Predicting post-IPO financial performance: a hybrid approach using logistic regression and decision trees

Abstract

Purpose: This study enhances the financial modelling of companies undergoing an Initial Public Offering (IPO) by focusing on internal capability determinants and IPO proceeds. Design/methodology/approach: A hybrid logistic regression and shallow-depth decision tree approach are employed to predict the initial three-year post-IPO performance of companies listed on the Stock Exchange of Thailand (SET) using data from 2002 to 2021. Findings: The results demonstrate that these models not only perform competitively against complex machine learning algorithms but also surpass them in terms of interpretability, an essential feature in financial modelling. The proposed approach effectively captures the effects of each determinant, offering valuable insights into strategic resource allocation and investment decision-making during transition years. Originality/value: This study introduces a novel application that integrates logistic regression with decision trees to predict multiclass financial performance, filling the gap between complex machine learning techniques and interpretable financial models. It offers practical tools for companies and investors to make informed decisions in challenging post-IPO environments.