Original Research Article

Predicting Financial Distress Using Artificial Neural Network Models for Industrial Shareholding Companies Listed on the Amman Stock Exchange

1 The World Islamic Sciences and Education University (WISE)

* Corresponding author: Lubna Natheer Oqdeh ([email protected])

Available online: 09 Nov 2026

Abstract

This study aimed to examine financial distress, a problem that affects many companies and threatens their continuity and survival. Over the past decades, researchers have developed several quantitative and statistical models to predict financial distress. More recently, developments in artificial intelligence have provided more accurate, flexible, and efficient prediction models. This study aimed to examine the ability of artificial neural network (ANN) models to predict financial distress. To achieve this objective, financial data were collected, and 25 financial ratios were calculated to reflect different aspects of financial performance. These ratios represented the independent variables, while the financial status of the company represented the dependent variable. The study population consisted of 48 industrial companies listed on the Amman Stock Exchange over ten years (2014–2023). The final sample included 25 industrial companies, providing a total of 250 observations. Several statistical analyses were conducted, including descriptive statistics, correlation analysis, and multicollinearity testing. Principal Component Analysis (PCA) was also used to reduce dimensionality. Two artificial neural network models were developed to predict financial distress: Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM). The performance of both models was evaluated using standard evaluation metrics. The findings demonstrated that both models were effective in predicting financial distress, with the LSTM model showing better performance. The MLP model achieved 94%, 100%, 93.3%, and 96.6% for accuracy, precision, recall, and F1 score, respectively. The LSTM model achieved higher results of 97.5%, 100%, 97.2%, and 98.6% for the same metrics, respectively. The study recommends conducting further research using other artificial intelligence models to predict financial distress in different sectors, improve understanding of this phenomenon, and provide comparable evidence across different contexts.

KeywordsFinancial DistressArtificial Neural NetworkMultilayer Perceptron (MLP)Long Short-Term Memory (LSTM).

Main Subjects

Applied Humanities

License

Journal License

This work is licensed under a Creative Commons Attribution 4.0 International license

Copyright (c) 2026 Lubna Natheer Oqdeh, Abdullah Salah Hardan