Earticle

초록

영어

This study employs a machine learning-based approach to identify and predict stock market trends using a convolutional neural network (CNN) to the Korean stock market. Building on the methodology introduced in Jiang et al. (2023), we transform historical price and volume data into chart images and utilize CNN to extract patterns predictive of stock returns. Our findings demonstrate that this image-based model can predict the future returns, also in the Korean stock market. Notably, we observe high short-term predictive accuracy, particularly over weekly horizons, which facilitates profitable investment strategies. This study represents the first application of a chart image-based deep learning model to the Korean stock market, contributing new insights into the potential of deep learning techniques in financial market predictability.

목차

Abstract
1. Introduction
2. Data and methodology
2.1. Data
2.2. Convolution Neural Network (CNN)
2.3. Variables Construction
3. Empirical analysis
3.1. Short-horizon (Weekly) Portfolio Performance
3.2. Long-horizon (Monthly/Quarterly) Portfolio Performance
4. Additional Analysis
4.1. Correlation with Other Return Predictors?
4.2. Logistic Regression
4.3. Effect of Stock Size
5. Conclusion
References
Appendix
Figure
Table

키워드

저자

• Jin-Gyu Jeong [ 정진규 | College of Business, KAIST, 85 Hoegiro, Dongdaemoon-gu, Seoul ]
• Suk-Joon Byun [ 변석준 | College of Business, KAIST, 85 Hoegiro, Dongdaemoon-gu, Seoul ]
• Donghoon Kim [ 김동훈 | College of Business, KAIST, 85 Hoegiro, Dongdaemoon-gu, Seoul ] Corresponding Author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  한국재무학회 학술대회

• 간기

  부정기

• 수록기간

  2006~2024

• 십진분류

  KDC 325 DDC 330