The Research Department of Economics and Management of Tadbir Nikan Business, Marketing, and Finance Open 2 Serial Number 11 2025 09 01 Application of Gradient Boosting Algorithm in Predicting Relative Pricing of Initial Public Offerings: A Case Study of the Iranian Stock Exchange Application of Gradient Boosting Algorithm in Predicting Relative Pricing of Initial Public Offerings: A Case Study of the Iranian Stock Exchange 1 21 10.61838/bmfopen.236 EN Journal Article 2025 03 21 <p>This study aimed to develop a model based on the Gradient Boosting algorithm to predict the relative pricing of initial public offerings (IPOs) in the Iranian stock market. Another objective of the research was to identify the factors influencing the relative pricing of IPO stocks. This study was conducted using data from 42 companies listed on the Tehran Stock Exchange and 121 companies listed on the Iran Fara Bourse over the period from 2013 to 2023. The three main dependent variables examined included the market-to-book value ratio, the enterprise value-to-asset ratio, and the enterprise value-to-sales ratio. The independent and control variables were extracted based on financial theories and previous studies, and were utilized in gradient boosting models and subsequently in a combined machine learning model based on gradient boosting algorithms. The combined gradient boosting model demonstrated a high capability in predicting the relative pricing variables. Internal financial factors (weighted average cost of capital, return on assets, financial leverage) and performance-related variables (operating profit margin, earnings per share) had the greatest impact. Firm size, firm age, and cash flow ratio were among the influential control variables. The model was able to identify interaction effects and nonlinear relationships among variables. The application of the combined gradient boosting method for the first time in Iran’s capital market to price initial public offerings constitutes the primary innovation of this study. The results showed that the combined model had a lower prediction error compared to standalone gradient boosting models. This innovative approach, applied for the first time in the Iranian capital market, demonstrated high efficiency in dealing with market complexities and can assist investors, underwriters, and regulatory institutions in making better decisions.</p> https://www.bmfopen.com/index.php/bmfopen/article/download/236/182