Predicting stock splits using ensemble machine learning and SMOTE oversampling

Ang LI; Mark LIU; Simon SHEATHER

doi:10.1016/j.pacfin.2023.101948

Predicting stock splits using ensemble machine learning and SMOTE oversampling

Ang LI^*, Mark LIU^*, Simon SHEATHER

^*Corresponding author for this work

FACULTY OF BUSINESS

Research output: Journal Publications › Journal Article (refereed) › peer-review

5 Citations (Scopus)

Abstract

This study predicts stock splits using two ensemble machine learning techniques: gradient boosting machines (GBMs) and random forests (RFs). The goal is to form implementable portfolios based on positive predictions to generate abnormal returns. Since splits are rare events, we use SMOTE oversampling to synthesize new observations of splits in the sample to improve predictions. When predicting stock splits in the next quarter, GBM and RF achieve area under the receiver operating characteristic curve (AUC) scores of around 0.86 and 0.87, respectively. GBM and RF predictions generate monthly five-factor alphas (Fama and French, 2015) of 0.26% and 0.95% among stocks in the smallest size quintile. Three important features for predicting stock splits in both ensemble ML methods are current price levels, the ratio of current price to the price at last split, and stock returns in the past twelve months. When predicting stock splits in the next year, GBMs generate monthly five-factor alphas of 0.38% among small stocks.

Original language	English
Article number	101948
Number of pages	24
Journal	Pacific-Basin Finance Journal
Volume	78
Early online date	28 Jan 2023
DOIs	https://doi.org/10.1016/j.pacfin.2023.101948
Publication status	Published - Apr 2023

Bibliographical note

Funding Information:
☆ We thank Leonce Bargeron, Alex Brefeld, Charlie Clarke, Grant Clayton, Koustav De, Will Gerken, Russell Jame, Paulo Manoel, Peter Qi, Fabrice Riva (discussant), Jake Smith (discussant), Spencer Stone, Mao Ye, seminar participants at the University of Kentucky, and session participants at the 2022 Commonwealth Computational Summit, 12th Financial Markets and Corporate Governance Conference, and the 2022 FMA Annual Meeting for helpful comments. All errors and omissions are our own.

Publisher Copyright:
© 2023 Elsevier B.V.

Keywords

Stock splits
Ensemble machine learning
Gradient boosting machines
Random forests
SHAP feature importance
Hyperparameter tuning
SMOTE oversampling

Access to Document

10.1016/j.pacfin.2023.101948

Cite this

@article{689d155379974793b99b0a4234a60aa7,

title = "Predicting stock splits using ensemble machine learning and SMOTE oversampling",

abstract = "This study predicts stock splits using two ensemble machine learning techniques: gradient boosting machines (GBMs) and random forests (RFs). The goal is to form implementable portfolios based on positive predictions to generate abnormal returns. Since splits are rare events, we use SMOTE oversampling to synthesize new observations of splits in the sample to improve predictions. When predicting stock splits in the next quarter, GBM and RF achieve area under the receiver operating characteristic curve (AUC) scores of around 0.86 and 0.87, respectively. GBM and RF predictions generate monthly five-factor alphas (Fama and French, 2015) of 0.26% and 0.95% among stocks in the smallest size quintile. Three important features for predicting stock splits in both ensemble ML methods are current price levels, the ratio of current price to the price at last split, and stock returns in the past twelve months. When predicting stock splits in the next year, GBMs generate monthly five-factor alphas of 0.38% among small stocks.",

keywords = "Stock splits, Ensemble machine learning, Gradient boosting machines, Random forests, SHAP feature importance, Hyperparameter tuning, SMOTE oversampling",

author = "Ang LI and Mark LIU and Simon SHEATHER",

note = "Funding Information: ☆ We thank Leonce Bargeron, Alex Brefeld, Charlie Clarke, Grant Clayton, Koustav De, Will Gerken, Russell Jame, Paulo Manoel, Peter Qi, Fabrice Riva (discussant), Jake Smith (discussant), Spencer Stone, Mao Ye, seminar participants at the University of Kentucky, and session participants at the 2022 Commonwealth Computational Summit, 12th Financial Markets and Corporate Governance Conference, and the 2022 FMA Annual Meeting for helpful comments. All errors and omissions are our own. Publisher Copyright: {\textcopyright} 2023 Elsevier B.V.",

year = "2023",

month = apr,

doi = "10.1016/j.pacfin.2023.101948",

language = "English",

volume = "78",

journal = "Pacific-Basin Finance Journal",

issn = "0927-538X",

publisher = "Elsevier",

}

TY - JOUR

T1 - Predicting stock splits using ensemble machine learning and SMOTE oversampling

AU - LI, Ang

AU - LIU, Mark

AU - SHEATHER, Simon

N1 - Funding Information: ☆ We thank Leonce Bargeron, Alex Brefeld, Charlie Clarke, Grant Clayton, Koustav De, Will Gerken, Russell Jame, Paulo Manoel, Peter Qi, Fabrice Riva (discussant), Jake Smith (discussant), Spencer Stone, Mao Ye, seminar participants at the University of Kentucky, and session participants at the 2022 Commonwealth Computational Summit, 12th Financial Markets and Corporate Governance Conference, and the 2022 FMA Annual Meeting for helpful comments. All errors and omissions are our own. Publisher Copyright: © 2023 Elsevier B.V.

PY - 2023/4

Y1 - 2023/4

N2 - This study predicts stock splits using two ensemble machine learning techniques: gradient boosting machines (GBMs) and random forests (RFs). The goal is to form implementable portfolios based on positive predictions to generate abnormal returns. Since splits are rare events, we use SMOTE oversampling to synthesize new observations of splits in the sample to improve predictions. When predicting stock splits in the next quarter, GBM and RF achieve area under the receiver operating characteristic curve (AUC) scores of around 0.86 and 0.87, respectively. GBM and RF predictions generate monthly five-factor alphas (Fama and French, 2015) of 0.26% and 0.95% among stocks in the smallest size quintile. Three important features for predicting stock splits in both ensemble ML methods are current price levels, the ratio of current price to the price at last split, and stock returns in the past twelve months. When predicting stock splits in the next year, GBMs generate monthly five-factor alphas of 0.38% among small stocks.

AB - This study predicts stock splits using two ensemble machine learning techniques: gradient boosting machines (GBMs) and random forests (RFs). The goal is to form implementable portfolios based on positive predictions to generate abnormal returns. Since splits are rare events, we use SMOTE oversampling to synthesize new observations of splits in the sample to improve predictions. When predicting stock splits in the next quarter, GBM and RF achieve area under the receiver operating characteristic curve (AUC) scores of around 0.86 and 0.87, respectively. GBM and RF predictions generate monthly five-factor alphas (Fama and French, 2015) of 0.26% and 0.95% among stocks in the smallest size quintile. Three important features for predicting stock splits in both ensemble ML methods are current price levels, the ratio of current price to the price at last split, and stock returns in the past twelve months. When predicting stock splits in the next year, GBMs generate monthly five-factor alphas of 0.38% among small stocks.

KW - Stock splits

KW - Ensemble machine learning

KW - Gradient boosting machines

KW - Random forests

KW - SHAP feature importance

KW - Hyperparameter tuning

KW - SMOTE oversampling

UR - http://www.scopus.com/inward/record.url?scp=85149781722&partnerID=8YFLogxK

U2 - 10.1016/j.pacfin.2023.101948

DO - 10.1016/j.pacfin.2023.101948

M3 - Journal Article (refereed)

SN - 0927-538X

VL - 78

JO - Pacific-Basin Finance Journal

JF - Pacific-Basin Finance Journal

M1 - 101948

ER -

Predicting stock splits using ensemble machine learning and SMOTE oversampling

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this