Evaluating ensemble learning techniques for stock index trend prediction: a case of China

Xiaolu Wei (), Yubo Tian, Na Li and Huanxin Peng
Additional contact information
Xiaolu Wei: Hubei University
Yubo Tian: CCB Frontier Capital (Hong Kong ) Limited
Na Li: Northeast Agricultural University
Huanxin Peng: Northeast Agricultural University

Portuguese Economic Journal, 2024, vol. 23, issue 3, No 8, 505-530

Abstract: Abstract Stock index trend prediction is a very important topic in the finance. The purpose of this paper is to compare six ensemble learning related techniques for stock index direction prediction, including four boosting methods (Categorical Boosting (CatBoost), Light Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting (XGBoost) and Gradient Boosting Decision Tree (GBDT)), one bagging method (Random Forest (RF)) and one tree-structured machine learning method (Decision Tree (DT)). The Shanghai Composite Index is chosen for experimental evaluation. A factor library of seventy-two technical factors, thirty-five macro factors and seven micro factors are our inputs. Our predictions are one month ahead, and each prediction model is evaluated by the Area Under Curve (AUC). The results indicate that ensemble learning techniques perform well in stock index prediction, with all AUC values above 0.5. RF is considered as the top algorithm with an AUC value of 0.7355 before feature selection and 0.6736 after feature selection. Also, we predict the stock index trend using a comprehensive factor library and three single factor libraries, respectively. The results show that forecasting stock index directions with a complete factor library is of great importance, which could achieve more stable forecasting results. This study contributes to literature in that it is, to the best of our knowledge, the first to make an extensive evaluation of ensemble learning related methods by constructing a comprehensive factor library and three single factor libraries.

Keywords: Evaluation; Ensemble learning techniques; Stock index; Trend prediction; C32; C45; C53 (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s10258-023-00246-1 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

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:spr:portec:v:23:y:2024:i:3:d:10.1007_s10258-023-00246-1

Ordering information: This journal article can be ordered from
http://www.springer. ... cs/journal/10258/PS2

DOI: 10.1007/s10258-023-00246-1

Access Statistics for this article

Portuguese Economic Journal is currently edited by Luís F. Costa

More articles in Portuguese Economic Journal from Springer, Instituto Superior de Economia e Gestao
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().