Link-based financial risk management method, system equipment and storage medium
By constructing dynamic financial networks and using machine learning models, the problem of insufficient real-time monitoring and systemic risk prediction capabilities of traditional financial product risk management methods is solved, and more accurate risk assessment and early warning is achieved, dynamically capture the risk transmission process, and provide quantitative risk management indicators.
Patent Information
- Application Number
- CN202411026768.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-22
AI Technical Summary
Traditional financial product risk management methods have limited ability to monitor real-time and predict systemic risk. Static network models cannot dynamically capture the risk transmission process. The correlation coefficient ignores high-order correlation, lacks an effective risk warning mechanism, and ignores the concentration risk of financial products. The impact of
By combining link prediction and financial product network analysis, we can obtain historical data of financial products to construct a dynamic financial network, use machine learning models to predict risk indicators, set early warning thresholds and trigger risk management suggestions, and dynamically capture the time-varying nature of the topology of financial product networks, and use Link prediction technology predicts risk propagation paths.
Achieve more accurate prediction and assessment of concentration risks, systemic risks and infectious risks of financial products, dynamically capture the risk transmission process, provide active warnings, consider high-order correlations, quantify the degree of risk exposure, and provide accurate risk management Quantitative indicators.
Smart Images

Figure CN119006168A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of wealth management and network science and technology, and in particular to the fields of financial product risk control and monitoring, and more particularly to a link-based financial risk management method, system equipment and storage medium. Background Art
[0002] In the field of traditional financial product risk management, risk assessment and early warning mainly rely on historical data analysis, market monitoring and expert experience. These methods usually include quantitative analysis of market volatility, credit ratings, liquidity indicators and operational risks. For example, financial product management companies use value-at-risk (VaR) models to assess market risk, credit scoring models to assess credit risk, and liquidity forecasting models to assess liquidity risk. These methods can help financial product management companies identify and prevent potential risks to a certain extent, but they often rely on the accuracy of historical data and the stability of market conditions, and have limited ability to monitor in real time and predict systemic risks.
[0003] Traditional financial product risk management methods have several major flaws: 1) Static network models cannot dynamically capture the risk propagation process; 2) Traditional similarity metrics such as correlation coefficients ignore high-order correlations and have difficulty capturing risk contagion in extreme cases; 3) Existing models mostly focus on descriptive analysis and lack an effective risk warning mechanism; 4) Most studies ignore the impact of financial product concentration risk.
[0004] Therefore, how to use complex network theory to analyze and understand the nonlinear behavior and risk propagation mechanism of the financial market; how to evaluate the interactions and potential risks between financial entities through link prediction technology; and how to build a real-time monitoring system to track the dynamic changes of the financial market and issue early warning signals in a timely manner are technical problems that need to be solved urgently. Summary of the invention
[0005] The present invention can more accurately predict and evaluate the concentration risk, systemic risk and contagion risk of financial products by combining link prediction and financial product network analysis. To this end, the present invention provides a link-based financial risk management method, system device and storage medium to solve at least one of the above technical problems.
[0006] To achieve the above objectives, in a first aspect, the present invention provides a link-based financial risk management method, comprising:
[0007] Obtain historical data of financial products as nodes in the network graph, calculate the distance matrix based on the daily logarithmic returns of financial products, and construct a dynamic financial network through network filtering methods;
[0008] Based on the link prediction and the financial network characteristics of the nodes, the financial market structure is predicted through a preset machine learning model to obtain risk indicator data of financial products;
[0009] Obtaining financial products identified as high-risk based on the risk indicator data of the financial products, and the risk indicator data serving as input for subsequent calculations of concentration risk, systemic risk and contagion risk indicators;
[0010] In combination with the high-risk financial products, the risk index of the dynamic financial network is calculated, and a warning threshold is set according to the risk index of the dynamic financial network, and a warning is triggered if it is higher than the threshold;
[0011] Provide corresponding risk management suggestions based on early warning information.
[0012] Furthermore, in the acquisition of financial product historical data as a node in the network diagram, the financial product historical data includes: historical holdings data, yield and risk indicators of public financial products, as well as quarterly, semi-annual and annual reports of each financial product, daily net value of financial products and public sentiment of financial products, etc.
[0013] Furthermore, the construction of a dynamic financial network through a network filtering method includes: the network filtering method estimates the financial market structure through three different methods: including a dynamic asset graph, a dynamic threshold network and a dynamic minimum spanning tree; the three methods estimate the asset distance matrix through the common movement measure of daily logarithmic returns.
[0014] Furthermore, setting a warning threshold according to the risk indicators of the dynamic financial network includes: setting a warning threshold, setting the distance matrix threshold of the concentration risk to 0; when it is detected that the distance of a financial product in the network exceeds 0.5, it is identified as a high concentration risk financial product, triggering a warning.
[0015] In a second aspect, the present invention also provides a link-based financial risk management system, the system comprising: a data collection module, a network construction module, an indicator calculation module, a product screening module, a risk warning module, and a risk management strategy formulation module;
[0016] The data acquisition module is used to obtain historical data of financial products;
[0017] The network construction module is used to use the acquired historical data of financial products as nodes in the network diagram, calculate the distance matrix according to the daily logarithmic returns of financial products, and construct a dynamic financial network through a network filtering method;
[0018] The indicator calculation module is used to predict the financial market structure through a preset machine learning model based on the link prediction and the financial network characteristics of the node, and obtain risk indicator data of financial products;
[0019] The product screening module is used to obtain financial products identified as high-risk according to the risk indicator data of the financial products, and the risk indicator data is used as input for the subsequent calculation of concentration risk, systemic risk and contagion risk indicators;
[0020] The risk warning module is used to calculate the risk index of the dynamic financial network in combination with the high-risk financial products, set a warning threshold according to the risk index of the dynamic financial network, and trigger a warning if it is higher than the threshold;
[0021] The risk management strategy formulation module is used to provide corresponding risk management suggestions based on the early warning information.
[0022] Furthermore, in the acquisition of financial product historical data as a node in the network diagram, the financial product historical data includes: historical holdings data, yield and risk indicators of public financial products, as well as quarterly, semi-annual and annual reports of each financial product, daily net value of financial products and public sentiment of financial products, etc.
[0023] Furthermore, the construction of a dynamic financial network through a network filtering method includes: the network filtering method estimates the financial market structure through three different methods: including a dynamic asset graph, a dynamic threshold network and a dynamic minimum spanning tree; the three methods estimate the asset distance matrix through the common movement measure of daily logarithmic returns.
[0024] Furthermore, the system also includes a monitoring and adjustment module; which is used to regularly update the financial product network and recalculate risk indicators; when an early warning is triggered, investors are notified, and new risk management suggestions are provided based on the new early warning.
[0025] In a third aspect, the present invention provides a computing device, comprising:
[0026] A memory for storing program instructions;
[0027] The processor is used to call the program instructions stored in the memory and execute the above-mentioned link-based financial risk management method according to the obtained program instructions.
[0028] In a fourth aspect, the present invention provides a computer-readable storage medium, including computer-readable instructions. When a computer reads and executes the computer-readable instructions, the above-mentioned link-based financial risk management method is implemented. The readable medium may be multiple, and the multiple readable media can operate independently of each other.
[0029] Compared with the existing technology, this solution obtains the historical data of financial products as nodes in the network diagram, calculates the distance matrix based on the daily logarithmic returns of financial products, and constructs a dynamic financial network through the network filtering method; predicts the financial market structure through the preset machine learning model based on the link prediction and the financial network characteristics of the nodes, and obtains the risk indicator data of financial products; obtains the financial products identified as high-risk based on the risk indicator data of financial products; calculates the risk indicators of the dynamic financial network in combination with high-risk financial products, sets the warning threshold based on the risk indicators of the dynamic financial network, and triggers the warning if it is higher than the threshold; provides corresponding risk management suggestions based on the warning information. It achieves a more accurate prediction and evaluation of the concentration risk, systemic risk and contagion risk of financial products, and dynamically captures the time-varying nature of the topological structure of the financial product network, more accurately describes the risk propagation process, uses link prediction technology to predict the risk propagation path, and realizes active risk warning. Considering high-order correlation, it can better capture the risk contagion in extreme cases. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0031] Figure 1 A flowchart of an embodiment of a link-based financial risk management method provided by the present invention;
[0032] Figure 2 This is a structural block diagram of an embodiment of the link-based financial risk management system provided by the present invention. DETAILED DESCRIPTION
[0033] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.
[0034] It will be understood by those skilled in the art that, unless otherwise stated, the singular forms "a", "an" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.
[0035] In order to make the objectives, technical solutions and advantages of the present application more clear, the specific implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.
[0036] The core of the present invention is to provide a link-based financial risk management method. Figure 1 The flowchart of the link-based financial risk management method provided by this embodiment is shown in FIG. Figure 1 The technical solution of this embodiment is described in detail:
[0037] Step S1: Calculate the distance matrix of the nodes in the table based on the daily logarithmic returns of financial products, and construct a dynamic financial network through network filtering method;
[0038] In a specific embodiment, the historical data of financial products can be obtained by collecting the historical holding data, yield and risk indicators of the public offering financial products available in the market. These data may include the quarterly, semi-annual and annual reports of each financial product, as well as the daily net value of the financial product and the public opinion of the financial product.
[0039] In a specific embodiment, the network filtering method is as follows:
[0040] Set the closing price of the financial product at time t to P(t). Calculate the return time based on the daily logarithmic return of the financial product R(t) = logP(t) - logP(t-1).
[0041] First, a distance matrix is calculated, which measures the relative relationship of daily log returns and is defined as:
[0042]
[0043] where ρt(i,j) is the Pearson correlation coefficient between the logarithmic return time series of financial products i and j over a time window ending at time t, Where V is the set of financial instruments. The distance matrix is constructed by partitioning the return time series R(t) into rolling windows of size L trading days and setting δT trading days (time step) between two consecutive windows. The choice of window width L and window time step δT is arbitrary, and it is a trade-off between too dynamic or too smooth analysis. The smaller the window width and the larger the window step, the more dynamic the data. Here we mainly take L∈{42,84,126} and δT=5 trading days. A dynamic financial network is defined as a time network:
[0044]
[0045] where vertex i∈V corresponds to the financial product of interest. For each pair in time window t<i,j> , There exists a corresponding edge (i,j)t∈Et, and each edge has a weight wi,j(t)=Di,j(t). Based on the distance matrix Di,j(t) defined above, a dynamic network is created by applying the network filtering method.
[0046] Step S2: According to the link prediction and the financial network characteristics of the node, the financial market structure is predicted by a preset machine learning model to obtain risk indicator data of financial products;
[0047] In a preferred embodiment, the risk indicator data of a financial product may be historical volatility, maximum drawdown, negative public opinion about the financial product, etc.
[0048] Step S3: Obtain financial products identified as high-risk based on the risk indicator data of the financial products, and the risk indicator data is used as input for the subsequent calculation of concentration risk, systemic risk and contagion risk indicators.
[0049] Specifically, high risks include the following three types of risks: concentration risk, systemic risk and contagious risk;
[0050] 1. Concentration risk
[0051] 1.1 Concentration risk standards:
[0052] The concentration risk standard is defined by factors such as the strength of association between financial products. Based on these standards, financial products or financial product portfolios with high concentration risk in the financial product network are screened out.
[0053] 1.2 Constructing concentration risk indicators:
[0054] After completing the construction of the financial product network, the following indicators are used to assess and monitor concentration risk.
[0055] The distance matrix Di,j,constructs the dynamic financial product network part, where the larger the value,the higher the concentration of financial products i and j.
[0056] Eigenvector Centrality: measures the importance of a node by considering not only the number of nodes directly connected to it, but also the importance of these nodes. In a financial product network, financial products with high eigenvector centrality have a significant impact on the concentration risk of the network.
[0057] Clustering Coefficient: It measures the degree of interconnection between neighboring nodes of a node. A high clustering coefficient indicates that a financial product has a high concentration with other financial products, which may lead to concentration risk.
[0058] Edge Overlap: Measures the degree of overlap in the investment targets of two financial products. A high edge overlap means that the investment concentration of the two financial products is high, which increases the concentration risk.
[0059] 1.3 Concentration risk assessment and early warning
[0060] The above indicators are used to analyze the financial product network and keep real-time monitoring of the dynamic changes of the financial product network. When it is found that the concentration risk of certain financial products or financial product groups exceeds the preset threshold, the system will issue a concentration risk warning.
[0061] When a client's financial product portfolio is too concentrated, it is recommended to reduce the concentration risk caused by a single investment by adding financial products of different categories, markets and strategies to the portfolio.
[0062] 2. Systemic risk
[0063] 2.1 Identify high-risk financial products
[0064] First, define the standards for high-risk financial products. You can screen and identify high-risk financial products based on historical volatility, loss rate, risk indicators, etc.
[0065] 2.2 Constructing network risk indicators
[0066] Construct a financial product network based on link prediction and calculate the following related indicators.
[0067] Modularity: It measures the tightness of node grouping in the network. The higher the Modularity value, the more obvious the node grouping.
[0068] Community Structure: Identify community structures in the network, which are combinations of nodes that are highly cohesive and poorly coupled.
[0069] Degree Centrality: Measures the number of direct connections a node has with other nodes. Highly central nodes may be key to information dissemination.
[0070] Betweenness Centrality: It measures the information transmission ability of nodes in the network. Nodes with high betweenness centrality are likely to be located on the information propagation path.
[0071] Closeness Centrality: Measures the average shortest path length from a node to other nodes. Nodes with high closeness centrality can spread information quickly.
[0072] 2.3 Systemic risk prediction
[0073] Based on the indicator evaluation results, when it is found that certain financial product groups have reached the preset risk threshold, such as high-risk financial products are clustered and occupy a core position in the network, a systemic risk warning will be issued.
[0074] 3. Infectious risks
[0075] Based on the identification of high-risk financial products and the construction of network risk indicators, we further analyze the contagion effect of high-risk financial products on other financial products, identify financial products that may be affected by contagion risks, and take corresponding risk control measures.
[0076] 3.1 Identify high-risk financial products
[0077] First, define the standards for high-risk financial products. You can screen and identify high-risk financial products based on historical volatility, maximum drawdown, public opinion, etc.
[0078] 3.2 Analysis of Contagion Effect Based on Link Prediction
[0079] Using the link prediction results, we analyze the potential connection between high-risk financial products and other financial products, and evaluate the contagion effect of high-risk financial products on other financial products.
[0080] Calculate the link weights between high-risk financial products and other financial products. The higher the weight, the greater the risk of contagion. Identify financial products that are closely linked to high-risk financial products, which are more susceptible to contagion risks.
[0081] At the same time, the position and centrality of high-risk financial products in the network are analyzed to evaluate their importance in the contagion process. Finally, the key node financial products on the contagion path are identified, which play an important intermediary role in the contagion process.
[0082] 3.3 Take risk control measures
[0083] For financial products identified as affected by contagion risks, take corresponding risk control measures to mitigate potential negative impacts. Strengthen monitoring and early warning of affected financial products, and pay close attention to changes in their risk profiles. Adjust the positions and allocations of affected financial products to reduce correlation with high-risk financial products. For financial products at key nodes on the contagion path, take key control measures to reduce their ability to spread risks. In portfolio construction, appropriately reduce the weight of affected financial products to diversify contagion risks.
[0084] Step S4: Calculate the risk index of the dynamic financial network in combination with the high-risk financial product, set an early warning threshold according to the risk index of the dynamic financial network, and trigger an early warning if it is higher than the threshold;
[0085] Step S5: Provide corresponding risk management suggestions based on the warning information.
[0086] In a specific embodiment, the network filtering method estimates the financial market structure through three different network filtering methods: including dynamic asset graph (DAG), dynamic threshold network (DTN) and dynamic minimum spanning tree (DMST); these methods estimate the asset distance matrix through the common movement measure of daily log returns.
[0087] 1.1 Dynamic Asset Graph (DAG)
[0088] The dynamic asset graph is a filtered financial network modeled by first arranging the edges in descending order of weights w1(t), w2(t), ..., wN(N-1) / 2(t). The resulting graph is obtained by selecting the strongest connected edges, and the number of edges is arbitrary. In this method, edges with weights in the top quartile are selected, i.e., w1(t), w2(t), ..., w[N(N-1) / 8](t).
[0089] 2.2 Dynamic Threshold Network (DTN)
[0090] Considering the distance matrix D(t) defined in equation (1), a filtered adjacency matrix A is created to construct the financial network using the following rules:
[0091]
[0092] Where assets i,j∈V, The key value rc transforms the matrix D into an undirected network, where Aij(t)=1 and Aij(t)=0 represent the presence and absence of an edge between i and j in time window t, respectively. The rc value is fixed at 0.65 because for rc≤0.65, the network characteristics are overwhelmed by large fluctuations. The goal of this method is to identify asset pairs that are highly correlated and above a threshold rc. This method is different from DAG, in which pairs with correlation values below rc can be added to the network.
[0093] 2.3 Dynamic Minimum Spanning Tree (DMST)
[0094] A dynamic minimum spanning tree is created based on the minimum asset distances in the matrix D(t) defined earlier. The Kruskal algorithm can be used to identify the minimum spanning tree (MST) in the fully connected graph D at time t. The number of edges is fixed to N-1, where N is the number of assets. This method provides the minimum distance, producing the minimum market structure to connect all assets.
[0095] In one embodiment, the information required for training link prediction is provided by extracting a set of network features at the node level and the link level. The network features are used as input to the machine learning model, and the output is two node connections.
[0096] The network features are summarized as follows: node-level features evaluate the position of nodes in the overall structure of a given graph G(V,E), and link-level features examine the content and pattern of the relationship between edges in a given graph G(V,E). Table 1 shows a set of node-level features used as input for machine learning models. Table 2 shows a set of link-level features used as input for machine learning models.
[0097]
[0098] Table 1
[0099]
[0100] Table 2
[0101] This embodiment helps predict the market structure itself by analyzing the topological information of the network. Therefore, the feature set is divided into two different subsets. Labels are given to these two subsets based on the source of information: (i) pairwise correlated features, which are features based on asset correlations and are not derived from any other network information, and (ii) non-pairwise correlated features, which are features derived from the network topology. Although pairwise correlated features are traditionally used in financial analysis, the importance of non-pairwise correlated features for predicting market structure is an important consideration for this embodiment and is part of the information gain in market structure prediction relative to existing methods. In Table 1, all features are non-pairwise correlated attributes. In Table 2, the first two features are pairwise correlated features.
[0102] In one embodiment, financial network features are used as input into a machine learning model, and market structure prediction is regarded as link prediction for predicting the market structure represented by a given financial product network. Given a snapshot of the financial network up to time t, the solution provided by this embodiment can accurately predict the edges that will appear in the network at a given future time t' (t' = t + h). By selecting three time points t0 < t < t', and providing an algorithm that accesses W[t0,t] = <V, Et0,..., Et> to estimate the likelihood of the edges that will appear in W[t'], where t' = t + h and h = {1, 2,..., 20} trading weeks. The similarity-based method and the classifier-based method are the two most common methods in link prediction. In the classifier-based method, link prediction is defined as a binary classification problem. Here, a feature vector is extracted for each pair of nodes, and a 1 / 0 label should be assigned based on the presence or absence of the link in the network. In this embodiment, the classifier-based method is applied to predict the financial product network structure, and the features used are the relevant link-level features and the node features.
[0103] In one embodiment, by dividing the dataset into a training set and a test set, the time series of the data is considered. The training set uses data generated during the period from January 1, 2015 to December 31, 2021, and the test set uses data from January 1, 2022 to December 31, 2023. The machine learning model is trained and tested using the rolling window method. Considering L as the size of the logarithmic return time series, t as the current time, t - k < t < t + h, the training set is created by using the network features in G(t - k), where k = 1, 2,..., 30. The test set contains the data of the current network G(t), where G(t + h) is the target, where h = 1, 2,..., 20 trading weeks. After training and testing the machine learning model, the rolling window is moved by considering a time step δT = 5 trading days (1 trading week) in order to restart the process and retrain the machine learning model. To evaluate the information rate that the machine learning model can extract from the feature set, this embodiment uses the XGboost algorithm.
[0104] The prediction performance of the link prediction method is evaluated by calculating the area under the ROC curve (AUC). It ranges from 0.5 to 1, where 0.5 represents the result of the random naive algorithm and 1 represents the best result. The AUC metric provides a summary indicator for the algorithm of this embodiment, and at the same time, the shape of the ROC curve reveals the prediction performance of the algorithm for each prediction set size.
[0105] In the scheme of this embodiment, by obtaining the historical data of financial products as nodes in the network diagram, the distance matrix is calculated according to the daily logarithmic returns of financial products, and a dynamic financial network is constructed through a network filtering method; according to the link prediction and the financial network characteristics of the nodes, the financial market structure is predicted through a preset machine learning model to obtain the risk indicator data of financial products; according to the risk indicator data of financial products, financial products identified as high-risk are obtained; in combination with high-risk financial products, the risk indicators of the dynamic financial network are calculated, and the warning threshold is set according to the risk indicators of the dynamic financial network. If it is higher than the threshold, the warning is triggered; according to the warning information, corresponding risk management suggestions are provided. It achieves a more accurate prediction and evaluation of the concentration risk, systemic risk and contagion risk of financial products, and dynamically captures the time-varying nature of the topological structure of the financial product network, more accurately describes the risk propagation process, uses link prediction technology to predict the risk propagation path, realizes active risk warning, considers high-order correlation, can better capture the risk contagion in extreme cases, and quantifies the risk exposure by using the topological structure characteristics of the financial product network, providing quantitative indicators for risk management.
[0106] In one embodiment, a link-based financial risk management system is provided. Figure 2 The figure shows a block diagram of a link-based financial risk management system provided by this embodiment, wherein the system comprises: a data acquisition module 10, a network construction module 20, an indicator calculation module 30, a product screening module 40, a risk warning module 50, and a risk management strategy formulation module 60;
[0107] The data acquisition module 10 is used to obtain historical data of financial products;
[0108] The network construction module 20 is used to use the acquired historical data of financial products as nodes in the network diagram, calculate the distance matrix according to the daily logarithmic returns of financial products, and construct a dynamic financial network through a network filtering method;
[0109] The indicator calculation module 30 is used to predict the financial market structure through a preset machine learning model according to the link prediction and the financial network characteristics of the node, and obtain risk indicator data of financial products;
[0110] The product screening module 40 is used to obtain financial products identified as high-risk according to the risk indicator data of the financial products, and the risk indicator data is used as input for subsequent calculation of concentration risk, systemic risk and contagious risk indicators;
[0111] The risk warning module 50 is used to calculate the risk index of the dynamic financial network in combination with the high-risk financial product, set a warning threshold according to the risk index of the dynamic financial network, and trigger a warning if it is higher than the threshold;
[0112] The risk management strategy formulation module 60 is used to provide corresponding risk management suggestions based on the early warning information.
[0113] In a preferred embodiment, in the acquisition of financial product historical data as a node in the network diagram, the financial product historical data includes: historical holdings data, yield and risk indicators of public financial products, as well as quarterly, semi-annual and annual reports of each financial product, daily net value of financial products and public sentiment of financial products, etc.
[0114] In a preferred embodiment, the construction of a dynamic financial network through a network filtering method includes: the network filtering method estimates the financial market structure through three different methods: including a dynamic asset graph, a dynamic threshold network and a dynamic minimum spanning tree; the three methods estimate the asset distance matrix through the common movement measure of daily logarithmic returns.
[0115] In a preferred embodiment, a monitoring and adjustment module is also included; it is used to regularly update the financial product network and recalculate risk indicators; when an early warning is triggered, investors are notified, and new risk management suggestions are provided based on the new early warning.
[0116] In the scheme of this embodiment, by obtaining the historical data of financial products as nodes in the network diagram, the distance matrix is calculated according to the daily logarithmic returns of financial products, and a dynamic financial network is constructed through a network filtering method; according to the link prediction and the financial network characteristics of the nodes, the financial market structure is predicted through a preset machine learning model to obtain the risk indicator data of financial products; according to the risk indicator data of financial products, financial products identified as high-risk are obtained; in combination with high-risk financial products, the risk indicators of the dynamic financial network are calculated, and an early warning threshold is set according to the risk indicators of the dynamic financial network, and an early warning is triggered if it is higher than the threshold; according to the early warning information, corresponding risk management suggestions are provided. It achieves a more accurate prediction and evaluation of the concentration risk, systemic risk and contagious risk of financial products, and dynamically captures the time-varying nature of the topological structure of the financial product network, more accurately describes the risk propagation process, uses link prediction technology to predict the risk propagation path, realizes active early warning of risks, considers high-order correlations, can better capture risk contagion in extreme cases, and quantifies the degree of risk exposure by using the topological structure characteristics of the financial product network, providing quantitative indicators for risk management.
[0117] The link-based financial risk management system of this embodiment is used to implement the aforementioned link-based financial risk management method. Therefore, the specific implementation method of the link-based financial risk management system can be seen in the specific implementation method part of the link-based financial risk management method in the previous text. For example, the data acquisition module 10 is used to implement step S1 in the above-mentioned link-based financial risk management method, and the network construction module 20 is used to implement step S2 in the above-mentioned link-based financial risk management method, etc. Therefore, its specific implementation method can refer to the description of the corresponding various partial embodiments. The system provided by the present invention is attached to the method provided by the present invention, and its specific effects can refer to the above-mentioned method, which will not be repeated here.
[0118] In one embodiment, a computing device is provided, comprising:
[0119] A memory for storing program instructions;
[0120] A processor is used to implement the steps of the link-based financial risk management method as described in any of the above when executing the computer program. A link-based financial risk management method of the present invention has obvious advantages and beneficial effects compared with the prior art. By obtaining historical data of financial products as nodes in a network diagram, a distance matrix is calculated based on the daily logarithmic returns of financial products, and a dynamic financial network is constructed through a network filtering method; based on link predictions and the financial network characteristics of the nodes, the financial market structure is predicted through a preset machine learning model to obtain risk indicator data of financial products; financial products identified as high-risk are obtained based on the risk indicator data of financial products; in combination with high-risk financial products, the risk indicators of the dynamic financial network are calculated, and a warning threshold is set based on the risk indicators of the dynamic financial network, and a warning is triggered if it is higher than the threshold; and corresponding risk management suggestions are provided based on the warning information. It achieves more accurate prediction and assessment of concentration risk, systemic risk and contagion risk of financial products, dynamically captures the time-varying nature of the topological structure of financial product networks, more accurately describes the risk propagation process, uses link prediction technology to predict the risk propagation path, and achieves active risk early warning. By considering high-order correlations, it can better capture risk contagion in extreme situations, quantify the degree of risk exposure by utilizing the topological structure characteristics of the financial product network, and provide quantitative indicators for risk management.
[0121] In one embodiment, a computer-readable storage medium is provided, comprising computer-readable instructions. When a computer reads and executes the computer-readable instructions, the steps of the link-based financial risk management method described above are implemented. The link-based financial risk management method of the present invention has obvious advantages and beneficial effects compared with the prior art. The method obtains historical data of financial products as nodes in a network diagram, calculates a distance matrix based on the daily logarithmic returns of financial products, and constructs a dynamic financial network through a network filtering method; predicts the financial market structure through a preset machine learning model based on link prediction and the financial network characteristics of the nodes, and obtains risk indicator data of financial products; obtains financial products identified as high-risk based on the risk indicator data of financial products; calculates the risk indicator of the dynamic financial network in combination with high-risk financial products, sets an early warning threshold based on the risk indicator of the dynamic financial network, and triggers an early warning if it is higher than the threshold; and provides corresponding risk management suggestions based on the early warning information. It achieves more accurate prediction and assessment of concentration risk, systemic risk and contagion risk of financial products, dynamically captures the time-varying nature of the topological structure of financial product networks, more accurately describes the risk propagation process, uses link prediction technology to predict the risk propagation path, and achieves active risk early warning. By considering high-order correlations, it can better capture risk contagion in extreme situations, quantify the degree of risk exposure by utilizing the topological structure characteristics of the financial product network, and provide quantitative indicators for risk management.
[0122] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the specific system can be divided into different functional modules to complete all or part of the functions described above.
[0123] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0124] In several embodiments covered by the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, or it can be an electrical, mechanical or other form of connection.
[0125] The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of the present application.
[0126] In addition, each functional module in each embodiment of the present application can be implemented in the form of hardware or in the form of software functional modules. If these functional modules are implemented in the form of software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application is essentially or part of the prior art that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a computer program product, which includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer program product is stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, a computer instruction can be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrations. The available media may be magnetic media (eg, floppy disks, hard disks, magnetic tapes), optical media (eg, DVDs), or semiconductor media (eg, solid state disks (SSDs)).
[0127] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.
Claims
1. A link-based financial risk management method, characterized in that: include: Obtain historical data of financial products as nodes in the network graph, calculate the distance matrix based on the daily logarithmic returns of financial products, and construct a dynamic financial network through network filtering methods; Based on the link prediction and the financial network characteristics of the nodes, the financial market structure is predicted through a preset machine learning model to obtain risk indicator data of financial products; Obtaining financial products identified as high-risk based on the risk indicator data of the financial products, and the risk indicator data serving as input for subsequent calculations of concentration risk, systemic risk and contagion risk indicators; In combination with the high-risk financial products, the risk index of the dynamic financial network is calculated, and a warning threshold is set according to the risk index of the dynamic financial network, and a warning is triggered if it is higher than the threshold; Provide corresponding risk management suggestions based on early warning information.
2. The link-based financial risk management method according to claim 1, characterized in that: In the acquisition of financial product historical data as a node in the network diagram, the financial product historical data includes: historical holdings data, yield and risk indicators of public financial products, as well as quarterly, semi-annual and annual reports of each financial product, daily net value of financial products and public sentiment of financial products, etc.
3. The link-based financial risk management method according to claim 1, characterized in that: The method of constructing a dynamic financial network through a network filtering method includes: the network filtering method estimates the financial market structure through three different methods: dynamic asset graph, dynamic threshold network and dynamic minimum spanning tree; the three methods estimate the asset distance matrix through the common movement measurement of daily logarithmic returns.
4. The link-based financial risk management method according to claim 1, characterized in that: The step of setting a warning threshold according to the risk indicators of the dynamic financial network includes: setting a warning threshold, setting the distance matrix threshold of the concentration risk to 0; when it is detected that the distance of a financial product in the network exceeds 0.5, it is identified as a high concentration risk financial product, triggering a warning.
5. A link-based financial risk management system, characterized in that: include: Data collection module, network construction module, indicator calculation module, product screening module, risk warning module, and risk management strategy formulation module; The data acquisition module is used to obtain historical data of financial products; The network construction module is used to use the acquired historical data of financial products as nodes in the network diagram, calculate the distance matrix according to the daily logarithmic returns of financial products, and construct a dynamic financial network through a network filtering method; The indicator calculation module is used to predict the financial market structure through a preset machine learning model based on the link prediction and the financial network characteristics of the node, and obtain risk indicator data of financial products; The product screening module is used to obtain financial products identified as high-risk according to the risk indicator data of the financial products, and the risk indicator data is used as input for the subsequent calculation of concentration risk, systemic risk and contagion risk indicators; The risk warning module is used to calculate the risk index of the dynamic financial network in combination with the high-risk financial products, set a warning threshold according to the risk index of the dynamic financial network, and trigger a warning if it is higher than the threshold; The risk management strategy formulation module is used to provide corresponding risk management suggestions based on the early warning information.
6. The link-based financial risk management system according to claim 5, characterized in that: In the acquisition of financial product historical data as a node in the network diagram, the financial product historical data includes: historical holdings data, yield and risk indicators of public financial products, as well as quarterly, semi-annual and annual reports of each financial product, daily net value of financial products and public sentiment of financial products, etc.
7. The link-based financial risk management system according to claim 5, characterized in that: The method of constructing a dynamic financial network through a network filtering method includes: the network filtering method estimates the financial market structure through three different methods: dynamic asset graph, dynamic threshold network and dynamic minimum spanning tree; the three methods estimate the asset distance matrix through the common movement measurement of daily logarithmic returns.
8. The link-based financial risk management system according to claim 5, characterized in that: It also includes a monitoring and adjustment module, which is used to regularly update the financial product network and recalculate risk indicators; when an early warning is triggered, it notifies investors and provides new risk management recommendations based on the new early warning.
9. A computing device, characterized in that include: A memory for storing program instructions; A processor is used to call the program instructions stored in the memory, and execute the method according to any one of claims 1 to 4 according to the obtained program instructions.
10. A computer-readable storage medium, characterized in that: It includes computer-readable instructions. When a computer reads and executes the computer-readable instructions, it implements the method as claimed in any one of claims 1 to 4. There can be multiple readable media, and the multiple readable media can operate independently of each other.
