Identification of Stock Violations in Chinese Listed Companies: Based on Machine Learning Algorithms

In recent years, frequent violations by publicly listed companies in China have eroded market trust. This study employs traditional financial and non-financial metrics, emerging ESG metrics, and composite indicators constructed from merger data to predict violations using these models. We identify the most effective data indicators and model algorithms through rigorous comparative analysis. The findings reveal that ESG rating metrics, along with conventional financial and non-financial metrics, provide robust predictive accuracy for regulatory breaches. Moreover, the composite metrics outperform individual sets. The XGBoost algorithm performs best among machine learning model algorithms, particularly trained by composite indicators.