Research Of Listed Company Financial Performance Evaluation Based On BP Neural Network

The China's listed companies have been developing with the reform of joint stock system in the state-owned enterprise since the mid 1980's. With economical step speeding up, listed compaines have gradually become an active power in the national economy. The performace evaluation of the listed companies appears more and more importantly in many aspects such as guiding the investor rationality and supervising listed companies to strength internal management. The performance evaluation of listed companies arises from above situation. On the basis of the significance and the development history of the performance evaluation of listed companies, this paper analyzes the domestic and foreign existing performance evaluation systems and discusses several problems which the performance evaluation of the listed companies apperear. The dissertation concludes below:(1) This paper is based on the entrusting theory, the management theory ,biggest and smallest principle. The entrusting theory tells us that the behalf and aim disagreement and information dissymmetry between clients and deputy decides the logic basis of performance evaluation systems. The level of entrusting desides the extension of performance evaluation systems. The management theory asks the companies to create an open and integrated guide line system which is based on a lots of investigations and advices. Biggest and smallest principle tells us that the most benefit is the fundamental purposeof companies and it also incarnates the performance evaluation.(2)There are many typical achievements appraisals methods such as EVA,REVE and Balanced Score card. The model of performance evaluation of listed companies system is a process to look for function mapping like other modelings. However, the relationship between guide lines of this system is non-linear. Typical achievements appraisals methods are not suitable straitly.(3) BP's neural network can learn agone experiences automatically from studying samples without any complicated search and description. It can approach the function automatically which describe the rule of the studying samples best whatever the form of the function is. Therefore, this paper puts forward a new method of the performance evaluation of listed companies , which is based on BP's neural network. Then uses the NNT based on MATLAB to make network design and calculate. Through the training and testing of lots of studying samples, the simulation model has been established. In the end, this paper proves the accuracy feasibility of the method with an example.