Network Modeling And Structural Analysis Of Stock Based On Complex Network

With the rapid development of Internet Finance, the global financial markets all over the world are getting closer and closer. The global stock market is a typical complex system. Complex network is exactly a powerful tool for studying complex systems, which is the high abstraction of complex systems. The global financial crisis caused by the subprime mortgage crisis of U.S.A led to the sharp fall for the global stock index, which had an important impact on the world economy. So it is important and urgent to study the global stock market. In this paper, the global stock index data in major countries during global financial crisis and before this stage and after this stage was taken as an important object of study. According to the fluctuation and correlation of price between the global stock index and the unique advantages of complex networks in system modeling and internal mechanism exploration and system evolution, the main works are as follows:(1)In view of the fact that the global stock market is not fully modeled during the financial crisis of the global stock market and from a comprehensive perspective of analysing of the global stock market, the paper took the open stock index data as object and and considered the single stock index as a network node and took the correlation between the index as a link. The global stock index network model during global financial crisis and before this stage and after this stage was built using the method of correlation coefficient threshold. Experimental results showed that the established network model is more reasonable and universal, compared with existing models.(2)In view of the fact that the existing local similarity link prediction algorithm had the bad prediction results and poor universality and from making full use of network local information, one kind of link prediction algorithm based on combing information is proposed in this paper. It is called by CI(Combining Index). Experimental results showed that the algorithm has high accuracy and strong adaptability, compared with the existing 10 kinds of local similarity indices. At the same time, the evaluation index of the link prediction algorithm is only a one-sided issue from the accuracy evaluation. From the perspective of evaluating the algorithm more comprehensively and scientifically, this paper puts forward an improved evaluation index— AEI(Actual Efficiency Index). That is to say that the evaluation index introduced the time factor into consideration. Experimental results showed that CN(Common Neighbours) index has higher AEI value, which is in good agreement with the results of the algorithm in practical application. Besides the result verified the scientificity and effectiveness of AEI index.(3)In view of the fact that the analysis of global stock market during global financial crisis and before this stage and after this stage is not comprehensively enough and from a comprehensive perspective of grasping the market rules and evolution mechanism of the period further, The key technologies of four kinds of complex networks are adopted for analysing the network. Firstly, the visual technique is used to visualize the model. Secondly, the network topology characteristics are analyzed, such as the node degree, average path length, clustering coefficient, the average number of neighbors and the network density, small world characteristics. Again, the motifs of the three types of network were explored. At last, the link prediction method and the proposed CI index are used to analyze the evolution of the network during the period of financial crisis. Comprehensive experimental results show that Using the above method can better explore characteristics of the global stock market network during global financial crisis and before this stage and after this stage. It has important practical significance for a better grasp of the rules of the stock market.