Value-based firm rating: The case of Turkish cement firms with machine learning methods

Authors : Müge Peştere-akçay, İlhan Küçükkaplan, Şaban Nazlıoğlu

Pages : 31-37

Doi:10.18070/erciyesiibd.1664645

View : 51 | Download : 73

Publication Date : 2025-12-31

Article Type : Research Paper

Abstract :This study conducts a value-based firm rating analysis utilizing advanced machine learning techniques. The analysis is based on balance sheet and income statement data spanning 2006 to 2019 for 12 Turkish cement firms. We develop a value-based rating model that integrates a comprehensive set of micro- and macroeconomic variables, including firm assets, country risk, sector risk, and market dynamics. Furthermore, we assess and compare the predictive performance of five alternative rating models: (i) discounted cash flow (DCF) value-based rating, (ii) market value-based rating, (iii) financial ratios-based rating, (iv) financial ratios and DCF value-based rating, and (v) financial ratios and market value-based rating. The empirical results indicate that the K-nearest neighbors (K-NN) machine learning method delivers superior predictive accuracy when combined with a value-based rating model; models relying solely on financial ratios exhibit the weakest performance. Moreover, incorporating firm value into financial ratio-based models significantly enhances their predictive power, while excluding firm value may lead to systematic biases. These findings highlight the importance of integrating valuation-based information into data-driven rating systems, suggesting that models capturing both firm fundamentals and market conditions offer a more reliable framework for evaluating corporate financial health, and thus provide deeper insights for decision-makers, particularly under uncertainty.
Keywords : irma derecelendirme, makine öğrenmesi, firma oranı, k-nn, k-ortalama

ORIGINAL ARTICLE URL