Model-based data processing method and device, equipment, medium and product

Abstract

The embodiment of the invention provides a model-based data processing method and device, equipment, a medium and a product, and belongs to the technical field of data processing. The method comprises the following steps: acquiring market economic data and institution data of a financial institution; determining a first influence factor according to the market economic data; determining a second influence factor according to the first influence factor and the mechanism data; determining a third influence factor according to the second influence factor and the market economic data; the third influence factor is used for representing the influence degree of the market economic environment on the asset return rate of the financial institution; according to the first influence factor, the second influence factor and the third influence factor, determining an asset demand quantity of the financial institution and a weight vector of optimal asset configuration; the weight vector is used for representing the optimal proportion of different asset categories in the financial institution investment portfolio. According to the embodiment of the invention, the accuracy of determining the asset demand quantity and the optimal asset configuration can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a model-based data processing method, apparatus, device, medium, and product. Background Technology

[0002] Asset adequacy management is a core element of sound business operations for insurance companies. In today's complex and volatile financial market environment, insurance companies must accurately determine their asset allocation strategies to ensure asset adequacy and achieve sound business operations.

[0003] However, with the increasing complexity of the insurance market environment and the continuous improvement of regulatory requirements, traditional methods have gradually revealed their limitations. Traditional methods typically employ simple statistical models or linear regression models, which are difficult to capture market fluctuations, the nonlinear relationships between risk factors, and the impact of extreme events. This results in inaccurate predictions of asset demand, and consequently, a lack of accuracy in determining asset allocation strategies. Summary of the Invention

[0004] The main objective of this application is to propose a model-based data processing method, apparatus, device, medium, and product, which aims to improve the accuracy of determining asset demand and optimal asset allocation.

[0005] To achieve the above objectives, a first aspect of this application proposes a model-based data processing method, the method comprising:

[0006] Obtain market economy data and institutional data from financial institutions;

[0007] Based on market economy data, the first influencing factor is determined; the first influencing factor is used to characterize the degree of impact of the market economic environment on the asset demand of financial institutions.

[0008] Based on the first influencing factor and institutional data, the second influencing factor is determined; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the asset demand of financial institutions.

[0009] Based on the second influencing factor and market economy data, the third influencing factor is determined; the third influencing factor is used to characterize the degree of impact of the market economic environment on the asset return rate of financial institutions.

[0010] Based on the first, second, and third influencing factors, the asset demand of financial institutions and the weight vector of the optimal asset allocation are determined; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

[0011] To achieve the above objectives, a second aspect of this application provides a model-based data processing apparatus, the apparatus comprising:

[0012] The acquisition module is used to acquire market economic data and institutional data from financial institutions;

[0013] The first determining module is used to determine a first influencing factor based on the market economic data; the first influencing factor is used to characterize the degree of influence of the market economic environment on the asset demand of the financial institution.

[0014] The second determining module is used to determine a second influencing factor based on the first influencing factor and the institutional data; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the financial institution's asset demand.

[0015] The third determining module is used to determine a third influencing factor based on the second influencing factor and the market economic data; the third influencing factor is used to characterize the degree of influence of the market economic environment on the asset return rate of the financial institution.

[0016] The fourth determining module is used to determine the asset demand of the financial institution and the weight vector of the optimal asset allocation based on the first influencing factor, the second influencing factor and the third influencing factor; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

[0017] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0018] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0019] To achieve the above objectives, a fifth aspect of the present application provides a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform the method described in the first aspect above.

[0020] This application proposes a model-based data processing method, apparatus, equipment, medium, and product. By acquiring market economic data and institutional data from financial institutions, it determines a first influencing factor characterizing the impact of the market economic environment on the asset demand of financial institutions. Further, combining the first influencing factor and institutional data, it determines a second influencing factor characterizing the marginal impact of changes in risk exposure on the asset demand of financial institutions. Based on this, using the second influencing factor and market economic data, it determines a third influencing factor characterizing the impact of the market economic environment on the asset return rate of financial institutions. Finally, based on these three influencing factors, it determines the asset demand of financial institutions and the weight vector for optimal asset allocation, thereby achieving precise determination of asset allocation strategies. Thus, by comprehensively considering multiple dimensions such as the macroeconomic environment, market volatility, risk exposure, and institutional data, it can more comprehensively and accurately reflect the impact of various factors on asset demand. This improves the accuracy of determining asset demand and optimal asset allocation. Attached Figure Description

[0021] Figure 1 This is one of the flowcharts of the model-based data processing method provided in the embodiments of this application;

[0022] Figure 2 This is the second flowchart of the model-based data processing method provided in the embodiments of this application;

[0023] Figure 3 This is the third flowchart of the model-based data processing method provided in the embodiments of this application;

[0024] Figure 4 This is the fourth flowchart of the model-based data processing method provided in the embodiments of this application;

[0025] Figure 5 This is a schematic diagram of the structure of the model-based data processing device provided in the embodiments of this application;

[0026] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0028] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0030] First, let's analyze some of the terms used in this application:

[0031] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0032] Asset adequacy management is a core component of sound business operations for insurance companies. Traditional methods rely on static models and historical data, ensuring asset adequacy through regular assessments of asset needs and allocation. However, with the increasing complexity of the insurance market and stricter regulatory requirements, traditional methods have revealed numerous problems: Firstly, they rely on a single data source, depending solely on historical financial data and static regulatory indicators, lacking comprehensive analysis of multi-dimensional data such as the macroeconomic environment, market fluctuations, risk exposure, and customer behavior, resulting in insufficient comprehensiveness and accuracy in asset adequacy assessments. Secondly, their predictive capabilities are limited, often employing simple statistical or linear regression models that struggle to capture market fluctuations, non-linear relationships between risk factors, and the impact of extreme events, leading to low accuracy and reliability in asset demand forecasting. Thirdly, they lack dynamic adjustment mechanisms, typically assessing and adjusting asset adequacy quarterly or annually, making it difficult to monitor market changes and dynamically adjust risk exposure in real time, resulting in insufficient flexibility and adaptability in asset allocation. Fourthly, current technologies are computationally inefficient, relying on manual operation and experience-based judgment, which fails to meet the demands of rapid response and precise management in the modern insurance market. These problems may lead to asset shortages or surpluses for insurance companies during market fluctuations or risk events, impacting their market competitiveness and risk control capabilities.

[0033] Based on this, embodiments of this application provide a model-based data processing method, apparatus, device, medium, and product, aiming to improve the accuracy of determining asset demand and optimal asset allocation.

[0034] The model-based data processing methods, apparatuses, devices, media, and products provided in this application are specifically described through the following embodiments. First, the model-based data processing method in this application is described.

[0035] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0036] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0037] The model-based data processing method provided in this application relates to the field of data processing technology. This model-based data processing method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the model-based data processing method, but is not limited to the above forms.

[0038] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0039] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0040] Figure 1 This is an optional flowchart of the model-based data processing method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S105.

[0041] S101, Obtain market economy data and institutional data from financial institutions;

[0042] S102, Based on the market economic data, determine the first influencing factor; the first influencing factor is used to characterize the degree of influence of the market economic environment on the asset demand of the financial institution.

[0043] S103, Based on the first influencing factor and the institutional data, determine the second influencing factor; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the financial institution's asset demand.

[0044] S104, Based on the second influencing factor and the market economic data, determine the third influencing factor; the third influencing factor is used to characterize the degree of impact of the market economic environment on the asset return rate of the financial institution.

[0045] S105, Based on the first influencing factor, the second influencing factor, and the third influencing factor, determine the asset demand of the financial institution and the weight vector of the optimal asset allocation; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

[0046] As illustrated in S101 to S105 of this application embodiment, by acquiring market economic data and institutional data of financial institutions, a first influencing factor is determined that characterizes the degree of impact of the market economic environment on the asset demand of financial institutions. Further, combining the first influencing factor and institutional data, a second influencing factor is determined to characterize the marginal impact of changes in risk exposure on the asset demand of financial institutions. Based on this, using the second influencing factor and market economic data, a third influencing factor is determined to characterize the degree of impact of the market economic environment on the asset return rate of financial institutions. Finally, based on these three influencing factors, the asset demand of financial institutions and the weight vector of the optimal asset allocation are determined, thereby achieving accurate determination of the asset allocation strategy. Thus, by comprehensively considering multiple dimensions such as the macroeconomic environment, market volatility, risk exposure, and institutional data, the impact of various factors on asset demand can be reflected more comprehensively and accurately. This improves the accuracy of determining asset demand and the optimal asset allocation.

[0047] In step S101 of some embodiments, market economic data refers to various types of data that reflect the macroeconomic situation, financial market fluctuations, and changes in the policy environment. It serves as a crucial external basis for financial institutions' asset allocation decisions. This includes macroeconomic indicators such as GDP growth rate, inflation rate, interest rates, and exchange rates, used to measure the overall economic development level and price fluctuations; market data such as stock market indices, bond yields, and market volatility indices, which directly reflect price changes in financial assets and market risks; and policy variables such as monetary policy adjustments, changes in fiscal policy, and industry regulatory requirements, reflecting the impact of external policies on the economy and financial markets. These data collectively constitute the external environmental foundation for financial institutions' asset demand analysis, providing multi-dimensional information support for determining the extent to which the market economic environment affects asset demand.

[0048] Financial institutions refer to organizations that are legally established and operate financial businesses. In this application scenario, they mainly refer to insurance companies. Their core businesses include undertaking risk protection (such as life insurance and property insurance underwriting) and conducting asset investment management to achieve sound operation and profit goals. As the main implementers of asset allocation strategies, the internal data of financial institutions is a key input for analyzing the impact of risk exposure on asset demand. Their operating conditions and risk characteristics directly determine the objectives and constraints of asset allocation.

[0049] Institutional data refers to the collection of data generated internally by financial institutions that reflects their operating conditions, risk characteristics, and business operations. It is core information reflecting the unique characteristics of an enterprise in asset demand analysis. This data can include financial and operational data, such as premium income, loss ratio, asset-liability structure, and asset adequacy ratio, used to assess the enterprise's financial health and business scale; it can also include risk exposure data, such as credit risk indicators and liquidity coverage ratio, quantifying the degree of exposure to various risks faced by the enterprise; it can also involve customer and business data, such as surrender rates, insurance preferences, and policy structure, reflecting the impact of customer behavior on asset demand structure; and it can also cover historical asset management data, such as past asset allocation plans and stress test results, providing historical references for model training and strategy optimization.

[0050] Optionally, in one feasible implementation of this application, for market economic data, professional data interfaces and platforms can be used to obtain macroeconomic indicators such as GDP growth rate, inflation rate, interest rate, and exchange rate from macroeconomic databases, as well as market data such as stock market index, bond yield, and market volatility index. At the same time, policy variable data such as monetary policy adjustments and fiscal policy changes can be collected to comprehensively reflect the macroeconomic operating trend, financial market fluctuations, and policy environment changes.

[0051] For institutional data from financial institutions, extraction can be performed from their internal databases, including financial and operational data, risk exposure data, customer and business data, and historical asset management data. This provides a comprehensive overview of the financial institution's operating status, risk characteristics, and business operations. This approach enables the effective acquisition of market economic data and institutional data, laying a data foundation for subsequently determining various influencing factors and asset allocation strategies.

[0052] In step S102 of some embodiments, the first influencing factor is a quantitative indicator used to characterize the degree of influence of the market economic environment on the asset demand of financial institutions. The first influencing factor aims to measure the strength of the effect of external factors such as the macroeconomic situation, financial market volatility, and changes in the policy environment on the scale of asset demand of financial institutions. For example, when GDP growth accelerates, it may drive financial institutions to expand their asset allocation to match market demand. At this time, the change in the value of the first influencing factor can reflect the specific direction and degree of the impact of such macroeconomic factors on asset demand, providing a basis for subsequent analysis of risk exposure and asset allocation.

[0053] Asset demand refers to the scale of assets that financial institutions rationally allocate to achieve a balance between risk and return, meet regulatory requirements, and fulfill business development needs under specific market economic conditions and their own operating circumstances. It is not a fixed value but a dynamic variable influenced by multiple dimensions of factors, including the macroeconomic environment, the institution's own risk exposure, asset adequacy, and business structure. For example, when market interest rates decline, financial institutions may need to increase their allocation to fixed-income assets to maintain profitability, and asset demand will adjust accordingly. Conversely, if downward economic pressure increases, financial institutions may shrink their asset size to control risk. This indicator is one of the core parameters for financial institutions' asset allocation decisions and directly affects the determination of the optimal asset allocation weight vector.

[0054] Optionally, in one feasible implementation of this application, firstly, the market economic data obtained in S101 is preprocessed. Data cleaning techniques are used to remove outliers and missing values, normalization methods are used to unify data dimensions, and key features are extracted through dimensionality reduction techniques such as principal component analysis. Next, the processed data is input into a time series forecasting model. This model uses a gating mechanism to capture the temporal dependencies and nonlinear relationships between data. During the training process, historical market economic data is used as input, and the actual asset demand of financial institutions in the corresponding period is used as a label to optimize model parameters to fit the data features. Finally, based on the trained time series forecasting model, predictions are made, and the quantitative value of the impact of market economic data on asset demand is output, i.e., the first influencing factor. For example, by calculating the change in asset demand for each unit change in different economic variables, an influence coefficient matrix is ​​constructed, intuitively reflecting the driving effect of the macroeconomic environment on the scale of asset demand of financial institutions, providing a basic quantitative basis for subsequent analysis.

[0055] In step S103 of some embodiments, the second influencing factor is a quantitative indicator used to measure the impact of changes in risk exposure on the asset demand of financial institutions. The second influencing factor focuses on the effect of changes in the internal risk situation of financial institutions on asset demand. It can reflect the degree to which the asset demand is adjusted when the institution's exposure to market risk, credit risk, liquidity risk, etc. changes. It is a key parameter connecting the external economic environment and the internal risk characteristics of the institution, providing a quantitative basis for the risk dimension for subsequent determination of asset allocation strategies.

[0056] Changes in risk exposure refer to the dynamic changes in the degree of exposure to various risks faced by financial institutions during their operations, caused by factors such as market fluctuations, business restructuring, and changes in customer behavior. Specifically, this includes changes in market risk exposure (such as changes in potential losses to investment portfolios like stocks and bonds due to market price fluctuations), changes in credit risk exposure (such as increased risk exposure due to rising customer default probabilities or downgrades in debtor credit ratings), and changes in liquidity risk exposure (such as changes in liquidity pressure caused by exacerbated maturity mismatches or decreased asset liquidity). These changes directly affect a financial institution's risk tolerance and asset demand, and are key internal risk dynamic factors that need to be focused on when assessing asset demand.

[0057] The marginal impact refers to the degree to which a financial institution's asset demand changes when the risk exposure changes by one unit, assuming other conditions remain unchanged. For example, when credit risk exposure increases by 1%, the asset demand may need to increase by a certain proportion to cover potential losses; this proportion is the marginal impact.

[0058] Optionally, in one feasible implementation of this application, a gradient boosting tree can be used to determine the second influencing factor. Specifically, a weak learner (regression tree) can be iteratively constructed to gradually optimize the mapping relationship between changes in risk exposure and asset demand. First, multidimensional risk features are extracted from institutional data, and these features are standardized in conjunction with the first influencing factor to eliminate dimensional differences. Second, the standardized risk features and the first influencing factor are used as input variables, and historical changes in asset demand are used as labels to initialize a constant model. Subsequently, multiple regression trees are iteratively trained, with each tree fitting the residual between the current model's predicted value and the actual value. The tree structure is optimized using gradient descent to gradually reduce prediction errors. Finally, by integrating the prediction results of all regression trees, the marginal impact coefficient of the risk exposure variable is calculated to form the second influencing factor.

[0059] In step S104 of some embodiments, the third influencing factor is a quantitative indicator used to characterize the impact of the market economic environment on the asset returns of financial institutions. The third influencing factor aims to measure the corresponding changes in the returns of various assets (such as stocks, bonds, and alternative investments) of financial institutions when economic variables such as GDP growth rate, interest rates, exchange rates, and asset market volatility change. For example, when market interest rates decline, the prices of fixed-income assets may rise, thereby increasing the return on the asset portfolio. The third influencing factor can quantify this dynamic relationship between economic variables and returns, providing a key basis for predicting the returns of subsequent asset allocation strategies.

[0060] Return on assets (ROA) is the ratio of the income earned by a financial institution on various assets over a specific period (such as an annual or quarterly period) to the value of those assets. It is expressed as a percentage and is a core indicator for measuring asset profitability. It encompasses various forms of income, including interest income and dividends, and can be specifically categorized into stock returns and bond returns.

[0061] Optionally, in one feasible implementation of this application, firstly, the risk exposure characteristics reflected by the second influencing factor are integrated with market economic data to construct a multi-factor return prediction model. This model can employ a reinforcement learning framework, dynamically adjusting the weights of each factor through a policy gradient algorithm. Using historical data as input and asset returns as output, the model is trained to capture the nonlinear mapping relationship between economic variables and asset returns. For example, gated recurrent units can be used to process time-series data, and an attention mechanism can be used to automatically allocate the importance weights of each factor under different economic cycles. Secondly, Monte Carlo simulations are performed based on the trained model to generate thousands of possible economic scenario paths, simulating the distribution of returns for various assets under different market economic environments. Finally, sensitivity analysis is used to calculate the rate of change in asset returns for each unit change in economic variables, forming a third influencing factor matrix. For example, the impact of a 50 basis point increase in interest rates on bond portfolio returns and the impact of a 1% decrease in GDP growth on equity asset returns are calculated. This matrix not only quantifies the direct impact of the economic environment on asset returns but also captures the synergistic effects between factors through cross-term analysis, providing a comprehensive quantitative basis for return-risk quantification for subsequent asset allocation optimization.

[0062] In step S105 of some embodiments, the asset demand is the reasonable asset size that a financial institution needs to hold to achieve sound operation after comprehensively considering the market economic environment, its own risk exposure, and return objectives. For example, when the economy is in an expansionary phase (the first influencing factor shows an increase in asset demand), but the institution's credit risk exposure increases significantly (the second influencing factor amplifies asset demand), and the expected market return rate declines (the third influencing factor suppresses asset demand), the system will dynamically balance these three factors to calculate the total amount of assets that can both cover risk and meet return objectives. This indicator is dynamic and will be adjusted in real time according to the market environment and the institution's risk status, making it a core parameter for financial institutions' asset planning and liquidity management.

[0063] The optimal asset allocation weight vector refers to the combination of various asset investments that achieves the optimal balance between risk and return characteristics for a financial institution's portfolio, given a certain amount of asset demand. For example, when the third influencing factor indicates that equity assets have a high expected return in the current economic environment, but the second influencing factor suggests a large market risk exposure, the system will dynamically adjust the allocation ratio of assets such as stocks, bonds, and cash through a multi-objective optimization algorithm to maximize the investment return while meeting regulatory requirements. Each dimension of the weight vector represents a specific asset class (e.g., a weight of 0.3 for government bonds indicates that 30% of the assets are allocated to government bonds). The solution process considers both the impact of the macroeconomic environment and the institution's own risk tolerance, ultimately providing financial institutions with a scientific basis for asset allocation decisions.

[0064] Optionally, in one feasible implementation of this application, firstly, the first, second, and third influencing factors are integrated to construct an optimization function aimed at maximizing expected returns and minimizing portfolio risk. Next, multiple constraints are set, including asset adequacy constraints based on the first influencing factor (e.g., solvency adequacy ratio must meet regulatory requirements), risk limit constraints driven by the second influencing factor (e.g., risk exposure to a single asset class does not exceed 20% of total assets), and non-negativity and normalization constraints on asset weights. Subsequently, an adaptive particle swarm optimization algorithm is used to solve the model, iteratively updating particle positions (i.e., asset weight combinations) and using simulated annealing to avoid getting trapped in local optima. For example, during an economic upswing, the algorithm increases the weight of risky assets based on the first influencing factor while dynamically adjusting the risk exposure threshold through the second influencing factor to ensure asset adequacy. Finally, the output asset demand is obtained by multiplying the optimal weight vector by the institution's available asset size.

[0065] In one embodiment, such as Figure 2 As shown, based on market economy data, the first influencing factor is determined, including:

[0066] S201, Extracting time-series volatility indicators from market economic data; Time-series volatility indicators include at least one of stock index volatility, yield curve change rate, and macroeconomic surprise index.

[0067] S202: Input the time series fluctuation index into the time series prediction model, capture the nonlinear mapping relationship between the time series fluctuation indexes through the time series prediction model, and output the first influencing factor.

[0068] The time series prediction model includes the Long Short-Term Memory (LSTM) network.

[0069] Optionally, in the embodiments of this application, the time-series fluctuation index is a quantitative indicator used to describe the dynamic characteristics of market economic data changing over time, reflecting the fluctuation range, trend of change and degree of uncertainty of economic variables at different points in time.

[0070] Stock index volatility is an indicator that measures the degree of price fluctuation in the stock market, reflecting investors' expectations of market risk. As a time-series volatility indicator, stock index volatility is used to capture changes in risk sentiment in asset markets and is an important market factor affecting the demand for assets by financial institutions.

[0071] The yield curve change rate is an indicator that describes how the relationship between yields on bonds of different maturities changes over time. It reflects market expectations of future interest rate trends and changes in the macroeconomic environment. The yield curve change rate can reflect information such as monetary policy trends, inflation expectations, and economic growth prospects, and is an important reference for financial institutions to assess the value of fixed-income assets and adjust asset allocation.

[0072] The Macroeconomic Surprise Index measures the difference between actual economic data and market expectations, reflecting the unexpected nature of economic performance. This index can capture sudden changes and uncertainties in the macroeconomic environment.

[0073] Optionally, in one specific implementation of this application, firstly, multi-dimensional time-series volatility indicators are extracted from market economic data. Subsequently, the extracted time-series indicators are preprocessed. Z-score standardization is applied to eliminate the influence of dimensions, moving averages are used to smooth high-frequency noise, and non-stationary sequences are processed through differencing. The processed indicators are then used to construct an input sequence according to time windows (e.g., 60 trading days), with each window containing historical values ​​of three indicators, and the target variable being the rate of change in the asset demand of financial institutions during the same period.

[0074] Next, a Long Short-Term Memory Network (LSTM) model was constructed and trained. The model adopts a three-layer architecture: the input layer receives the preprocessed time-series index matrix, the hidden layer contains 128 LSTM units, and the output layer is a fully connected layer that generates single-value predictions. The forget gate controls the degree of retention of historical information, the input gate filters important features of the current input, and the output gate determines the final output value, effectively capturing the long-term dependencies and nonlinear relationships between indices.

[0075] In the prediction phase, the latest time-series volatility index window is input into the trained model. The LSTM processes the sequence information through a series of gating mechanisms: first, a forget gate filters out short-term noise in interest rate changes; then, an input gate strengthens the synergistic effect between stock volatility and the macroeconomic surprise index; finally, the output layer integrates this information to generate the first influencing factor. This factor quantifies the overall impact of the market economic environment on asset demand. For example, an output value of 0.8 indicates that when stock volatility increases by one standard deviation, asset demand is expected to increase by 0.8%.

[0076] In these alternative embodiments, the multi-dimensional indicator system comprehensively covers the uncertainties of the asset market, interest rate environment, and macroeconomy, enhancing the accuracy of the characterization of the market economic environment; the gating mechanism of the LSTM model effectively handles the long-term and short-term dependencies in time series data, accurately capturing the complex nonlinear dynamic correlations between economic variables; the output first influencing factor can quantify the degree of impact of the economic environment on asset demand, providing financial institutions with data-driven decision-making basis for dynamically adjusting asset allocation, and enhancing the foresight and adaptability of risk management.

[0077] In one embodiment, such as Figure 3 As shown, determining the second impact factor based on the first impact factor and the institutional data includes:

[0078] S301, Extract multidimensional risk characteristics from institutional data; the multidimensional risk characteristics include at least one of credit risk exposure, market risk sensitivity, and liquidity coverage ratio;

[0079] S302, the first influencing factor is used as the risk feature weight adjustment coefficient to weight the multidimensional risk features and obtain the weighted features;

[0080] S303 involves inputting weighted features into a random forest model, and then evaluating the marginal effect of changes in risk exposure through a decision tree integrated from the random forest model to obtain the second influencing factor.

[0081] Optionally, in the embodiments of this application, multidimensional risk characteristics refer to a set of quantitative indicators extracted from the internal data of financial institutions to comprehensively depict their risk status, which can reveal the potential risk exposures faced by financial institutions and their changing trends from different perspectives.

[0082] Credit risk exposure refers to the maximum potential loss a financial institution may suffer due to borrower default, and it is a core indicator for measuring credit risk. Credit risk exposure reflects the degree of risk exposure a financial institution has in its credit business and is an important basis for assessing its asset quality and soundness.

[0083] Market risk sensitivity refers to the degree to which a financial institution's asset or liability value is affected by market price fluctuations, and is used to measure the potential impact of market risk. By quantifying market risk sensitivity, financial institutions can assess the impact of changes in different market factors on their asset value, providing a reference for risk management and asset allocation.

[0084] Optionally, in one specific implementation of this application, firstly, multidimensional risk features are extracted from institutional data. Then, feature weighting is performed. The first influencing factor is used as a dynamic weight adjustment coefficient, adjusting the importance of each risk feature according to its value. For example, when the first influencing factor indicates an economic expansion, the weight of credit risk exposure is reduced while the weight of market risk sensitivity is increased; conversely, the opposite adjustment is made during an economic downturn. Specifically, this can be achieved through matrix operations: multiplying the initial risk feature matrix by the diagonal matrix formed by the first influencing factor yields a weighted feature matrix, allowing the importance of risk features to dynamically change with the economic environment.

[0085] Finally, the marginal effect is evaluated using a random forest model. Specifically, multiple decision trees are constructed: each tree randomly selects some features and samples, and recursively splits nodes using indicators such as Gini impurity until the purity of the leaf nodes reaches the target. For example, a decision tree might determine the impact path of changes in risk exposure on asset demand based on a combination of market risk sensitivity and liquidity coverage ratio. The prediction results of all decision trees are integrated, and the marginal contribution of each risk feature is determined by ranking the features by their importance. For example, the percentage change in asset demand for every 1% increase in credit risk exposure is calculated, ultimately forming the second influencing factor.

[0086] In these alternative embodiments, the multi-dimensional risk feature system comprehensively covers the main types of risks faced by financial institutions, enhancing the three-dimensional characterization of risk conditions; using the first influencing factor as the weight adjustment coefficient, it realizes the dynamic linkage between the external economic environment and internal risk features, making risk assessment more timely; the ensemble learning mechanism of the random forest model effectively handles nonlinear relationships, quantifies the marginal effect of each risk feature through decision tree splitting nodes, and improves the accuracy of factor calculation.

[0087] In one embodiment, such as Figure 4 As shown, based on the second influencing factor and market economy data, the third influencing factor is determined, including:

[0088] S401 extracts return characteristics from market economic data; return characteristics include economic growth rate, inflation rate, and the slope of the asset return curve.

[0089] S402, using the second influencing factor as a risk adjustment coefficient, corrects the return characteristics to obtain the corrected characteristics;

[0090] S403 modifies the input of features to a reinforcement learning model based on a self-attention mechanism, and captures the dynamic dependencies between features through the multi-head attention layer of the reinforcement learning model to obtain the third influencing factor.

[0091] Optionally, in the embodiments of this application, return characteristics refer to a set of economic variables extracted from market economic data to characterize the potential drivers of asset returns.

[0092] The economic growth rate is an indicator that measures the speed at which the total economy grows over a certain period. It reflects the expansion or contraction of the macroeconomy and is a core indicator for assessing the health of the economy.

[0093] The inflation rate refers to the rate of increase in the general price level over a given period, typically measured by the rate of change in the Consumer Price Index (CPI) or Producer Price Index (PPI). The inflation rate is used to characterize the impact of price level changes on the returns of financial institutions' asset allocation and is an important component of return characteristics.

[0094] The slope of the yield curve refers to the degree of difference between the returns of assets with different maturities. As a return characteristic, the slope of the yield curve is used to capture the impact of changes in the term structure of interest rates on the returns of financial institutions' portfolios.

[0095] Optionally, in one specific implementation of this application, firstly, return characteristics are extracted from market economic data. Then, a second influencing factor is used for risk adjustment. The second influencing factor is resolved into a risk-adjusted coefficient matrix, where each element corresponds to the marginal impact weight of a specific risk exposure change on the return characteristics. For example, if the second influencing factor shows that a 1% increase in market risk sensitivity leads to a 0.5% increase in asset demand, then a higher weight is assigned when adjusting the slope of the asset yield curve. Correction is achieved through matrix multiplication: the original return characteristic matrix is ​​multiplied element-wise by the risk-adjusted coefficient matrix to obtain a corrected characteristic matrix, making the return characteristics more closely reflect the actual risk-return relationship faced by financial institutions.

[0096] Finally, a reinforcement learning model based on a self-attention mechanism is constructed. The model adopts an Actor-Critic architecture, where the Actor network generates asset allocation strategies, and the Critic network evaluates the value of these strategies. In the feature processing layer, a multi-head attention mechanism is designed: the modified feature matrix is ​​mapped to three spaces—query (Q), key (K), and value (V). Attention weights are generated by calculating the similarity matrix between Q and K and applying the softmax function, then weighted and summed with the V matrix to form multiple parallel attention heads. For example, one attention head might focus on the correlation between economic growth rate and the slope of the asset return curve, while another head focuses on the long-term trend of inflation. By stacking multiple attention layers, the model automatically captures the dynamic dependencies between features at different time scales and economic cycles. During training, a policy gradient algorithm is used, with maximizing the risk-adjusted asset return as the objective function. Economic scenario paths are generated through Monte Carlo simulation, and network parameters are iteratively optimized. Ultimately, the gradient of the state-value function output by the model is the third influencing factor, quantifying the impact of the market economic environment on asset returns and providing accurate return prediction for subsequent asset allocation decisions.

[0097] In these alternative embodiments, the multi-dimensional return feature system comprehensively covers the macroeconomic, inflation, and interest rate environments, enhancing the accuracy of characterizing the drivers of asset returns; using the second influencing factor as the risk adjustment coefficient achieves dynamic coupling between changes in risk exposure and return features, improving the risk sensitivity of factor calculation; the multi-head architecture of the self-attention mechanism automatically captures the nonlinear and time-varying dependencies between features, breaking through the limitations of traditional models on fixed parameter assumptions, and enhancing the reliability of return prediction and the effectiveness of risk management.

[0098] In one embodiment, determining the financial institution's asset demand and the optimal asset allocation weight vector based on the first influencing factor, the second influencing factor, and the third influencing factor includes:

[0099] The first influencing factor, the second influencing factor, and the third influencing factor are input into the asset demand forecasting model. The objective function of the asset demand forecasting model is solved by an optimization algorithm to obtain the weight vector. The objective function aims to maximize the expected return of the asset portfolio and minimize the risk of the asset portfolio.

[0100] Based on the weight vector, the asset demand of the financial institution for each asset class is determined.

[0101] Optionally, in one specific implementation of this application, firstly, an objective function for the asset demand forecasting model is constructed. The first, second, and third influencing factors are integrated into a three-dimensional feature tensor, which is then input into a forecasting model composed of a three-layer neural network. The objective function adopts a weighted sum form: maximizing the expected return of the asset portfolio (calculating the expected rate of return of various assets under different economic scenarios based on the third influencing factor) while minimizing portfolio risk (quantified through the risk exposure covariance matrix in the second influencing factor). The weight coefficients are determined using the analytic hierarchy process (AHP) to reflect the risk preferences of financial institutions.

[0102] Subsequently, a multi-objective optimization problem is solved. A genetic algorithm is used to initialize a population containing 200 solutions, each representing an asset allocation scheme (i.e., a weight vector). During the iteration process, new solutions are generated through crossover and mutation operations, and the solutions are sorted according to Pareto dominance. Constraints include: 1) weight normalization constraint (the sum of the weights of all asset classes is 1); 2) regulatory constraints (e.g., the investment ratio of a single asset class does not exceed 30%); 3) risk constraints (VaR limits are set based on the second influencing factor). For example, during an economic expansion, the algorithm will increase the weight of risky assets based on the first influencing factor, while ensuring the maximization of expected returns through the third influencing factor, and controlling the risk exposure within the limit through the second influencing factor.

[0103] Next, the optimal weight vector is determined. The TOPSIS method is applied to the Pareto front solution set to calculate the distance of each solution to the ideal solution (maximum return and minimum risk) and the negative ideal solution, selecting the solution with the highest relative proximity as the optimal weight vector. For example, when market volatility is high, the algorithm may choose a more balanced allocation scheme, reducing the weight of high-risk assets.

[0104] Finally, the asset demand is calculated. The optimal weight vector is multiplied by the financial institution's available asset size to obtain the specific demand for each type of asset. For example, if the optimal weight vector shows a 40% equity allocation and the institution's available assets are 10 billion yuan, then the equity demand is 4 billion yuan. Stress testing verifies the stability of the allocation plan, simulating portfolio value changes under extreme market scenarios (such as a sudden rise in interest rates or a stock market crash), ensuring that losses do not exceed a preset threshold at a 95% confidence level. This process achieves full-chain quantification from market environment analysis to asset allocation decisions, providing financial institutions with scientific and dynamic asset management solutions.

[0105] In these alternative embodiments, the optimal weight vector is accurately calculated using optimization algorithms, making asset allocation more scientific and reasonable; with the goal of maximizing returns and minimizing risks, the risk and return of the asset portfolio are effectively balanced, and the risk-return ratio is improved.

[0106] In one embodiment, the asset demand forecasting model is trained through the following steps:

[0107] Obtain historical market economy data and historical institutional data;

[0108] The historical market economy data and the historical institutional data are preprocessed to obtain preprocessed data; the preprocessing includes data cleaning, standardization, and feature engineering.

[0109] Based on the preprocessed data, the training sample set and the validation sample set are determined;

[0110] Construct an initial asset demand forecasting model;

[0111] The initial asset demand prediction model is trained using the training sample set to obtain the trained asset demand prediction model.

[0112] Based on the validation sample set, the model parameters of the trained asset demand prediction model are determined through cross-validation.

[0113] If the model parameters do not meet the preset training conditions, the hyperparameters of the trained asset demand prediction model are adjusted based on the validation sample set to obtain the adjusted asset demand prediction model. The model training and validation steps are then returned until the model parameters of the adjusted asset demand prediction model reach the preset training stop condition, thus obtaining the trained asset demand prediction model.

[0114] Optionally, in one specific implementation of this application, data collection and preprocessing are performed first. The Z-score standardization formula is applied: Outliers are handled, where Z represents the Z-score; x represents the observed value; μ represents the mean; and σ represents the standard deviation. Observations with |Z|>3 are considered outliers and corrected using interpolation. Core features such as market volatility, risk exposure, and asset return are constructed through feature engineering, forming a multidimensional feature tensor.

[0115] Next, a dynamic asset demand forecasting model is constructed. A hybrid architecture is adopted: an LSTM time series model is used to process market volatility data to capture the long-term dependencies of variables such as interest rates and stock indices; a random forest model is applied to assess multi-dimensional risk exposures such as credit risk, market risk, and liquidity risk; and the self-attention mechanism of the Transformer model is utilized. Where Q, K, and V are the query, key, and value matrices, respectively, and d k Using macroeconomic conditions and investment strategies as key dimensions, this study analyzes the complex impact of these factors on asset returns.

[0116] Then, model training and optimization are performed. The preprocessed data is divided into training and validation sets in a 7:3 ratio, and cross-validation is performed using a rolling time window. During training, the mean squared error (MSE) loss function and a custom penalty term are combined to apply a higher weight to the prediction error in the asset-deficient scenario. The hyperparameters are optimized using a grid search algorithm: θ**argminθ∈eL(DDval,θ), where Θ is the candidate set of parameters, and D... val For the validation set, we focused on adjusting the hidden layer dimension of the LSTM, the tree depth of the random forest, and the number of attention heads in the Transformer. In each iteration, we generated 10 sets of candidate parameters and selected the combination with the highest Sharpe ratio on the validation set.

[0117] Next, model validation and parameter tuning are performed. The K-fold cross-validation formula is used: The evaluation assesses the model's generalization ability, where K is the fold number and Accuracy is the standard deviation. k This represents the accuracy at the Kth fold. If RMSE > 0.05 or Sharpe ratio < 1.2 (preset training conditions), the learning rate scheduling strategy is adjusted or the regularization strength is increased. Ensemble learning is used to fuse multiple optimal models, thereby improving prediction stability.

[0118] Finally, complete model training and deployment. When the model's performance on the validation set reaches a preset threshold (e.g., RMSE < 0.03, Sharpe ratio > 1.5), stop training and save the optimal parameters. Deploy the trained model to the production environment, obtain market dynamics through a real-time data interface, and automatically trigger a model update every two weeks to ensure that predictive capabilities are continuously optimized as the market environment changes. The entire training process is containerized using Docker and utilizes GPU-accelerated computation, achieving full automation from data processing to model deployment.

[0119] In these alternative embodiments, multi-source data preprocessing integrates macro and institutional data, and improves data quality through cleaning, standardization, and feature engineering; cross-validation mechanism combines training and validation sets to ensure the generalization ability of model parameters under different data distributions; hyperparameter dynamic tuning is iteratively optimized based on validation results to avoid overfitting and improve model adaptability; closed-loop training mechanism continuously iterates until convergence, enabling the model to capture the complex relationship between market environment and institutional characteristics, providing reliable support for asset allocation decisions.

[0120] It should be noted that the various optional implementation methods described in the embodiments of this application can be combined with each other or implemented individually without conflict, and the embodiments of this application do not limit this.

[0121] To facilitate understanding of the model-based data processing method provided in the above embodiments, the following describes the model-based data processing method using a specific scenario embodiment.

[0122] Optionally, this application constructs a comprehensive data foundation by integrating macroeconomic data (such as GDP growth rate, interest rate, and inflation rate), market volatility data (such as stock market index and bond yield), internal insurance company data (such as premium income, loss ratio, and investment returns), and customer behavior data (such as insurance preferences and surrender rates). Simultaneously, through data cleaning and feature engineering, key features (such as market volatility, risk exposure, and asset return) are extracted, providing high-quality data support for the training and optimization of the dynamic management model.

[0123] This application employs various machine learning models (such as time series models, random forest models, and deep learning models) to dynamically predict asset demand. Specifically, it includes:

[0124] 1. Market volatility forecasting: Using LSTM time series models to analyze market volatility trends and assess their impact on asset demand.

[0125] 2. Risk Exposure Assessment: Random forest model is used to analyze the risk exposure of insurance companies (such as credit risk, market risk, and liquidity risk) and predict the impact of potential risks on asset demand.

[0126] 3. Asset return prediction: Utilize deep learning models (such as Transformer) to analyze the impact of the macroeconomic environment and investment strategies on asset returns, and optimize asset allocation strategies.

[0127] 4. Dynamic Adjustment and Optimization: This application dynamically adjusts asset adequacy targets and asset allocation strategies by monitoring market changes and risk exposure in real time. Combined with sensitivity analysis, it assesses the impact of changes in asset demand on asset adequacy, optimizes asset management strategies, and ensures the insurance company's asset adequacy and risk controllability.

[0128] Optionally, in this embodiment, the asset demand of insurance companies is calculated based on the forecast results of market volatility and risk exposure. Asset allocation strategies are optimized based on the asset return forecast results to ensure efficient asset utilization. Finally, the latest data on market volatility and risk exposure can be obtained through a real-time data interface to dynamically adjust the asset demand forecasting model. Furthermore, asset adequacy targets and asset allocation strategies are dynamically adjusted based on the forecast results of market changes and risk exposure.

[0129] In these optional embodiments, traditional asset adequacy management methods are optimized through dynamic management models, achieving accurate assessment and dynamic adjustment of asset adequacy. Multi-source data integration enhances the comprehensiveness and accuracy of asset adequacy assessment. Machine learning models improve the accuracy and reliability of asset demand forecasting. Real-time monitoring of market changes allows for dynamic adjustment of asset adequacy targets and asset allocation strategies, improving the flexibility and adaptability of asset management. Automated data processing and model training enhance the efficiency and responsiveness of asset management. This application helps insurance companies better cope with market fluctuations and risk events, ensuring asset adequacy while improving asset utilization efficiency, providing strong support for the company's sound operation and sustainable development.

[0130] Please see Figure 5 This application also provides a model-based data processing apparatus that can implement the above-described model-based data processing method. The apparatus includes:

[0131] The acquisition module is used to acquire market economic data and institutional data from financial institutions;

[0132] The first determining module is used to determine a first influencing factor based on the market economic data; the first influencing factor is used to characterize the degree of influence of the market economic environment on the asset demand of the financial institution.

[0133] The second determining module is used to determine a second influencing factor based on the first influencing factor and the institutional data; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the financial institution's asset demand.

[0134] The third determining module is used to determine a third influencing factor based on the second influencing factor and the market economic data; the third influencing factor is used to characterize the degree of influence of the market economic environment on the asset return rate of the financial institution.

[0135] The fourth determining module is used to determine the asset demand of the financial institution and the weight vector of the optimal asset allocation based on the first influencing factor, the second influencing factor and the third influencing factor; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

[0136] The specific implementation of this model-based data processing device is basically the same as the specific implementation of the model-based data processing method described above, and will not be repeated here.

[0137] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described model-based data processing method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0138] Please see Figure 6 , Figure 6 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0139] The processor 601 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0140] The memory 602 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 602 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 602 and is called and executed by the processor 601 using the model-based data processing method of the embodiments of this application.

[0141] The input / output interface 603 is used to implement information input and output;

[0142] The communication interface 604 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0143] Bus 605 transmits information between various components of the device (e.g., processor 601, memory 602, input / output interface 603, and communication interface 604);

[0144] The processor 601, memory 602, input / output interface 603, and communication interface 604 are connected to each other within the device via bus 605.

[0145] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described model-based data processing method.

[0146] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0147] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0148] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0149] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0150] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0151] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0152] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0153] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0154] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0155] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0156] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0157] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A model-based data processing method, characterized in that, The method includes: Obtain market economy data and institutional data from financial institutions; Based on the market economic data, a first influencing factor is determined; the first influencing factor is used to characterize the degree of impact of the market economic environment on the asset demand of the financial institutions. Based on the first influencing factor and the institutional data, a second influencing factor is determined; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the financial institution's asset demand. Based on the second influencing factor and the market economic data, a third influencing factor is determined; the third influencing factor is used to characterize the degree of impact of the market economic environment on the asset return rate of the financial institution. Based on the first influencing factor, the second influencing factor, and the third influencing factor, the asset demand of the financial institution and the weight vector of the optimal asset allocation are determined; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

2. The method according to claim 1, characterized in that, The determination of the first influencing factor based on the market economy data includes: Extract time-series volatility indicators from the market economic data; the time-series volatility indicators include at least one of stock index volatility, yield curve change rate, and macroeconomic surprise index. The time-series fluctuation index is input into the time series prediction model, and the nonlinear mapping relationship between the time-series fluctuation index is captured by the time series prediction model to output the first influencing factor. The time series prediction model includes a long short-term memory network.

3. The method according to claim 1, characterized in that, The step of determining the second impact factor based on the first impact factor and the institutional data includes: Multidimensional risk features are extracted from the institutional data; the multidimensional risk features include at least one of credit risk exposure, market risk sensitivity, and liquidity coverage ratio. The first influencing factor is used as the risk feature weight adjustment coefficient to weight the multidimensional risk features, thereby obtaining the weighted features. The weighted features are input into a random forest model, and the marginal effect of changes in risk exposure is evaluated through the decision tree integrated by the random forest model to obtain the second influencing factor.

4. The method according to claim 1, characterized in that, The determination of the third influencing factor based on the second influencing factor and the market economic data includes: Extract return characteristics from the market economic data; the return characteristics include at least one of the following: economic growth rate, inflation rate, and slope of the asset yield curve. Using the second influencing factor as a risk adjustment coefficient, the return characteristics are corrected to obtain the corrected characteristics; The modified features are input into a reinforcement learning model based on a self-attention mechanism. The dynamic dependencies between features are captured through the multi-head attention layer of the reinforcement learning model to obtain the third influencing factor.

5. The method according to claim 1, characterized in that, The step of determining the financial institution's asset demand and the optimal asset allocation weight vector based on the first influencing factor, the second influencing factor, and the third influencing factor includes: The first influencing factor, the second influencing factor, and the third influencing factor are input into the asset demand forecasting model. The objective function of the asset demand forecasting model is solved by an optimization algorithm to obtain the weight vector. The objective function aims to maximize the expected return of the asset portfolio and minimize the risk of the asset portfolio. Based on the weight vector, the asset demand of the financial institution for each asset class is determined.

6. The method according to claim 5, characterized in that, The asset demand forecasting model is trained through the following steps: Obtain historical market economy data and historical institutional data; The historical market economy data and the historical institutional data are preprocessed to obtain preprocessed data; the preprocessing includes data cleaning, standardization, and feature engineering. Based on the preprocessed data, the training sample set and the validation sample set are determined; Construct an initial asset demand forecasting model; The initial asset demand prediction model is trained using the training sample set to obtain the trained asset demand prediction model. Based on the validation sample set, the model parameters of the trained asset demand prediction model are determined through cross-validation. If the model parameters do not meet the preset training conditions, the hyperparameters of the trained asset demand prediction model are adjusted based on the validation sample set to obtain the adjusted asset demand prediction model. The model training and validation steps are then returned until the model parameters of the adjusted asset demand prediction model reach the preset training stop condition, thus obtaining the trained asset demand prediction model.

7. A model-based data processing device, characterized in that, The device includes: The acquisition module is used to acquire market economic data and institutional data from financial institutions; The first determining module is used to determine a first influencing factor based on the market economic data; the first influencing factor is used to characterize the degree of influence of the market economic environment on the asset demand of the financial institution. The second determining module is used to determine a second influencing factor based on the first influencing factor and the institutional data; the second influencing factor is used to characterize the marginal impact of changes in risk exposure on the financial institution's asset demand. The third determining module is used to determine a third influencing factor based on the second influencing factor and the market economic data; the third influencing factor is used to characterize the degree of influence of the market economic environment on the asset return rate of the financial institution. The fourth determining module is used to determine the asset demand of the financial institution and the weight vector of the optimal asset allocation based on the first influencing factor, the second influencing factor and the third influencing factor; the weight vector is used to represent the optimal proportion of different asset classes in the financial institution's investment portfolio.

8. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the model-based data processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the model-based data processing method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, When the instructions in the computer program product are executed by an electronic device, the electronic device performs the model-based data processing method as described in any one of claims 1-6.

Images