Research Article
Stock Return Prediction in the Tokyo Stock Exchange using Machine Learning Methods
@INPROCEEDINGS{10.4108/eai.22-5-2026.2365110, author={Yuhaobai Yuan}, title={Stock Return Prediction in the Tokyo Stock Exchange using Machine Learning Methods}, proceedings={Proceedings of the 4th International Conference on Image, Algorithms, and Artificial Intelligence, ICIAAI 2026, 22-24 May 2026, Singapore, Singapore}, publisher={EAI}, proceedings_a={ICIAAI}, year={2026}, month={8}, keywords={Machine learning Stock market prediction Japan Exchange Group LightGBM Sharpe ratio}, doi={10.4108/eai.22-5-2026.2365110} }- Yuhaobai Yuan
Year: 2026
Stock Return Prediction in the Tokyo Stock Exchange using Machine Learning Methods
ICIAAI
EAI
DOI: 10.4108/eai.22-5-2026.2365110
Abstract
This paper studies how three machine learning models work in the Japan Exchange Group (JPX) stock market. The models are Ridge regression, Multi-Layer Perceptron (MLP), and LightGBM. It uses past trading data from 2,000 stocks. The study builds a ranking strategy across stocks. The goal is to increase the daily spread return Sharpe ratio. The results show that the linear model (Ridge) and the MLP model do not produce strong risk-adjusted returns. But the LightGBM model performs better. It reaches the highest Sharpe ratio of 0.039272. The study also uses simulation tests. This helps remove forward-looking bias. The Sharpe ratio after this test is 0.028086.
Copyright © 2026–2026 EAI
