Customer Classification And Risk Management Of Futures Company

With continuously development of financial market, futures-trading is a high-risk and high-return investment behavior, which more and more become the focus of investment and financing, the futures customers are increasing. It is important that how to strengthen the risk management and improve the quality of service in the development of companies in futures. At present, the researches of the futures customers are subjective and qualitative in domestic market, which have few papers from the point of customer transaction behavior which reflects the customer situation angle exactly, to analyze the customer transaction situation. There are many places that can be improved. Therefore, this paper from a new point of view, based on the customer transactions data, using the basic statistical analysis methods and the prevailing data mining technologies carries on the cluster and the classification to the customers, from a quantitative point and the deeper level to study the trading behaviours of futures customers, to find the trading characteristics of different kinds of customers. And this paper apply the total score method, Decision Trees method, Bayesian Network method and Neural Network method to provide reference advices from different angles for the future customers.In order to strengthen the customers of futures companies risk management,and to help customers and companies to recognize customers transaction pattern. The risk and the cost of futures companies'management will reduce.The service of customers will improve and enhance.Then, these measures also can attract new customers and promote the further development of the futures trades.Firstly, the transaction behaviors of customers are divided into four basic abilities of risk ability, contribution, profitability and ability to stop by using factor analysis which base on the customer transaction dates. And on this basis, the futures customers are divided into three categories customers who are concerned in the company by applying clustering analysis method. Also, the author analyzes the advantages and disadvantages in each of ability of three kinds of customers. Then the paper apply the total score method, the Decision Trees method, the Bayesian Network method and Neural Network method of data mining which combine with the basic abilities and transaction indicators to classify customers. After that, the author evaluates and analysis the results of classification which refered to the results of clustering.The author also find out the characteristic of the trading behaviors of different kinds of customers. Meanwhile, the four methods of this paper are very convenient applied in risk management of futures companies in actual. The people can see comprehensive trade situation of customers intuitively by the total score method. Applying the Decision Tree method, you can find out the trading rules of each type customers .The effective of distinguishing is good. The scope of applicationg of The Bayesian Method is wide.This method can deal with the change of external factors. Using the Neural Network method, the management can not only judge the types of customers, but also can compare with the indexes weight in the customer categories. So the customers can be managed in target. The applications of these methods will have guide significance in the company risk management. Finally, the paper draws some conclusions and puts forward the direction research in future. The author hopes to further strengthen risk management of futures companies and to serve customers in better through this paper.