Security market mobility risk metering method, device, equipment, medium and product

Abstract

The invention discloses a security market mobility risk metering method, device and equipment, a medium and a product, and relates to the field of mobility risk metering, and the method comprises the steps: obtaining to-be-metered stock mobility data; wherein the stock mobility data to be measured comprises a listed company investment relation graph, news information and stock information data in a preset time period, and the preset time period comprises a plurality of continuous time nodes; based on the to-be-metered stock mobility data, a trained stock mobility risk metering model is adopted to meter stock mobility risks of all listed companies, and stock mobility risk indexes of all listed companies at the next time node are obtained; wherein the stock mobility risk measurement model comprises a graph isomorphic neural network, a directed graph neural network, a recurrent neural network and a fusion unit. According to the invention, the accuracy of stock mobility risk measurement in the security market can be improved.

Description

Technical Field

[0001] The present application relates to the field of liquidity risk measurement, and particularly to a method, device, equipment, medium and product for measuring the liquidity risk of the securities market. Background Art

[0002] Financial security is the cornerstone for maintaining the stability of the financial system and ensuring the sustainable and healthy development of the economy. As an important part of financial security, the increase in liquidity risk will affect the macroeconomic level, including the effectiveness of monetary policy and the stability of economic growth. In the securities market, liquidity risk is particularly important, which is related to the effective allocation of capital, the confidence of investors and the stability of the financial system. The deterioration of its risk may trigger a financial crisis in extreme cases. Therefore, reasonably managing and controlling the liquidity risk of the securities market plays a crucial role in ensuring financial security, maintaining the stability of the financial system and promoting the healthy development of the economy.

[0003] As an important part of the securities market, the stock market has higher trading transparency and more easily accessible data compared with the bond market. These characteristics make the assessment and monitoring of the liquidity risk of the stock market more direct and effective. In related technologies, traditional statistical analysis methods are still adopted, or the liquidity risk is measured only based on the information of the stock itself without considering external information, resulting in a problem of poor accuracy in measuring stock risk. Summary of the Invention

[0004] The purpose of the present application is to provide a method, device, equipment, medium and product for measuring the liquidity risk of the securities market, which can improve the accuracy of measuring the liquidity risk of stocks in the securities market.

[0005] To achieve the above purpose, the present application provides the following solutions:

[0006] In the first aspect, the present application provides a method for measuring the liquidity risk of the securities market, including:

[0007] Obtain the liquidity data of the stock to be measured; wherein, the liquidity data of the stock to be measured includes the investment relationship diagram of listed companies, news information and stock information data within a preset time period, and the preset time period includes a plurality of consecutive time nodes;

[0008] Based on the liquidity data of the stock to be measured, use the trained stock liquidity risk measurement model to measure the stock liquidity risk of each listed company, and obtain the stock liquidity risk index of each listed company at the next time node; wherein, the stock liquidity risk measurement model includes a graph isomorphism neural network, a directed graph neural network, a recurrent neural network and a fusion unit.

[0009] Further, obtain the stock liquidity data to be measured, specifically including:

[0010] Remove or fill in missing values, remove outliers, process duplicate data, and normalize the stock information data;

[0011] Perform text cleaning, text normalization, and semantic enhancement on the news information;

[0012] Perform missing value processing, feature normalization, and discretization on the investment relationship graph of listed companies;

[0013] Based on the processed investment relationship graph of listed companies, news information, and stock information data, obtain the stock liquidity data to be measured.

[0014] Further, based on the stock liquidity data to be measured, use the trained stock liquidity risk measurement model to measure the stock liquidity risk of each listed company, and obtain the stock liquidity risk index of each listed company at the next time node, specifically including:

[0015] Input the investment relationship graph of listed companies into the graph isomorphism neural network to obtain multiple local graph features of the investment relationship; among them, the local graph features of the investment relationship include the features of multiple nodes, and the nodes include listed company nodes, non-listed company nodes, and individual nodes, and the local graph features of the investment relationship characterize the investment relationship between a listed company and its shareholders;

[0016] Construct a news directed graph according to the news information; among them, the news directed graph includes multiple central nodes; the central nodes are listed company nodes;

[0017] Input the news directed graph and the stock information data into the directed graph neural network to obtain the global graph features of the stock liquidity risk among listed companies; among them, the global graph features include the features of multiple central nodes, the content attribute of the central node features is the news stock features at each time node, and the global graph features characterize the investment relationship among listed companies;

[0018] Input the time series composed of the news stock features into the recurrent neural network to obtain the stock liquidity risk features of each listed company at the next time node;

[0019] Based on the stock liquidity risk features of each listed company at the next time node and the local graph features of the investment relationship, use a fusion unit to obtain the stock risk liquidity index of each listed company at the next time node.

[0020] Further, the news information includes news content and news theme;

[0021] Among them, constructing a news directed graph according to the news information specifically includes:

[0022] Based on the news content and news theme, using a large language model to extract the positive or negative correlation relationships between listed companies involved in the news information;

[0023] Determine the news directed graph according to the positive or negative correlation relationship.

[0024] Furthermore, the recurrent neural network is composed of multiple GRU sequence predictors;

[0025] Among them, inputting the time series composed of the news stock features into the recurrent neural network to obtain the stock liquidity risk features of each listed company at the next time node, specifically including:

[0026] Based on the time series composed of the news stock features of each listed company, using the corresponding GRU sequence predictor to measure the stock liquidity risk features of the listed company at the next time node, and obtaining the stock liquidity risk features of each listed company at the next time node.

[0027] Furthermore, inputting the listed company investment relationship graph into the graph isomorphism neural network to obtain multiple investment relationship local graph features, including:

[0028] Obtain multiple investment relationship local graphs according to the listed company investment relationship graph; among them, the investment relationship local graph includes multiple nodes and a central node;

[0029] Initialize the structural attributes of each node in the investment relationship local graph; among them, the structural attributes include: node degree, node type, and the distance from each node to the central node;

[0030] According to the structural attributes, perform conversion using an embedding layer to obtain the features of each node;

[0031] Use the graph isomorphism neural network to update the features of the nodes to obtain multiple investment relationship local graph features.

[0032] In a second aspect, the present application provides a securities market liquidity risk measurement device, including:

[0033] An acquisition module, configured to acquire the stock liquidity data to be measured; among them, the stock liquidity data to be measured includes the listed company investment relationship graph, news information, and stock information data within a preset time period, and the preset time period includes multiple consecutive time nodes;

[0034] A measurement module, configured to measure the stock liquidity risk of each listed company based on the to-be-measured stock liquidity data by using a trained stock liquidity risk measurement model, and obtain the stock liquidity risk indicators of each listed company at the next time node; wherein, the stock liquidity risk measurement model includes a graph isomorphism neural network, a directed graph neural network, a recurrent neural network, and a fusion unit.

[0035] In a third aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the above-mentioned method for measuring the liquidity risk of the securities market.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned method for measuring the liquidity risk of the securities market is implemented.

[0037] In a fifth aspect, the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the above-mentioned method for measuring the liquidity risk of the securities market is implemented.

[0038] According to the specific embodiments provided by the present application, the following technical effects are disclosed in the present application:

[0039] The present application provides a method, device, equipment, medium and product for measuring the liquidity risk of the securities market. The to-be-measured stock liquidity data is obtained, and the to-be-measured stock liquidity data includes the investment relationship graph, news information and stock information data of listed companies within a preset time period. The trained stock liquidity risk measurement model is used to process the investment relationship graph, news information and stock information data of listed companies, and the stock liquidity risk indicators of each listed company are obtained. The stock liquidity risk measurement model respectively considers the influence of news information and market investment relationship on the stock liquidity risk, and combines the historical stock information data to measure the stock liquidity risk indicators. This measurement method fully considers the influence of external information such as market investment relationship and news information on the stock liquidity risk, and improves the accuracy of the stock liquidity risk measurement. Description of the Drawings

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1It is an application environment diagram of a method for measuring the liquidity risk of the securities market in an embodiment of the present application;

[0042] Figure 2 It is a schematic flowchart of a method for measuring the liquidity risk of the securities market provided in an embodiment of the present application;

[0043] Figure 3 It is a schematic diagram of the functional modules of a device for measuring the liquidity risk of the securities market provided in an embodiment of the present application;

[0044] Figure 4 It is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Detailed implementation manners

[0045] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0046] To make the objectives, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0047] The method for measuring the liquidity risk of the securities market provided in the embodiments of the present application can be applied to an application environment as shown in Figure 1 . Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, placed in the cloud or on other servers. The terminal 102 can send the stock liquidity data to be measured to the server 104. After receiving the stock liquidity data to be measured, for the stock liquidity data to be measured, the server 104 measures the stock liquidity risk of each listed company based on the stock liquidity data to be measured by using the trained stock liquidity risk measurement model, and obtains the stock liquidity risk indicators of each listed company at the next time node. The server 104 can feedback the obtained stock liquidity risk indicators of each listed company at the next time node to the terminal 102. In addition, in some embodiments, the market liquidity risk measurement method can also be implemented by the server 104 or the terminal 102 alone. For example, the terminal 102 can directly process the stock liquidity data to be measured, or the server 104 can obtain the stock liquidity data to be measured from the data storage system and process the stock liquidity data to be measured.

[0048] Among them, the terminal 102 can be, but is not limited to, various desktop computers, laptop computers, smart phones, and tablet computers. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0049] In an exemplary embodiment, as Figure 2 shown, a method for measuring the liquidity risk of the securities market is provided. This method is executed by a computer device, and specifically can be executed alone by a computer device such as a terminal or a server, or can be jointly executed by a terminal and a server. In the embodiment of the present application, taking this method applied to Figure 1 the server 104 in it as an example for illustration, it includes the following steps 201 to step 202. Among them:

[0050] Step 201, obtain the stock liquidity data to be measured; among them, the stock liquidity data to be measured includes the investment relationship diagram of listed companies, news information, and stock information data within a preset time period, and the preset time period includes a plurality of consecutive time nodes.

[0051] Step 202, based on the stock liquidity data to be measured, use the trained stock liquidity risk measurement model to measure the stock liquidity risk of each listed company, and obtain the stock liquidity risk index of each listed company at the next time node; among them, the stock liquidity risk measurement model includes a graph isomorphism neural network, a directed graph neural network, a recurrent neural network, and a fusion unit.

[0052] Implementing the above steps 201 to 202 improves the accuracy of stock liquidity risk measurement.

[0053] In an exemplary embodiment, step 201 specifically includes:

[0054] Remove or fill in missing values, remove outliers, process duplicate data, and perform normalization processing on the stock information data; among them, the stock information data obtains the daily stock data through the Tushare data interface.

[0055] Perform text cleaning, text normalization, and semantic enhancement processing on the news information; among them, the news information is obtained by querying the daily consultation provided by the Choice financial terminal.

[0056] Perform missing value processing, feature normalization, and discretization processing on the investment relationship diagram of listed companies; among them, the investment relationship diagram of listed companies can obtain corresponding data through the API (Application Programming Interface) data interface of an open platform.

[0057] Based on the processed investment relationship graph, news information, and stock information data of listed companies, the stock liquidity data to be measured is obtained.

[0058] In an exemplary embodiment, step 202 specifically includes steps 301 - 305:

[0059] Step 301: Input the investment relationship graph of the listed company into the graph isomorphism neural network to obtain multiple local graph features of the investment relationship.

[0060] The local graph features of the investment relationship include the features of multiple nodes. The nodes include listed company nodes, unlisted company nodes, and individual nodes. The local graph features of the investment relationship represent the investment relationship between a listed company and its shareholders. Then, step 301 specifically includes steps 401 - 404:

[0061] Step 401: Obtain multiple local graphs of the investment relationship according to the investment relationship graph of the listed company. Among them, each local graph of the investment relationship includes multiple nodes and a central node, and the central node is a listed company node.

[0062] Step 402: Initialize the structural attributes of each node in the local graph of the investment relationship. The structural attributes include: node degree, node type, and the distance from each node to the central node.

[0063] Step 403: According to the structural attributes, use an embedding layer for transformation to obtain the features of each node.

[0064] Step 404: Use the graph isomorphism neural network to update the features of the nodes to obtain multiple local graph features of the investment relationship.

[0065] Specifically, contrastive learning is used to train the graph isomorphism neural network. During the training process, the InfoNCE loss is used as the objective to optimize the parameters of the graph isomorphism neural network. The training process includes:

[0066] Collect N training samples, with each local investment relationship graph as a training sample. For each training batch, prepare 1 positive sample pair and N - 1 negative sample pairs. A positive sample pair refers to two representations generated from the same local investment relationship graph, and positive sample pairs are considered similar; a negative sample pair refers to a pair of representations generated from different local investment relationship graphs, and negative sample pairs are considered dissimilar. Input the local investment relationship graphs of each batch into the graph isomorphism neural network twice to obtain the Query representation and Key representation of the local investment relationship graphs. Therefore, use the Query representation and Key representation of the same local investment relationship graph as the positive sample pair, and use the representations of different local investment relationship graphs to construct negative sample pairs. Furthermore, within a training batch of N training samples, obtain 1 positive sample pair and N - 1 negative sample pairs; finally, adopt the InfoNCE loss as the objective to optimize the parameters of the graph isomorphism neural network.

[0067] Step 302: Construct a news directed graph according to the news information.

[0068] If the news information includes news content and news theme, then step 302 specifically includes:

[0069] Based on the news content and news theme, use a large language model to extract the positive and negative correlation relationships between the listed companies involved in the news information; determine the news directed graph according to the positive or negative correlation relationship; wherein, the news directed graph includes multiple central nodes; the central nodes are listed company nodes.

[0070] Specifically, news information belonging to one company often mentions several other companies, and there may be a relationship between this company and any of the companies mentioned again in the news information, such as a partnership.

[0071] For example, if the news information reports that Company A and Company B jointly invested in a failed project, it reflects that there may be a certain potential risk between Company A and Company B at the same time; then, for the companies that exist in the same news information, connect them to construct a news graph to indicate the risk association between Company A and Company B.

[0072] In this embodiment, obtain the news information of each company within a certain period of time programmatically. If the news information mentions different companies (for example, contains the stock code of Company H), then regard such news information as the news information belonging to Company H.

[0073] For example, assume that a news information belonging to Company K mentions Company K once and Company G twice. Then, for Company K, it can be derived that the directed connection from Company K to Company G is 2.

[0074] If a news item is organized under Company A and mentions Company B, then the news item is considered to belong to Company A, and there is a directed link from A to B, denoted as (A, B). If Company B is cited multiple times in the same news item, each citation increases the cumulative weight of the directed edge. Additionally, self-citations are not counted. Thus, if a citation of Company A appears in a news item belonging to Company A, that citation is ignored.

[0075] Step 303: Input the news directed graph and the stock information data into a directed graph neural network to obtain the global graph features of the stock liquidity risk among listed companies.

[0076] Among them, the global graph features include multiple central node features. The content attribute of the central node features is the news stock features at each time node, and the global graph features characterize the investment relationships among listed companies.

[0077] DiGCN (Digraph Inception Convolutional Networks) utilizes the internal connection between the Laplacian distribution of the graph and the stationary distribution of PageRank to extend spectral-based graph convolution to directed graphs and can still perform convolution when the graph is not strongly connected.

[0078] In DiGCN, the news directed graph convolution operation is mainly based on two matrices: the simplified Laplacian matrix L appr and the k-hop proximity matrix P (k)DiGCN first introduces an auxiliary node as the teleport node of personalized PageRank, which effectively simplifies the calculation process of the Laplacian matrix, maintains the sparsity of the matrix, and significantly reduces the demand for computing resources. The convolution output obtained based on the simplified Laplacian matrix is used as the convolution result of one scale. The output of this scale takes into account the relationships of all central nodes in the news directed graph. Through the symmetrization process of the Laplacian matrix, the global structural information in the news directed graph can be captured. Next, the k-order proximity matrix is used to describe the k-order proximity relationship between central nodes in the news directed graph. Through the matrix of the k-order proximity relationship, the receptive field of the central node can be expanded, thereby learning multi-scale features. The convolution operation of the k-order proximity matrix mainly captures multi-scale local structural information. Through the k-order proximity matrix, the local proximity relationship of each listed company node can be better processed. On this basis, based on different k-order proximity matrices, multi-scale directed graph convolution operations with different k values are designed, and an Inception block is constructed by fusing the convolution results based on the Laplacian and the convolution results of the k-order proximity matrices at different scales, integrating scale features at different levels to form a richer representation of the listed company nodes. Multiple Inception blocks are stacked to form a multi-layer network structure, and the output of each layer is used as the input of the next layer to gradually extract higher-level features to more effectively learn the complex patterns in the news directed graph.

[0079] Therefore, based on the news directed graph, graph convolution operations can be used to capture the complex relationships between each core node, including the information of the neighbor core nodes of the central node, so as to learn the vector representation of the listed company nodes.

[0080] All listed companies share a news directed graph to more effectively spread market information constructed based on news information. Specifically, at each time step t, the news directed graph spreads the stock data of the t-th day corresponding to each listed company to obtain the feature representation of the t-th day corresponding to all listed companies, that is, the news stock features of each listed company. This representation learns the news information relationship and its stock data features between listed companies.

[0081] Step 304: Input the time series composed of the news stock features into a recurrent neural network to obtain the stock liquidity risk features of each listed company at the next time node. The recurrent neural network consists of multiple GRU (Gated Recurrent Unit) sequence predictors.

[0082] Specifically, based on the time series composed of the news stock characteristics of each listed company, the corresponding GRU sequence predictor is used to measure the stock liquidity risk characteristics of the listed company at the next time node, and the stock liquidity risk characteristics of each listed company at the next time node are obtained.

[0083] Each listed company has its own corresponding GRU sequence predictor to learn its unique liquidity risk prediction. Specifically, for listed company i, there are its own relevant GRU update gates and its own relevant GRU reset gates. Among them, each time step t is a time node. The input vector of the listed company on the t-th day consists of the news stock characteristics of the listed company on the t-th day. Based on the news stock characteristics of the listed company on the t-th day and the previous hidden state adjusted by the GRU reset gate, a new hidden state candidate can be obtained. Finally, GRU determines how much of the old state to retain and how much new state to introduce through the update gate to form the current state, that is, the stock liquidity risk characteristics at the t+1 moment.

[0084] Step 305, based on the stock liquidity risk characteristics of each listed company at the next time node and the characteristics of the local investment relationship graph, a fusion unit is used to obtain the stock risk liquidity index of each listed company at the next time node.

[0085] Specifically, the stock liquidity risk characteristics of each listed company on the (t + 1)-th day obtained are fused with the characteristics of the local investment relationship graph, and the investment relationship information of the listed company is selectively obtained. First, the attention weights of the characteristics of the local investment relationship graph and the stock liquidity risk characteristics are calculated, and then they are normalized through the softmax function to obtain the fusion vector of the central node. The fusion vector of the central node is spliced through a fully connected layer to obtain the stock liquidity index of each listed company at the next time node.

[0086] Finally, the directed graph neural network and the recurrent neural network are trained. Using the stock information data of each listed company in the news directed graph as samples and the corresponding stock liquidity index of each listed company as labels, the SmoothL1Loss smooth L1 loss function is used to update the parameters of the directed graph neural network and the recurrent neural network.

[0087] Based on the same inventive concept, the embodiments of the present application also provide a device for realizing the above-mentioned securities market liquidity risk measurement. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, for the specific limitations in the following one or more embodiments of the securities market liquidity risk measurement device, reference can be made to the limitations on the securities market liquidity risk measurement method in the above text, which will not be repeated here.

[0088] In an exemplary embodiment, as Figure 3As shown, a securities market liquidity risk measurement device is provided, including:

[0089] An acquisition module 31, configured to acquire the liquidity data of the stocks to be measured; wherein, the liquidity data of the stocks to be measured includes the investment relationship diagram of listed companies, news information, and stock information data within a preset time period, and the preset time period includes a plurality of consecutive time nodes.

[0090] A measurement module 32, configured to measure the stock liquidity risk of each listed company based on the liquidity data of the stocks to be measured, using the trained stock liquidity risk measurement model, to obtain the stock liquidity risk indicators of each listed company at the next time node; wherein, the stock liquidity risk measurement model includes a graph isomorphism neural network, a directed graph neural network, a recurrent neural network, and a fusion unit.

[0091] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal, and its internal structure diagram can be as Figure 4 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the liquidity data of the stocks to be measured. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a method for measuring the liquidity risk of the securities market.

[0092] Those skilled in the art can understand that Figure 4 the structure shown in

[0093] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0094] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0095] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0096] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0097] Those of ordinary skill in the art can understand that all or part of the processes of implementing the above method embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium, and when the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memories can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0098] In each of the embodiments provided in the present application, the database involved may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on a blockchain, etc., and is not limited thereto. In each of the embodiments provided in the present application, the processor may be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., and is not limited thereto.

[0099] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0100] Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. A method for measuring liquidity risk in a securities market, characterized in that: The securities market liquidity risk measurement method includes: Acquire the stock liquidity data to be measured; wherein the stock liquidity data to be measured includes an investment relationship diagram of listed companies, news information and stock information data within a preset time period, and the preset time period includes multiple continuous time nodes; Based on the stock liquidity data to be measured, the trained stock liquidity risk measurement model is used to measure the stock liquidity risk of each listed company to obtain the stock liquidity risk index of each listed company at the next time node; wherein, the stock liquidity risk measurement model includes a graph isomorphic neural network, a directed graph neural network, a recurrent neural network and a fusion unit.

2. The method for measuring securities market liquidity risk according to claim 1, characterized in that: Obtain the liquidity data of the stocks to be measured, including: Remove or fill missing values, remove outliers, process duplicate data, and normalize stock information data; Perform text cleaning, text normalization and semantic enhancement on news information; Perform missing value processing, feature normalization and discretization processing on the investment relationship diagram of listed companies; Based on the processed investment relationship diagram of listed companies, news information and stock information data, the stock liquidity data to be measured is obtained.

3. The method for measuring securities market liquidity risk according to claim 1, characterized in that: Based on the stock liquidity data to be measured, the stock liquidity risk of each listed company is measured using the trained stock liquidity risk measurement model to obtain the stock liquidity risk index of each listed company at the next time point, specifically including: Inputting the listed company investment relationship graph into a graph isomorphism neural network to obtain a plurality of investment relationship local graph features; wherein the investment relationship local graph features include features of a plurality of nodes, the nodes including listed company nodes, non-listed company nodes and individual nodes, and the investment relationship local graph features represent the investment relationship between a listed company and its shareholders; Constructing a news directed graph according to the news information; wherein the news directed graph includes a plurality of central nodes; the central nodes are nodes of listed companies; The news directed graph and the stock information data are input into a directed graph neural network to obtain a global graph feature of stock liquidity risk among listed companies; wherein the global graph feature includes a plurality of central node features, the content attributes of the central node features are news stock features at each time node, and the global graph feature represents the investment relationship between listed companies; Input the time series composed of the news stock features into the recurrent neural network to obtain the stock liquidity risk features of each listed company at the next time node; Based on the stock liquidity risk characteristics of each listed company at the next time node and the characteristics of the local graph of investment relations, a fusion unit is used to obtain the stock risk liquidity index of each listed company at the next time node.

4. The method for measuring securities market liquidity risk according to claim 3, characterized in that: The news information includes news content and news subject; The step of constructing a news directed graph according to the news information specifically includes: Based on the news content and news topics, using a large language model to extract positive or negative correlations between listed companies involved in the news information; A news directed graph is determined according to the positive or negative correlation.

5. The method for measuring securities market liquidity risk according to claim 3, characterized in that: The recurrent neural network is composed of multiple GRU sequence predictors; The time series composed of the news stock features is input into the recurrent neural network to obtain the stock liquidity risk features of each listed company at the next time node, specifically including: Based on the time series composed of the news stock features of each listed company, the corresponding GRU sequence predictor is used to measure the stock liquidity risk features of the listed company at the next time node, and the stock liquidity risk features of each listed company at the next time node are obtained.

6. The method for measuring securities market liquidity risk according to claim 3, characterized in that: The investment relationship graph of the listed company is input into the graph isomorphism neural network to obtain multiple investment relationship local graph features, including: According to the listed company investment relationship graph, a plurality of investment relationship partial graphs are obtained; wherein the investment relationship partial graphs include a plurality of nodes and a central node; Initializing the structural attributes of each node in the local investment relationship graph; wherein the structural attributes include: node degree, node type and the distance from each node to the central node; According to the structural attributes, an embedding layer is used to perform conversion to obtain the characteristics of each node; A graph isomorphic neural network is used to update the features of the nodes to obtain multiple investment relationship local graph features.

7. A device for measuring liquidity risk in a securities market, characterized in that: The securities market liquidity risk measurement device comprises: An acquisition module is used to acquire the stock liquidity data to be measured; wherein the stock liquidity data to be measured includes an investment relationship diagram of listed companies, news information and stock information data within a preset time period, and the preset time period includes multiple continuous time nodes; The measurement module is used to measure the stock liquidity risk of each listed company based on the stock liquidity data to be measured, using the trained stock liquidity risk measurement model to obtain the stock liquidity risk index of each listed company at the next time node; wherein the stock liquidity risk measurement model includes a graph isomorphic neural network, a directed graph neural network, a recurrent neural network and a fusion unit.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for measuring liquidity risk in the securities market according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for measuring securities market liquidity risk according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for measuring securities market liquidity risk according to any one of claims 1 to 6 is implemented.

Images