Adapting Altman's bankruptcy prediction model to the compositional data methodology

Adapting Altman's bankruptcy prediction model to the compositional data methodology

Adapting Altman's Bankruptcy Prediction Model to the Compositional Data Methodology

What Happened?

A team of researchers has successfully adapted the widely used Altman bankruptcy prediction model to the compositional data methodology, a statistical approach that addresses issues with standard financial ratios. The study compared the results of this adapted model with those obtained from standard financial ratios, using a dataset of 31,131 firms from the Spanish wholesale trade sector, with 97 of them being bankrupt. The researchers employed three machine learning tools - logistic regression models, k-nearest neighbors, and random forests - in combination with compositional log-ratios. The results showed that the compositional methods outperformed standard ratios in terms of sensitivity, with mixed results regarding specificity.

Background & Context

The Altman bankruptcy prediction model, developed in 1968, is a widely used tool for predicting corporate bankruptcy. However, its reliance on standard financial ratios has been criticized for its limitations, including the presence of extreme outliers, asymmetry, non-normality, and non-linearity. The compositional data methodology, on the other hand, has been successfully applied to various fields to address these issues. By adapting the Altman model to this methodology, researchers aimed to improve its predictive performance and provide a more accurate tool for identifying firms at risk of bankruptcy.

Impact on Swiss SMEs & Finance

The findings of this study have significant implications for Swiss SMEs and the finance industry. The use of compositional methods in bankruptcy prediction can lead to more accurate assessments of credit risk, enabling lenders to make more informed decisions. This, in turn, can improve the overall stability of the financial system. Additionally, the study's results can be applied to other sectors and economies, potentially providing a more robust tool for predicting bankruptcy and insolvency.

What to Watch

As the study's findings suggest that compositional methods can improve the predictive performance of bankruptcy models, it is likely that researchers and practitioners will continue to explore the application of this methodology in finance. The use of machine learning tools in combination with compositional log-ratios is an area that warrants further investigation, particularly in the context of Swiss SMEs and the broader financial industry.

Source

Original Article: Adapting Altman's bankruptcy prediction model to the compositional data methodology

Published: March 25, 2026

Author: Fatemeh Keivani

---

Disclaimer: This article is for informational purposes only and does not constitute financial advice. Consult a licensed financial advisor before making investment decisions.

Related Articles

• Mislearning of Factor Risk Premia under Structural Breaks: A Misspecified Bayesian Learning Framework
• Mislearning of Factor Risk Premia under Structural Breaks: large A Misspecified Bayesian Learning Framework
• Adaptive Regime-Aware Stock Price Prediction Using Autoencoder-Gated Dual Node Transformers with Reinforcement Learning Control