Research On Financial Crisis Prediction Based On DD And SVM

With the development of market economy and economy globalization, there have been not only more opportunities but also more risks and uncertainties for the listed companies in China. Listed companies may suffer from pressures of growing competition in external and poor management in internal and many of them are involved in financial crisis and even go bankrupt. How to predict financial crisis of the listed companies especially those in the GEM of China has been becoming a hot topic to be discussed in the fierce competitive market. In order to survive and keep the market run regularly, the prediction of financial crisis becomes necessary and urgent.Support Vector Machine (SVM) has better learning ability and generalization performance, which has been drawn wide attention in the financial sector in recent years. Financial crisis prediction is researched based on DD-FWSVM model in the study. The main topics are followed.Firstly, establish a new risk rating system based on distance to default (DD) and order statistics (OS). According to the new risk rating system, listed companies are classified into three ratings marked by T1, T2 and T3, which represent high risk, moderate risk and low risk respectively. Then, the new risk rating system is applied to the listed companies in the GEM of China and further financial crisis prediction in this paper.Secondly, it is the research of feature weighting for financial indicators based on DD and grey relational degree. Grey relational degrees between DD and financial indicators are the feature weights on account of that DD can provide effective predicting information for the financial crisis of the listed companies.Lastly, it is the research of financial crisis prediction based on biorthogonal wavelet hybrid kernels DD-FWSVM model with three classifications. A new biorthogonal wavelet hybrid kernels function of weighted features based on DD is proposed. In the function, grey relational degrees between DD and financial indicators are the feature weights of corresponding financial indicators. Considering the complexity of the financial situation, the DD-FWSVM financial crisis prediction model with three classifications based on the new kernel function is further constructed. The experiments analysis is conducted based on the listed companies in the GEM of China at last.On the one hand, prediction of financial crisis can reduce operational risk, contribute to specification as well as optimize the allocation of social resources based on an effective prediction model, On the other hand, financial crisis prediction can make operators be able to respond risk flexibly and provides effective means of supervision for government and defends the national interests.