Predicting stock market trends using convolutional neural networks: A deep learning approach

Ege Özkul

Junior Management Science (JUMS), 2026, vol. 11, issue 1, 195-226

Abstract: Technical analysis aims to predict stock returns based on price and volume patterns and has seen growing adoption of machine learning methods. However, most approaches rely on hand-crafted features. This paper investigates whether deep learning applied to stock chart images can predict future returns without manual feature engineering. Building on Jiang et al. (2023), this study applies Convolutional Neural Networks (CNNs), a computer vision architecture, to predict stock returns from price charts and extends their approach by implementing Vision Transformers, specifically the Class-Attention in Image Transformer (CaiT). Stock prices, volumes, and moving averages are encoded into images, which are used to train models that classify future stock returns as either positive or negative. Results show that both CNN and CaiT models outperform traditional technical indicators such as momentum and reversal strategies when applied to US stocks. Moreover, combining the two models yields incremental predictive power. An investment strategy based on their joint predictions achieves higher returns and Sharpe ratios than either model alone.

Keywords: technical analysis; stock return prediction; deep learning; convolutional neural networks; vision transformers (search for similar items in EconPapers)
Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://www.econstor.eu/bitstream/10419/341437/1/1972505653.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:zbw:jumsac:341437

DOI: 10.5282/jums/v11i1pp195-226

Access Statistics for this article

Junior Management Science (JUMS) is currently edited by Dominik van Aaken, Gunther Friedl, Christian Koziol, Sascha Raithel

More articles in Junior Management Science (JUMS) from Junior Management Science e. V.
Bibliographic data for series maintained by ZBW - Leibniz Information Centre for Economics ().