Institute of Communications and Information Sciences, 49, 9, (2024), 1264-1273. DOI: 10.7840/kics.2024.49.9.1264.

[KICS Style]

Sungju Yoo, Juhyeon Jang, Jaeyun Kim, "Development of a Stock Market Data Simulation and a Machine Learning Trading System Using GANs," The Journal of Korean Institute of Communications and Information Sciences, vol. 49, no. 9, pp. 1264-1273, 9. 2024. (https://doi.org/10.7840/kics.2024.49.9.1264)
Vol. 49, No. 9 Index

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880