Key Drivers of Volatility in BISTt100 Firms Using Machine Learning Segmentation

Abstract

This study conducts a comprehensive volatility analysis among firms listed on the BIST100 index using machine learning techniques and panel regression models. Focusing on the period from 2006 to 2023, the study excludes financial firms, resulting in a dataset of 46 companies. The methodology follows a two-step process: First, firms are clustered into low and high-volatility groups using Principal Component Analysis (PCA) and the K-means algorithm; second, panel regression models are applied to determine the financial ratios influencing stock price volatility. The Parkinson Volatility measure is used as the dependent variable, while independent variables include Return on Assets (ROA), Return on Equity (ROE), liquidity ratios, firm beta, and leverage ratios. Results indicate that firm beta has a statistically significant positive impact on volatility across all models, while the current ratio negatively affects volatility in the model 1. These findings provide valuable insights for investors and policymakers regarding risk management in the Turkish stock market. Applying machine learning and advanced econometric techniques adds to the literature on volatility forecasting and financial decision-making. © 2025 Elsevier B.V., All rights reserved.