Risk Early Warning Modeling And System Implementation Of Small High-frequency Trading

In recent years,the application of financial big data has become a hot trend in the industry with the widespread popularization of the Internet and big data technologies.In this thesis,the small amount high-frequency trading risk of e-commerce system is analyzed for early warning modeling,and an early warning system is developed based on Spring + SpringMVC + SpringJDBC framework.The main work of this thesis includes:(1)Based on the problem in this thesis and the comparative analysis of the related work,the integration method such as the random forest algorithm and AdaBoost algorithm are selected as the early warning algorithm.Firstly,four early warning model schemes are designed by analyzing the two algorithms and tuning the parameters.Then the model trained by the random forest is selected as the most optimal model after establishing and evaluating the early warning models among four schemes.(2)Considering that the users of the system are internal professionals,a system integration method that can be compiled and run online is designed,and the offline warning model established by Python is integrated into the risk early warning systemimplemented by JavaWeb.After analysis of business requirements the system architecture is designed and the whole system includes three subsystems,seven sub-functions and four database entities.Then,UML is used for the system development such as the design of class diagram and the design of sequence diagram of the main functions.Finally one risk warning system is implemented.(3)According to the system requirements and software test specifications,unit test,functional test and system performance test are designed and executed.The test results proves the effectiveness and the practicability of the implemented system.