Abstract
As one of the basic tools of statistical analysis, random matrix theory builds a bridge for the study of microscopic and macroscopic properties of complex networks. The eigenvalues of the adjacency matrix of the complex network after mapping correspond to the energy spectrum in the random matrix theory, so the characteristics of the complex network are concentrated in the volatility of the eigenvalue sequence. The eigenvalues of complex networks are analyzed through random matrix theory to find the relationship between their structure and properties. This paper discusses the design and realization of the comprehensive application platform of group enterprise financial management. Based on the current situation of the development of financial management software in domestic and foreign group companies, the thesis gives the goals and tasks of the platform research and design. The enterprise financial management platform has made a business and technical architecture design and has made an in-depth analysis and design of modules such as fund management, budget management, statement management, and accounting management. This paper finds that the critical point threshold makes the results of network community division below this threshold poor, and the results above this threshold are good. Furthermore, the tipping point threshold for the NYSE network is significantly lower than that of the CSM network. This paper further analyzes the dynamic cross-correlation matrix and examines the time-dependent changes in the number of outliers, the number of communities, the degree of network modularity, and the degree of decentralization of industry composition under the condition of two market critical point thresholds. From the magnitude of fluctuation, it is found that the NYSE network is much more stable than the CSM network, which may also be related to the basic characteristics of the two markets, because the capital chain relationship is much more stable than the supply chain relationship.
1. Introduction
Stochastic matrix theory can calculate and give the average results of all interactions contained in complex systems. By comparing the different characteristics of real systems and the universal properties of random matrix theory, the special nonrandom properties of real systems can be obtained [1]. The nature of interactions within the system provides an important approach. Scholars systematically expounded the random matrix theory [2]. Subsequently, researches on combining random matrices in the fields of communication, biology, and finance have been widely carried out [3].
Accelerating the construction of financial management informatization has very important practical and strategic significance for strengthening the scientific management of corporate finance and establishing a modern financial management system. However, due to the lack of awareness of its importance by the management of some enterprises, and the fact that financial management informatization is a systematic project, there is a lot of arduous work to be done, so a considerable number of enterprises are still in the construction and improvement of financial accounting management information systems [4]. Due to the limitations of other objective conditions in the enterprise, systems of various architectures coexist within the enterprise. Many enterprises have begun to take the system construction based on business flow as the strategic goal and focus of the long-term development of financial management informatization [5]. As a brand-new management information system that combines the latest information technology development achievements, the business flow system integrates the internal and external resources of the enterprise, optimizes all aspects of the enterprise activities, and scientifically designs the whole system through the overall planning of the business processes inside and outside the enterprise, so as to achieve a high contribution of enterprise resources and promote the efficient and orderly flow of logistics, capital flow, and information flow in the system, which can greatly improve the economic benefits of the enterprise [6].
This paper introduces the theory of random matrices and studies its analysis methods, including energy level sequence analysis, expansion, ensemble of random matrices, and so on. This paper describes the design and implementation of the financial management integrated application platform and makes a business and technical framework for the group enterprise financial management integrated application platform from the perspective of architecture. Aiming at the in-depth analysis of the data processing of the platform, the database is designed, and the relevant functional models of the platform are designed. The business architecture, functional architecture, and technical architecture of the platform are described. The design constraints and data table structure design of the system database are described. This paper adopts the random matrix method to analyze the static structural characteristics and dynamic evolution characteristics of the two markets. The random matrix analysis includes the distribution of eigenvalues, the distribution of eigenvector components, the contribution of the industry to large eigenvalues, the antiparticipation rate of eigenvectors, the evolution of the largest eigenvalue, the average eigenvalue over time, etc. The analysis of complex network community division includes network visualization analysis under different thresholds, the change trend of indicators such as the number of isolated points, modulus, community number, and the degree of decentralization of community composition with the increase of the threshold, as well as the above indicators under a given threshold with time.
2. Related Work
The random matrix theory uses statistical analysis of the energy spectrum and empirical characteristic spectrum of a complex system to obtain the degree of randomization of the actual empirical data, and to display the behavioral characteristics of the overall correlation in the empirical data [7]. Due to the high complexity of the financial system itself, it is very difficult to directly study its internal interactions. With the development of random matrix theory applied to nuclear physics, scholars have found that there is a relationship between the complex system of economics and the financial system [8]. There is a large degree of similarity, so the theory of random matrix can be introduced into the market research of the financial system [9].
Relevant scholars have analyzed the topological structure of the network constructed by 20 global financial indices before and during the crisis and divided the components of the second largest eigenvector, which contains two clusters, into positive correlation and negative correlation [10]. The two will transform, and the structure of the network will change from the star before the crisis to the chain during the crisis.
From the software structure point of view, each module of the enterprise management information system will operate independently but they must also be integrated. From the perspective of software functions, it not only includes accounting management, salary management, fixed asset management, purchasing and accounts payable management, sales and accounts receivable management, and inventory management but also includes material demand planning, production process, cost, and management of human resources [11]. From the perspective of software development platform and development technology, large-scale enterprise management information systems mainly use 32-bit development tools, and running on platforms above Windows 98, the database will no longer use the desktop database but must use the server database [12].
The network architecture mainly adopts three-layer (database server/application server or transaction processing server/client) or multilayer structure to overcome the defect that the traditional C/S structure can easily cause network bottleneck phenomenon [13]. In addition, in the large-scale management information system, the browser and Web server technology in the Internet/Intranet technology should be used to realize the standardization of the data structure of the software system, and the cross-regional and cross-platform operation. At the same time, the application of e-commerce in software functions should also be considered.
Relevant scholars believe that the arrival of the era of big data brings both opportunities and challenges to financial management [14]. While summarizing the disadvantages of traditional financial management, they propose changes in corporate financial management brought about by big data. The researchers conducted further research on the application of big data in financial management, combined with the current problems faced by Chinese enterprises in financial management, and put forward countermeasures for enterprises to improve their financial management level in the era of big data [15]. In an article published in the journal Innovation Forum, relevant scholars analyzed the close relationship between big data and the financial work of enterprises and then discussed how financial work has been affected by the era of big data [16].
In an article published in a financial and accounting research journal, the researchers proposed that the advantages of big data can be used to transform the financial information system and improve the level of financial informatization [17]. Relevant scholars have further studied the transformation direction of enterprise group financial management in the era of big data, starting from analyzing the financial problems that appear in the continuous development of enterprise groups, and put forward the conclusion that making full use of the advantages brought about by big data can help enterprise groups successfully achieve financial management [18].
Relevant scholars pointed out that the informatization construction of enterprise accounting management needs to involve various applications, such as personnel management system, financial accounting system, inventory management system, financial information exchange, and online financial submission system [19]. In these systems, the C/S and B/S architectures coexist, the programming languages used vary from Visual Basic to Delphi to Java, and the databases used vary from Access database to SQL Server, etc. It is well shared, the coding is not uniform, and the security is low, which greatly reduces the efficiency of various work. However, the operation of the enterprise is a whole, and each system needs to cooperate with each other, so the data exchange interface between the application systems has become a major problem that plagues the implementation of integration.
In the process of comprehensively building “enterprise financial information digitization,” if the platform is restructured and the existing information systems of various departments are planned in a unified manner, not only are a lot of manpower and material resources required, but also time is not allowed [20].
3. Application Platform Organizational Construction
3.1. Commercial Random Matrix
Random matrix theory is one of the important mathematical tools for statistical analysis of complex systems. Through statistical analysis of the energy spectrum and eigenstates of complex systems, it obtains the randomness of the actual data and reveals the behavioral characteristics of the overall correlation in the actual data.
How a thing is ordered by some dimension (time, space, or other abstract dimensions) is a problem common to many disciplines. We can map this to a list of values.
The average density of each energy level distribution is structurally the same, and we have to find other “numerical” ways to distinguish the different distributions. Probably the easiest and most intuitive way is to look at the interval distribution of successive energy levels.
We define the nearest neighbor interval (NNS) probability density function p(s), and p(s)ds is the probability that a particular si is in the interval (s, s + ds), or, given an energy level xi, the probability that the distance between the next energy level and xi is in the interval (s, s + ds). For a uniformly distributed sequence, the NNS distribution function is as follows:
The NNS distribution function of a completely random sequence is as follows:
The NNS distribution function of the “Wigner conjecture” is
Another information obtained from the energy level sequence is the number variance. Denote nL(x) as the number of energy levels in the interval with x as the starting point and length L; then, the variance of the number is
Random matrix theory (RMT) is not concerned with the overall distribution of the sequence of energy levels, but rather the volatility exhibited by a single energy level density compared to some average energy level density. Therefore, for a particular system, before analyzing the volatility of its energy level sequence, it becomes meaningful to estimate its overall energy level density.
The average distance between its energy levels is constant over the entire sequence range. We calculate the distance of a sequence from the average energy level density, a process called unfolding.
For an energy level sequence, we assume that its energy level density is given exactly and has the following form:
The cumulative energy level distribution is
As the complexity of the system under study increases, most problems are difficult to solve strictly, and even some approximation methods are ineffective. So people have to turn to statistical methods.
Statistical methods include two methods: “top-down” and “bottom-up.” “Top-down” means ignoring the dynamics of small-scale details and constructing theoretical models from only some of the more important aspects, while “bottom-up” means constructing theoretical models by considering the dynamics of all microscopic details.
We can then compare the theoretical results with the actual results and even infer some details. Obviously, the “top-down” approach is more realistic.
The nearest neighbor spacing distribution (NNSD) P(s) of a system with time-reversal symmetry and a Gaussian orthogonal ensemble is a Wigner function of the following form:
The nearest neighbor spacing distribution (NNSD) P(s) function for a system without time-reversal symmetry is similar to a Gaussian unitary ensemble:
3.2. Objectives and Principles of Platform Design
The platform design goals are reflected in the following points:(1)It is necessary to formulate a sound financial accounting mechanism to ensure that accounting can be implemented and operated in an environment of implementation, concentration, science, and speed. It is necessary to formulate a sound model management system for accounting management, statement collection, asset accounting, payroll accounting, and inventory accounting, and to unify all financial accounts within the entire enterprise.(2)Always ensure the smooth operation of the comprehensive budget management mechanism. On the premise of ensuring the balance of funds on the revenue and expenditure bus, improve various fund management methods, effectively control the accumulated funds of the enterprise, and try to speed up the activity of funds, so that the funds of the enterprise can be improved. Benefit can grow steadily and, in this way, can enhance the internal control ability of the enterprise. While promoting the overall budget management mechanism of the enterprise, it monitors the overall process of the enterprise and forms a complete and scientific fund management system. The most important thing is to ensure the effective implementation of capital occupancy and capital handling.(3)According to the current development trend of the market economic system, a fund management system that meets the survival and development of modern enterprises should be formulated, comprehensive budget management should be implemented within the enterprise, and a fund preassessment mechanism should be formulated. For the production budget, expenditure budget, and income budget of the enterprise, it has established hierarchical goals and established a capital management mechanism. Predevelopment assessment for large-scale capital possession projects mainly includes initialization, review, execution, and control of capital data. This plan is an important foundation for the smooth implementation of the comprehensive budget, and it is also one of the indispensable important facilities for a complete centralized management system of corporate funds.
The design principles of the comprehensive application platform for financial management of group enterprises are reflected in the following points.
3.2.1. Practicality
After the platform goes online, it can fully adapt to the current and future business development, meet the basic needs of group companies in terms of functions, reduce the burden on staff, and improve the main productivity of labor. The platform should have the characteristics of convenient operation and easy management.
3.2.2. Advanced Nature
The advanced nature mainly refers to the advanced nature of technology and management concept. When developing a financial management platform, whether in terms of structure, design, or management ideas, it should grasp the mainstream of technology and industry development and conform to the current development direction of information technology and software development.
3.2.3. Scalability
Scalability refers to the ease with which software can extend new functions. The better the scalability, the stronger the ability of the software to adapt to “changes.” In the current era of changing on demand, customer needs are often changing, and more and more new requirements are added, so the scalability of software is a very important measure of software performance. The group financial management platform needs to be able to better adapt to the new application requirements brought about by business development, changes, and institutional increase, so that the platform can maintain good scalability, upgradeability, and compatibility.
3.2.4. Safety and Reliability
The safety and reliability of software are an important criterion to measure the quality of software. Security refers to software attributes related to the ability to prevent unauthorized intentional or accidental access to programs and data; reliability refers to a set of attributes related to the ability of software to maintain its performance level under specified periods of time and conditions.
The orderly operation of the group’s financial management platform has an important influence on the development of the enterprise itself. If the enterprise’s financial management platform is damaged, financial information will be stolen or damaged. Therefore, efficient and safe protection measures must be taken to strengthen the network security of the enterprise financial management platform. In addition, software and hardware recovery measures in the event of a platform failure should also be considered.
3.3. How to Build a Financial Management Model
Group enterprises have obvious characteristics of cross-regional, cross-industry, and diversified operations, and the emergence of the Internet has effectively broken the limitation of time and space, which has fundamentally created favorable conditions for the improvement of centralized management efficiency of enterprises. Therefore, based on the development dimension of the group enterprise itself, it is necessary to effectively integrate the attributes and characteristics of the group itself and scientifically and rationally allocate software resources to ensure that the information of various types of enterprises can be of high quality. Based on the current Internet environment, the corresponding financial management models of group companies generally include the following three categories.
3.3.1. Real-Time Concentration
Real-time centralization is a management platform built by using B/S technology, which provides favorable conditions for centralized management. The group headquarters can rely on the corresponding financial accounting and management software, so that each unit can transmit data to the headquarters in time.
The group and member units can effectively meet the standards of information sharing, data and information can also be coordinated and circulated within the members, and the performance of each member unit can be directly evaluated.
Based on the dimension of an industrial group, the group generally allocates core resources such as capital and human resources in a centralized manner and conducts corresponding supervision; when the group conducts specific monitoring of the business of its subsidiaries, it mainly adopts a real-time centralized mode. Referring to Figure 1, it can be seen that group companies have broken the limitations of the traditional layer-by-layer reporting model, which is of great significance for them to grasp the overall information more quickly and ensure the authenticity and accuracy of the information.
3.3.2. Regular Concentration
Regular centralization refers to the use of C/S or B/S technology to build a corresponding platform, so that the group can control each member unit specifically, and the member units can use reports and other modes to transmit data, in order to achieve centralized management. The purpose is to provide favorable conditions.
From the perspective of an investment group, the relationship between the group and its subsidiaries is relatively scattered. If the group can regularly use reports, financial analysis, etc. to grasp the operation and development of the subsidiary, it needs to ensure that the group’s own information is of high quality. The process can be performed using a periodic centralized mode.
3.3.3. Mixed Concentration
Hybrid concentration refers to the use of C/S or B/S technology: the enterprise group can store the data and software of each member unit in the local and then use the network for specific transmission, so that the enterprise group can control the entire group.
From the perspective of a managed group, the corresponding management models of different subsidiaries are significantly different. Some subsidiaries are very poor in centralized management, while other subsidiaries need to use decentralized management to achieve their goals. Therefore, this type of enterprise can adopt a hybrid and centralized model to control member units in real time, while other members can adopt a regular centralized method to meet their own management standards and promote the improvement of information quality.
3.3.4. Factors Affecting the Financial Management Model of Group Companies
The financial centralized management mode will be affected by many factors in the specific selection process. The article starts with corporate culture, organization, tax treatment, industrial structure, equity structure, and other factors and divides the application of the above three modes (see Table 1).
3.4. Technical Architecture of the Platform
Technical architecture is an idea, a system blueprint, planning, and responsibility setting for software structure composition. The significance is to separate processing calculations, business logic, security rules, etc. and combine them with the agreed upon interfaces and protocols to form a software structure with clear responsibilities and clear structure.
The B/S architecture and middleware technology constitute the software architecture of the comprehensive financial management platform of the group enterprise, which realizes the centralized information management of business flow and data flow and greatly reduces the operating costs of the daily operation, application, and maintenance of the system. The business logic is encapsulated by the MVC-based technical framework, which realizes the display and visualization of the internal content elements—the separation between the graphical interface, the business logic (the information technology structure of the business process), and the data storage. In this way, according to the technical requirements of the system’s own operation and the number and management needs of the group’s financial business, a business platform can be built conveniently and quickly, and the specific business system can be well reflected in terms of reliability, stability, and reusability. Figure 2 shows the overall hierarchical status of the system, which is mainly divided into five layers, including page display layer, control layer, business logic layer, data access layer, and integration layer. Combined with the characteristics of the E7 project, the interface display adopts the display layer, the interface transfer control adopts the control layer, the business processing adopts the business logic layer, and the database operation adopts the data access layer.
3.5. Database Design
The database occupies a very important position in the comprehensive application platform of financial management. The quality of the database structure design will directly affect the efficiency and effect of the platform. Reasonable database structure design can improve the efficiency of data storage and ensure the integrity and consistency of data.
The requirements of cross-database products constitute a restriction on the structure design of the database; that is to say, the basic data types can be supported by general relational databases, so abstract database field types should be defined in the design. The naming of database fields should be readable, concise, organic, consistent, etc. which are the main principles that types should follow. Table 2 lists the definitions of the field types to be used according to the design requirements of the business platform for enterprise group financial management.
3.6. Design of Functional Modules
3.6.1. Design of Credential Management
The management of vouchers is to input the data of the financial accounting system into the quality gate, and it is the first confirmation of the information quality in the accounting business, so the design of voucher management is very important.
When the user fills in the key information that needs to be entered in accounting, it is necessary to ensure that they can enter the accounting interface. When entering data on the accounting page, JSP should receive data and call the accounting Vouehers() method. Finally, the data is saved in the background by accounting vouchers, and a success message should be returned to the user after the saving is successful.
When the administrator needs to query data, he first needs to enter the relevant query conditions and, after receiving it, call the information query.
When performing a credential query, firstly, you need to verify the identity and record the login log before entering the credential management interface. To use this function, the user must enter the relevant conditions, query the information that meets the conditions through the background database, and then return the information to the credential management interface.
3.6.2. Design of Account Books and Statements
The ultimate purpose of the account book report is to be able to query and print financial data and other functions. When the administrator needs to perform accounting on the account data, he first needs to enter the interface of the account book report, select the accounting conditions in the interface, and then send a data exchange request. When the relevant data entered by the administrator meets the relevant query conditions of the system, it will automatically run in the background and return relevant results. When the query conditions cannot be satisfied, a failure message will be returned.
4. Experimental Results and Analysis
4.1. Experimental Analysis of Complex Network with Static Cross-Correlation Matrix
The differences between the CSM network and the NYSE network are mainly manifested in two aspects: (1) judging from the growth of the number of communities divided by the community, the community division of the CSM network began to be substantially carried out when ξ = 0.48. The community division of the NYSE network starts to manifest at ξ = 3.0 (the second set of thresholds), which indicates that the community division of the NYSE network is much easier than that of the CSM network. (2) From the comparison between the results of community division and the actual industry division, almost all the divided communities in the CSM network contain the finances of several industries, and the community does not correspond to the actual industry. While the NYSE network has multiple divided societies that only contain the finances of one industry, the societies overlap well with the actual industry. Therefore, we can preliminarily conclude that the community division results of the CSM network are not as ideal as those of the NYSE network.
The visual network analysis roughly and qualitatively compares the CSM network and the NYSE network community division showing different structural characteristics. However, since the comparison and analysis results depend on the selection of the threshold ξ of each group of comparison networks, a small number of network comparisons cannot accurately reflect the difference between the two, and it is necessary to further investigate some important parameters of the network in order to more accurately quantitatively describe the threshold ξ.
According to the preliminary conclusions of the visual analysis, it is concluded that the following network parameters are more important.
4.1.1. The Number of Isolated Points in the Network Is O(ξ)
Outliers refer to the points where the network G(ξ) is moderately zero, and the correlation coefficients between the finances corresponding to these points and all other finances are less than the threshold ξ. Therefore, the number of outliers O(ξ) reflects the number of relatively weakly correlated financials in the market.
4.1.2. The Number of Communities M(ξ) Obtained by Dividing F(ξ) into Communities
The network after removing outliers is composed of multiple connected subgraphs. For each connected subgraph, it can also be divided into more communities by running the community partition algorithm. The final number of communities M(ξ) is an important parameter to view the evolution characteristics of the network structure.
First, we observe the change in the number of outliers O(ξ): the number of O(ξ) for the threshold ξ of the CSM network to increase from 0.36 to 0.55 is roughly equivalent to the number of O(ξ) for the threshold ξ of the NYSE network to increase from 0.25 to 0.55. Therefore, although the threshold ξ takes values in different ranges of the two networks, the two groups of networks are still comparable. Under the premise of comparability, we run the GN community division algorithm for each network separately and calculate the modularity Q(ξ) of the network and the number of communities M(ξ) after the network is divided. The change of the NYSE network modularity Q(ξ) is also consistent, but the number of communities M(ξ) also has certain similarities.
We compare the most important parameter of community division—the degree of decentralization of the industry composition of the community I(ξ) changes with the threshold ξ. From the trend of the mean value mean I(ξ), we also found that there is a critical change phenomenon. When the threshold value is smaller than the critical value, the mean I(ξ) is larger, indicating that the industry composition of most associations is highly decentralized, and the results of the division of finance and associations in different industries are not ideal. When the threshold value is greater than the critical value, mean I(ξ) suddenly decreases to a lower level, indicating that the financial industry of the community is less decentralized and the community division results are ideal. In addition, the critical point of mean I(ξ) mutation corresponds to the critical point of M(ξ) mutation. From the phenomenon that max I(ξ) does not decrease with the increase of the threshold ξ, we can think that the community division algorithm cannot completely distinguish the finances of all different industries. The evolution of CSM network and NYSE network with threshold ξ is shown in Figures 3 and 4.
4.2. Modularity Evolution Analysis of Complex Network with Dynamic Cross-Correlation Matrix
In this subsection, we analyze the community division of the respective dynamic cross-correlation matrices Ct of CSM and NYSE. Since there are M dynamic cross-correlation matrices, it requires a huge amount of computation to examine the characteristics of each matrix evolving with the threshold ξ in turn, and the actual economic significance is not great. Therefore, we only choose specific thresholds to observe the dynamic evolution of the network. From the analysis in the previous section, we found that when ξ is set to 0.50 and 0.38, the number of isolated points O(ξ) of the CSM network and NYSE network and the mean I(ξ) of the industry dispersion of the community are both the smallest.
In order to investigate the influencing factors of network parameter changes, the trends of the financial index returns of CSM and NYSE in corresponding periods are also given. The analysis shows that O(t) and Q(t) have a strong correlation with the change of time t and are consistent with the change of the mean value of the eigenvalue of the cross-correlation matrix, which is further positively correlated with the change of the mean value of the correlation coefficient ρ(t) relation. In addition, compared with the evolution of I(t), the fluctuation range of min I(t) and mean I(t) of CSM is larger. The results of the CSM network community division are sometimes good and sometimes bad, and the network structure fluctuates violently, while the NYSE network community division has consistently good results, and the network structure fluctuation is weak. The evolution results of the dynamic cross-correlation matrix complex network modularity are shown in Figures 5 and 6.
4.3. High-Frequency Dynamic Financial Network Evolution Statistics
The analyzed financial complex networks are established by thresholding the cross-correlation matrix, and the method of community division is also used to effectively extract the structural features of these networks. However, we know that when analyzing a network, some conventional statistical indicators such as average degree and average clustering coefficient can effectively reflect the basic properties of the network. A high-frequency dynamic financial network with static threshold and dynamic threshold is established, which is very different from the static and dynamic financial network constructed in this paper in construction method. This method does not use the cross-correlation matrix directly but uses the real-time cross-correlation matrix. Compared with the cross-correlation matrix, the real-time cross-correlation matrix only uses the instantaneous price at a certain moment, rather than the correlation coefficient of a certain time window. Therefore, the resulting network will be more high-frequency and contain different statistical laws.
Figures 7 and 8 show the static threshold Qs and the dynamic threshold Qd(t). Overall, the static threshold Qs is much lower than the dynamic threshold. The dynamic threshold Qd(t) varies greatly with time. It is not difficult to see that its volatility is highly paroxysmal, but it still remains within a certain range, which shows that the impact of the overall market volatility on the network can be effectively eliminated when building the network.
5. Conclusion
As one of the statistical tools for studying complex objects, random matrix theory has laid a solid foundation for the development of economics and other disciplines. The random matrix theory studies complex networks from a macro perspective, which opens up a new way for the study of complex networks. Therefore, random matrix theory is of great significance to the research and analysis of complex networks. In the construction of the platform, based on the analysis of the group enterprise financial management theory and related computer software technology, network technology, and the group financial accounting process, this paper proposes a comprehensive application platform for group enterprise financial management based on the J2EE architecture. Based on the idea of centralized financial management, this paper analyzes the needs and main business processes of group enterprise financial management, including budget management, statement management, accounting, and so on. According to the needs of the financial management of the group enterprise, this paper makes a business and technical framework for the system and designs the database according to the needs of the comprehensive application platform of financial management. In this paper, a visual analysis of complex networks constructed by static cross-correlation matrices under several comparable thresholds is carried out, and it is found that the NYSE network can more easily divide the community than the CSM network, and the community can be divided when the threshold is small. In general, the results of the NYSE network community division are significantly better than those of the CSM network. The main reason may be that the NYSE is a market centered on the financial tertiary industry, while the CSM is a market centered on manufacturing. This paper finds that the financial algorithm classification of the CSM market does not correspond to the standard market sector classification, while the financial algorithm classification of the NYSE market is more consistent with the standard sector classification. In terms of the reasons for their differences, this paper gives two explanations for their economic implications: first, there are differences in the investment decisions of investors in the two markets. Secondly, there are differences in the fundamentals of the two markets: the primary market is the secondary industry represented by nondaily consumer goods, industries, and other industries as the pillar industry; the secondary market is represented by the financial industry and the tertiary industry is the NYSE market.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.
Conflicts of Interest
The authors declare that they have no conflicts of interest or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by Guoyuan Securities Company Limited.