Fifteen years of overall performance of Tehran Stock Exchange: a hybrid descriptive and predictive analysis

Ghiyasi, Mojtaba; Valizadeh, Omid; Lotfi, Mohsen

doi:10.22059/ijms.2026.413741.678559

Fifteen years of overall performance of Tehran Stock Exchange: a hybrid descriptive and predictive analysis

Articles in Press

Document Type : Research Paper

Authors

Mojtaba Ghiyasi ¹

Omid Valizadeh ²

Mohsen Lotfi ¹

¹ Faculty of Industrial Engineering and Management Science, Shahrood University of Technology

² Department of Management, Faculty of Economics and Administrative Sciences, Ferdowsi University of Mashhad, Mashhad, Iran

10.22059/ijms.2026.413741.678559

Abstract

Evaluating the performance of publicly listed firms is essential for understanding capital market dynamics, particularly in volatile environments. This study proposes a hybrid framework integrating Dynamic Network Data Envelopment Analysis (DEA) with machine learning to assess and predict the performance of 130 firms listed on the Tehran Stock Exchange over the period 2007–2022. In the first stage, DNDEA is employed to measure efficiency by capturing both the internal multi-stage structure of firms and their intertemporal dynamics. In the second stage, XGBoost and LightGBM models are applied to identify key determinants of efficiency and improve prediction accuracy under different feature selection scenarios. The results indicate that the proposed approach achieves high predictive performance, with LightGBM outperforming XGBoost after addressing multicollinearity using the Variance Inflation Factor (VIF). This study contributes by providing an integrated framework that combines dynamic and network-based efficiency analysis with advanced predictive modeling, offering valuable insights for investors and policymakers in emerging markets.

Keywords

DEA

Dynamic Network DEA

Machine Learning

Stock Exchange

Subjects