Dynamic Financial Early Warning Research Based On Support Vector Machine

Chinese security market is developing but not mature. Many listed companies have poor compression and they are in distress. This situation impair related interests of investors and creditors. So, how to build a financial prediction model has become a topic subject. It not only has high academic value but also enormous practical value. If we can generalize the research results to other no-listed companies, we can prevent various types of enterprises from financial crisis and the research is much more practical significance. The focus point of the paper is to establish a financial early warning model of listed companies and to improve the level of company's financial management.Firstly, the paper reviews previous empirical results of the research on the financial predicting models. On this bases, sixty-six A Publicly listed companies, which are special treated(ST) are selected as samples. And according to matching principle of the same industry and assets size, the equal quantity firms whose financial position are good are selected as control samples. Sencondly, the paper chooses elevn financial indicators to set up the new financial index system, At last, we use support vector machines to set up a new financial distress prediction model, which gives the direction of the enterprise operation.The main conclusions are as follows:①The support vector machine dynamic warning models has high accuracy and Hign practical economic value in reading a listed company financial position.②The BP neural network model itself is instability.③Make a comparison between SVM warning model and the BP neural network model, it is concluded that the support vector machine is more superior.