<p>In the context of global financial markets, the goal of this study is to explore machine learning-based methods that provide new solutions for early warning of financial risk and decision support, thereby maintaining market stability and protecting investors’ rights and interests. This is in response to the financial risk identification and accounting analysis needs of listed companies. The method adopted involves collecting a large amount of financial statement data and market information from listed companies, constructing a multidimensional feature set through data cleaning and preprocessing, and then utilizing machine learning technology to build a financial risk identification system. Key empirical results show that the machine learning model is significantly better than traditional statistical methods in financial risk identification, with an accuracy rate of 92%, a recall rate of 88%, an F1 score of 0.90, and a prediction error of 15% compared with traditional machine learning models when processing complex financial data, effectively improving the accuracy and efficiency of risk warning. Its theoretical meaning is to verify the advantages of machine learning technology in processing complex financial data and improve the accuracy of risk identification, enriching the methodology of financial risk modeling, and the practical meaning is to provide investors and regulators with a more accurate financial risk reference, which is helpful to optimize decision-making and enhance market risk management and control capabilities.</p>