Quality Inspection Of Fund Rating That Based On Neural Network

With the deepening of the economic globalization, the prosperity of capital market such as securities investment funds, substandard loan, stock and so on,the function of credit rating in our country’s financial market has become increasingly apparent. Supervision departments and the investors have become more depend on the rating agencies. At the same time they also put forward higher request to the theoretical research workers. How to test the information content of credit rating? How to improve the credibility of rating results and information of the investors? How to draw lessons from foreign advanced experience and combining with China’s national situations to establish scientific and reasonable index system? All of these questions have become the basic questions of our country’s future credit rating research.The function of credit rating is to reduce information asymmetry and reveal the risk. In order to achieve this, it’s important to guarantee the rating quality. Based on this kind of situation, this article attempts to establish a method to inspect the quality of rating results. We can know the merits of the quality from two aspects:one is the accuracy of the rating point, authentication, the basic assumptions of rating agencies about"high level, low default", another is the stability of the rating results, the level changes only when the basic credit risk changes, that is because the basic credit risk that the level based on changes slow, the change of the level will not strong. Therefore we inspect the quality of rating result from this two point. This article adopts the method of self-organizing neural network combined with the index data that describe the quality of the funds, and compare the results with the results of rating agencies to verify its accuracy. Based on fuzzy evaluation method and entropy weight method, combining with the transfer matrix method to test the stability, and then inspect the method by the following2months’performance of the funds. It is concluded that the method has good prediction and stability.At this point, inspecting the quality of rating results from the points of accuracy and stability is the perspective innovation; adopting the method of neural network to test the accuracy and using the transfer matrix to test the stability is the technological innovation. We hope this method can perfect our country’s fund rating quality inspection system. And all of these bring the rating agencies to play the role risk disclosure and measuring.