@article{ART003169013},
author={Jean-Ho Kim and Eun-Hong Park and Ha Young Kim},
title={Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={1},
pages={41-51}
TY - JOUR
AU - Jean-Ho Kim
AU - Eun-Hong Park
AU - Ha Young Kim
TI - Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 1
PB - The Korean Society Of Computer And Information
SP - 41
EP - 51
SN - 1598-849X
AB - The Korea Composite Stock Price Index exhibits high sensitivity to global market conditions, as net exports form a conerstone of the economic structure in Korea. However, most existing studies on stock price prediction limited in two key aspects: First, they primarily focus on indicators within the domestic market, thereby omitted the influence of global market dynamics. Second, they lack of thorough consideration of feature selection methods, which hinders performance and efficiency of deep prediction models in data-hungry financial time series. In this paper, we propose extended stock data that integrates stock market indicators from major export destinations along with macroeconomic variables and Filter-DQN, a novel feature selection method that incorporates mutual information theory and reinforcement learning approach. The proposed method enhance both prediction performance and computational efficiency by effectively filtering out redundant variables.
Experimental results demonstrate that the Filter-DQN method improved performance of baseline models by 8.28%, 45.28%, compared to full variables and naive mutual information based selection, respectively.
Futhermore, the reduced dimensionality of variables leads to 2.6% reduction in computational cost.
KW - Feature selection;Filter methods;Reinforcement Learning;Stock price prediction;Deep Learning
DO -
UR -
ER -
Jean-Ho Kim, Eun-Hong Park and Ha Young Kim. (2025). Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework. Journal of The Korea Society of Computer and Information, 30(1), 41-51.
Jean-Ho Kim, Eun-Hong Park and Ha Young Kim. 2025, "Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework", Journal of The Korea Society of Computer and Information, vol.30, no.1 pp.41-51.
Jean-Ho Kim, Eun-Hong Park, Ha Young Kim "Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework" Journal of The Korea Society of Computer and Information 30.1 pp.41-51 (2025) : 41.
Jean-Ho Kim, Eun-Hong Park, Ha Young Kim. Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework. 2025; 30(1), 41-51.
Jean-Ho Kim, Eun-Hong Park and Ha Young Kim. "Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework" Journal of The Korea Society of Computer and Information 30, no.1 (2025) : 41-51.
Jean-Ho Kim; Eun-Hong Park; Ha Young Kim. Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework. Journal of The Korea Society of Computer and Information, 30(1), 41-51.
Jean-Ho Kim; Eun-Hong Park; Ha Young Kim. Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework. Journal of The Korea Society of Computer and Information. 2025; 30(1) 41-51.
Jean-Ho Kim, Eun-Hong Park, Ha Young Kim. Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework. 2025; 30(1), 41-51.
Jean-Ho Kim, Eun-Hong Park and Ha Young Kim. "Optimizing Stock Price Prediction via Macro context and Filter-DQN Framework" Journal of The Korea Society of Computer and Information 30, no.1 (2025) : 41-51.
