MACHINE LEARNING MODELS FOR UNDERSTANDING CORPORATE RISKS AND PERFORMANCE PATTERNS

This research addresses the question of how the size of a firm influences its financial performance and risk-taking
behavior by analyzing the first 100 companies of the FTSE index over the period 2018-2023. This period is
characterized by economic uncertainty due to Brexit and the Covid-19 pandemic. The main question is whether
large firms perform better and manage their risks more efficiently than smaller ones.
Previous research has been criticized for using only traditional methodologies like OLS regression, whereas this
study employs machine learning techniques, thus, moving one step further. The researchers employed seven
different models: Logistic Regression, Support Vector Machine (SVM), Random Forest, Decision Tree, Gradient
Boosting, Extra Trees, and K-Nearest Neighbors (KNN). Stock price changes were used as a proxy for firm
performance, stock return volatility as a measure for risk-taking, and log of market capitalization as a proxy for
firm size.
The results firm size is highly positively correlated with financial performance showing that large firms generally
more profitable and stable. Moreover, they engage in fewer unstructured risk activities, which implies that their
financial operations are more intentional and under control. Logistic Regression was the best-performing model
in terms of predictive accuracy with a 97% success rate, among all machine learning models tested, showing the
effectiveness of machine learning in financial research.
This paper is a great contribution in terms of new insights from the UK market and it addresses the gap in the
existing literature which is mostly based on the US and Asia. Besides, it proves that machine learning can be a
great tool in corporate finance analysis by discovering the complex patterns that traditional methods may fail to
uncover.