Asset allocation strategy recommendation method and device, equipment, storage medium and product

Abstract

The embodiment of the invention provides an asset allocation strategy recommendation method and device, equipment, a storage medium and a product, and relates to the field of financial science and technology. The method comprises the following steps: acquiring macroeconomic data in a current configuration period, and determining a corresponding macroscopic factor feature based on the macroeconomic data; acquiring historical investment behavior data of a user in a current configuration period, and determining corresponding investment features based on the historical investment behavior data; based on the macroscopic factor features and the investment features, generating an asset configuration strategy recommendation result suitable for the next configuration period; wherein the asset configuration strategy recommendation result comprises an asset category, a configuration proportion and an adjustment suggestion. According to the method provided by the invention, the technical problems of poor real-time performance, lack of personalized adaptation capability and insufficient dynamic risk control of an existing asset allocation strategy recommendation result are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of financial technology, in particular to an asset allocation strategy recommendation method and device, equipment, storage medium and product. BACKGROUND

[0002] In modern financial markets, asset allocation strategy recommendation is a core link of investment management, which helps investors achieve the balance between risk and return in different market environments, and is a key means to improve portfolio performance. Reasonable asset allocation not only helps to diversify risks, but also captures investment opportunities in diversified markets and meets diverse financial goals.

[0003] Currently, common asset allocation methods include Markowitz mean-variance model, Black-Litterman model and risk parity strategy, etc. These methods are mostly based on historical statistical data and preset risk preferences, and recommend static or semi-static asset allocation strategies for investors through mathematical modeling.

[0004] However, with the increasing complexity and uncertainty of the market, traditional methods are difficult to adapt to the dynamic changes of the macro environment, so that the recommended allocation strategy often lags behind the actual market performance, resulting in a deviation between the investment strategy and the expected target, and limitations in personalized customization and risk control ability, which cannot meet the real-time and adaptive needs of investors. SUMMARY

[0005] The present application provides an asset allocation strategy recommendation method, device, equipment, storage medium and product, to solve the technical problems of poor real-time performance, lack of personalized adaptation ability and insufficient dynamic risk control of existing asset allocation strategy recommendation results, and to dynamically generate asset allocation strategies based on macro factors and customer behavior, to improve the real-time response ability, personalized recommendation accuracy and multi-period risk control level of the strategy.

[0006] In a first aspect, the present application provides an asset allocation strategy recommendation method, comprising:

[0007] Obtaining macroeconomic data in the current allocation period, and determining corresponding macro factor features based on the macroeconomic data, wherein the macro factor features are used to reflect the market environment in the next allocation period;

[0008] Obtaining historical investment behavior data of a user in the current allocation period, and determining corresponding investment features based on the historical investment behavior data, wherein the investment features are used to reflect the user's preferences and risk tolerance;

[0009] Based on the macro factor features and the investment features, an asset allocation strategy recommendation result suitable for the next allocation period is generated.

[0010] Optionally, the corresponding macro factor features are obtained based on the macroeconomic data, including:

[0011] The macroeconomic data is preprocessed and factor constructed to generate a historical macro factor feature sequence in a current configuration period;

[0012] The historical macro factor feature sequence is input into a pre-trained macro factor prediction model to obtain an initial macro factor feature set corresponding to the next configuration period;

[0013] The initial macro factor feature set is subjected to significance analysis and screening to obtain the macro factor features corresponding to the next configuration period.

[0014] Optionally, historical investment behavior data of a user in a current period is obtained, including:

[0015] The historical investment behavior data of a plurality of users in a financial system is obtained, and the historical investment behavior data of the plurality of users is cleaned and processed to form a historical investment behavior dataset;

[0016] The historical investment behavior dataset is subjected to investment behavior index extraction according to a preset period to obtain corresponding multi-dimensional investment indexes; the multi-dimensional investment indexes are used to reflect investment preferences and risk attributes;

[0017] According to the multi-dimensional investment indexes, the plurality of users are subjected to investment behavior feature clustering to identify a user group having similar investment behavior features;

[0018] The aggregated investment behavior data of the user group in which the user is located and the historical investment behavior data of the user are fused to generate the historical investment behavior data of the user.

[0019] Optionally, the historical investment behavior data includes at least one of the fund transaction data, investment questionnaire data, and investment preference data;

[0020] Based on the historical investment behavior data, corresponding investment features are determined, including:

[0021] Based on the fund transaction data of the user group, dynamic investment behavior indexes of the user in the next configuration period are determined;

[0022] Based on the investment questionnaire data and the investment preference data, static risk preference indexes of the user in the next configuration period are determined;

[0023] The dynamic investment behavior indexes and the static risk preference indexes are input into a preset feature fusion model to obtain corresponding investment features.

[0024] Optionally, the investment characteristics include the yield preference characteristics, the risk tolerance characteristics, and the liquidity demand characteristics; the characteristic fusion model includes a yield preference characteristic extraction sub-model, a risk tolerance characteristic extraction sub-model, and a liquidity demand characteristic extraction sub-model;

[0025] The dynamic investment behavior indicators and the static risk preference indicators are input into a preset characteristic fusion model to obtain corresponding investment characteristics, including:

[0026] The dynamic investment behavior indicators and the static risk preference indicators are input into the yield preference characteristic extraction sub-model to obtain the yield preference characteristics corresponding to the user;

[0027] The dynamic investment behavior indicators and the static risk preference indicators are input into the risk tolerance characteristic extraction sub-model to obtain the risk tolerance characteristics corresponding to the user;

[0028] The dynamic investment behavior indicators and the static risk preference indicators are input into the liquidity demand characteristic extraction sub-model to obtain the liquidity demand characteristics corresponding to the user;

[0029] The yield preference characteristics, the risk tolerance characteristics, and the liquidity demand characteristics are vectorized and spliced to generate the investment characteristics.

[0030] Optionally, based on the macro factor characteristics and the investment characteristics, an asset allocation strategy recommendation result suitable for the next allocation period is generated, including:

[0031] A pre-trained asset allocation strategy recommendation model is obtained;

[0032] The macro factor characteristics and the investment characteristics are input into the asset allocation strategy recommendation model to output an asset allocation strategy recommendation result suitable for the next allocation period; wherein the asset allocation strategy recommendation result includes asset categories, allocation proportions, and adjustment suggestions.

[0033] Optionally, the asset allocation strategy recommendation model is a strategy generation model based on reinforcement learning, including a state space, an action space, and a reward and punishment function;

[0034] The macro factor characteristics and the investment characteristic vector are jointly input into the asset allocation strategy recommendation model, and multi-dimensional fusion and strategy reasoning are performed through the model to output an individualized asset allocation strategy recommendation result, including:

[0035] The macro factor characteristics and the investment characteristic vector are spliced to form a comprehensive state representation and are input into the state space;

[0036] define the generated asset allocation action as the action space, the action including an asset class allocation proportion adjustment operation;

[0037] construct a reward function including a basic reward term and a penalty term: wherein,

[0038] the basic reward term is calculated based on a risk-adjusted return indicator; the indicator includes a Sharpe ratio and a Kama ratio;

[0039] the penalty term includes a transaction cost penalty, a turnover rate penalty, and a risk deviation penalty; wherein,

[0040] the transaction cost penalty is calculated based on the rate and the transaction amount generated by asset transactions;

[0041] the turnover rate penalty is calculated based on the ratio of the total amount of buy and sell transactions of the investment portfolio to the average asset size within a preset period;

[0042] the risk deviation penalty is calculated based on the difference between the actual investment portfolio volatility, the risk value, and the user's target risk level;

[0043] Through the asset allocation strategy recommendation model, the environment state in the state space is perceived, the configuration strategy in the action space is explored, and the optimization strategy selection is fed back according to the reward function, to obtain the asset allocation strategy recommendation result of the next configuration period.

[0044] In a second aspect, the present application provides an asset allocation strategy recommendation device, comprising:

[0045] a macro factor feature acquisition module for acquiring macroeconomic data in the current configuration period, and determining corresponding macro factor features based on the macroeconomic data; wherein the macro factor features are used to reflect the market environment in the next configuration period;

[0046] an investment feature acquisition module for acquiring historical investment behavior data of a user in the current configuration period, and determining corresponding investment features based on the historical investment behavior data; wherein the investment features are used to reflect the user's preferences and risk tolerance;

[0047] a recommendation result generation module for generating an asset allocation strategy recommendation result suitable for the next configuration period based on the macro factor features and the investment features.

[0048] In a third aspect, the present application provides an electronic device, comprising a processor and a memory in communication with the processor;

[0049] the memory stores computer execution instructions;

[0050] The processor executes computer-executed instructions stored in the memory to implement the method of the first aspect.

[0051] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executed instructions, and the computer-executed instructions are executed by a processor to implement the method of the first aspect.

[0052] In a fifth aspect, the present application provides a computer program product, comprising a computer program, and the computer program is executed by a processor to implement the method of the first aspect.

[0053] The asset allocation strategy recommendation method, device, equipment, storage medium and program product provided by the present application effectively improve the adaptability of the strategy to the dynamic changes of the market environment by introducing a macro factor prediction mechanism, overcoming the lagging defect of the traditional method relying on historical data. At the same time, by fusing user individual and group behavior data to construct personalized investment characteristics, the accurate portrayal of user risk preference is realized, and the customization ability and risk control precision of the strategy are enhanced. On this basis, relying on automatic modeling and multi-objective optimization, the asset allocation strategy with market adaptability and user matching degree is generated while reducing human bias and transaction cost, improving investment efficiency and user satisfaction. BRIEF DESCRIPTION OF DRAWINGS

[0054] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0055] Figure 1 An application scenario diagram of the asset allocation strategy recommendation method provided by the present application;

[0056] Figure 2 Another application scenario diagram of the asset allocation strategy recommendation method provided by the present application;

[0057] Figure 3 A flowchart of the asset allocation strategy recommendation method provided by the embodiment of the present application;

[0058] Figure 4 A structural diagram of the asset allocation strategy recommendation device provided by the embodiment of the present application;

[0059] Figure 5 A block diagram of an electronic device provided by the embodiment of the present application.

[0060] The specific embodiments of the application have been shown by the above drawings, and will be described in more detail hereinafter. These drawings and the written description are not intended to restrict the scope of the inventive concept in any way, but to illustrate the inventive concept by reference to specific embodiments. DETAILED DESCRIPTION

[0061] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The description of the exemplary embodiments is intended to apply to any embodiment of the application, unless specified otherwise. It should be understood that every embodiment need not necessarily include all of the features shown in the drawings. The exemplary embodiments described herein are not intended to be exhaustive or to be limited to the precise form disclosed. They are chosen and described in order to best explain the principles of the application and its best mode of operation.

[0062] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards of relevant countries and regions, necessary security measures are taken, public order and good customs are not violated, and appropriate operation portals are provided for users to choose authorization or refusal.

[0063] And the present application involves big data analysis of user information (including but not limited to personal biological characteristics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and uses artificial intelligence technology for automatic decision-making, and makes technical solutions based on automatic decision-making results that have a significant impact on personal rights and interests, provides appropriate operation portals for users to choose to agree or refuse automatic decision-making results; if the user chooses to refuse, the expert decision-making process is entered.

[0064] It should be noted that the asset allocation strategy recommendation method, device, equipment, storage medium and product provided by the present application can be used in the field of financial technology, and can also be used in any field other than financial technology. The application field of the asset allocation strategy recommendation method, device, equipment, storage medium and product in the present application is not limited.

[0065] Technical terms:

[0066] Macro factor characteristics: refer to macroeconomic variables that have a systematic impact on financial markets or asset prices, including but not limited to economic growth rate, inflation rate, interest rate level, exchange rate fluctuation, monetary policy adjustment, fiscal policy orientation, international trade data and geopolitical events, etc. These factors are converted into calculable numerical indicators through quantitative models for dynamic asset allocation strategy recommendation decisions.

[0067] Investment characteristics: A multi-dimensional data representation that integrates customer historical transaction records, fund flow patterns, risk preference questionnaire results, holding period, and rebalancing frequency to generate a feature set reflecting the customer's dynamic investment behavior. This vector is used to quantify the customer's risk tolerance, investment style, and behavioral preferences.

[0068] Asset allocation strategy recommendation method: An intelligent asset allocation strategy recommendation method based on macro factor and user behavior dual perspectives. The core innovation lies in the integration of macro factor characteristics reflecting future market environment and investment characteristics reflecting user's personalized preferences and risk tolerance, and generating customized asset allocation strategies for the next period.

[0069] Reinforcement learning model: A machine learning paradigm that simulates the interaction between agents and the environment to maximize long-term rewards and dynamically optimize asset allocation strategy recommendations. In this invention, the environment state is defined by macro factors and customer behavior data, and the action space is the adjustment of asset weights.

[0070] In modern financial markets, asset allocation strategy recommendation is the core of investment management, helping investors balance risk and return in different market environments, and is a key means to improve portfolio performance. Reasonable asset allocation not only helps to diversify risk, but also captures investment opportunities in multiple markets and meets diverse financial goals.

[0071] Currently, common asset allocation methods include Markowitz mean-variance model, Black-Litterman model, and risk parity strategy. Specifically, asset allocation strategy recommendation can use the Markowitz mean-variance model for allocation, i.e., calculating the optimal asset weight based on historical returns and volatility. However, the above allocation process relies on low-frequency macroeconomic data (such as quarterly GDP) or historical price data, resulting in allocation strategies lagging behind real-time market changes and failing to respond quickly to unexpected events.

[0072] Alternatively, another asset allocation strategy recommendation method can also categorize customers into conservative, balanced, or aggressive types based on their risk questionnaire results and assign a pre-set fixed proportion of asset portfolios. However, static questionnaires cannot reflect the customer's real behavior (such as high-risk questionnaire users who actually trade low-volatility assets frequently), leading to allocation results deviating from the customer's actual needs.

[0073] Optionally, another asset allocation strategy recommendation method can also adjust the bond allocation ratio according to interest rate changes, or adjust the stock position according to GDP growth. However, the above allocation methods often analyze macro factors separately (such as only focusing on interest rates), ignore the nonlinear interaction between factors (such as the superimposed effect of inflation and exchange rate depreciation), that is, do not systematically integrate the synergistic effect of multiple factors, and ignore the feedback effect of customer behavior on the strategy.

[0074] Optionally, another asset allocation strategy recommendation method can also predict asset returns using linear regression or time series analysis. However, in the allocation process, the model assumptions are strong (such as normal distribution, linear correlation of factors), which is difficult to adapt to nonlinear market fluctuations, thereby triggering irrational portfolio adjustment and exacerbating portfolio drawdown.

[0075] In summary, these methods are mostly based on historical statistical data and preset risk preferences, and recommend static or semi-static asset allocation strategies for investors through mathematical modeling.

[0076] However, with the increasing complexity and uncertainty of the market, traditional methods are difficult to adapt to the dynamic changes of the macro environment, making the recommended allocation strategy often lag behind the actual market performance, leading to a deviation between the investment strategy and the expected target, while there are limitations in individual customization and risk control capabilities, which cannot meet the real-time and adaptive needs of investors for investment strategies.

[0077] The asset allocation strategy recommendation method provided by the present application aims to solve the above technical problems of the prior art. Specifically, by introducing a macro factor prediction mechanism, the adaptability of the strategy to dynamic changes in the market environment is effectively improved, overcoming the lagging defect of traditional methods that rely on historical data; at the same time, by fusing user individual and group behavior data to construct personalized investment features, the user's risk preference is accurately described, and the customization ability and risk control precision of the strategy are enhanced; on this basis, relying on automated modeling and multi-objective optimization, the asset allocation strategy that combines market adaptability and user matching degree is generated while reducing human bias and transaction costs, improving investment efficiency and user satisfaction.

[0078] The asset allocation strategy recommendation method provided by the present application is suitable for various real scenarios that require dynamic, personalized and intelligent asset allocation, especially suitable for wealth management, investment consulting and personal finance business needs.

[0079] For example, in financial institutions such as securities firms, funds, or banks, investment advisors can quickly generate dynamic asset allocation plans that closely match clients' risk preferences based on their historical behavioral data and the current macroeconomic environment, thereby improving service efficiency and client satisfaction. In high-net-worth client wealth management scenarios, the proportion of major asset classes can be adjusted in real time according to market changes, and portfolio risk and transaction costs can be automatically controlled to achieve stable return targets. In internet financial platforms or robo-advisor applications, ordinary users can also use this technology to obtain personalized and adaptive allocation services that were originally reserved for professional investors, thereby lowering the investment threshold.

[0080] Furthermore, this method can be embedded into investment decision support systems, portfolio management systems, or financial cloud service platforms. For example, in family offices or pension fund management institutions, the system can generate cross-cycle asset allocation strategies based on long-term macroeconomic forecasts and client lifecycle characteristics; in quantitative investment platforms, it can be combined with automated trading systems to achieve integrated management of strategy recommendation and execution.

[0081] In summary, any financial business scenario that requires integrating macroeconomic judgment with personalized needs, providing dynamic asset allocation advice, and emphasizing risk and cost control falls within the scope of application of this technical solution.

[0082] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0083] For easier understanding, the following will be combined with Figure 1 and Figure 2 The application scenarios used in the embodiments of this application are described below. See also... Figure 1 and Figure 2 The asset allocation strategy recommendation method provided in this application involves an information acquisition device 11 and an asset allocation strategy recommendation device 12.

[0084] Specifically, see Figure 1 The system acquires macroeconomic data for the current allocation period and historical investment behavior data of the user within the current allocation period through the information acquisition device 11, and sends the two types of data to the asset allocation strategy recommendation device 12 for asset allocation strategy recommendation.

[0085] In the asset allocation strategy recommendation device 12, the received macroeconomic data of the current allocation period is processed to obtain macroeconomic factor characteristics that reflect the market environment of the next allocation period. At the same time, the received historical investment behavior data is also processed to determine investment characteristics that reflect the user's preferences and risk tolerance. Subsequently, the asset allocation strategy is generated and recommended by combining the macroeconomic factor characteristics and investment characteristics, so as to achieve more accurate and personalized asset allocation strategy recommendations for users.

[0086] Optionally, in some scenarios, see Figure 2 The information acquisition device 11 acquires macroeconomic data for the current allocation period and historical investment behavior data of users within the current allocation period. It processes the received macroeconomic data for the current allocation period to obtain macroeconomic factor characteristics that reflect the market environment for the next allocation period. It also processes the received historical investment behavior data to determine investment characteristics that reflect users' preferences and risk tolerance. Furthermore, it sends the generated macroeconomic factor characteristics and investment characteristics to the asset allocation strategy recommendation device 12 for asset allocation strategy generation and recommendation.

[0087] In the asset allocation strategy recommendation device 12, asset allocation strategies are generated and recommended by combining macroeconomic factors and investment characteristics, so as to provide users with more accurate and personalized asset allocation strategy recommendations.

[0088] In other scenarios, the information acquisition device 11 and the asset allocation strategy recommendation device 12 may be the same device. This embodiment does not make specific limitations on this.

[0089] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0090] Figure 3 This is a flowchart illustrating an asset allocation strategy recommendation method provided in an embodiment of this application. The method can be executed by an asset allocation strategy recommendation device, which can be a server or an electronic device. The following description uses an electronic device as an example. The method in this embodiment can be implemented through software, hardware, or a combination of both. Figure 3 As shown, the method includes the following steps:

[0091] S301. Obtain macroeconomic data for the current configuration period and determine the corresponding macroeconomic factor characteristics based on the macroeconomic data.

[0092] In this embodiment, macroeconomic data can be divided into data corresponding to different allocation periods to facilitate more accurate asset allocation strategy generation and recommendation decisions. The allocation period can be flexibly set according to the needs of the actual investment strategy, such as being divided by quarter, semi-annually, or annually.

[0093] By dividing macroeconomic data into data for different allocation periods, the system can dynamically adjust asset allocation strategy recommendations based on data from different periods, ensuring that the investment portfolio can respond promptly to market changes and economic trends.

[0094] Building upon the above, for each configuration period, multiple pre-configured data interfaces can be used to integrate multi-source macroeconomic data within the current configuration period. This multi-source macroeconomic data can include macroeconomic indicators, market dynamics, policy changes, and market sentiment.

[0095] Specifically, multi-source data interfaces include, but are not limited to, macroeconomic databases to provide global economic indicators and trend analysis. In addition, data sources include financial market regulatory documents to obtain the latest policy changes and compliance requirements. To enhance understanding of market sentiment, a news and public opinion API is also integrated to enable real-time analysis of media reports and public sentiment. Furthermore, alternative data sources, such as satellite imagery and supply chain logs, can be introduced to capture economic activities and market dynamics that traditional data cannot cover.

[0096] In this way, the integration of multi-source data enables the system to comprehensively and accurately analyze the current economic environment, providing a solid data foundation for asset allocation strategy recommendations.

[0097] Because macroeconomic data exhibits specific patterns of change across different investment cycles, it is possible to predict the economic situation of the next cycle by acquiring current-cycle macroeconomic data and extracting key features to generate macroeconomic factor characteristics that characterize the future market environment. These factors can include indicators that significantly impact asset returns and risks, such as economic growth rates, inflation rates, interest rate levels, unemployment rates, and trade data.

[0098] By using macroeconomic factors as data for asset allocation strategy recommendations, we can not only capture the evolving trends of economic dynamics, but also provide forward-looking and precise support for the formulation of asset allocation strategies. In addition, by introducing the characteristics of macroeconomic factors, we can gain a deeper understanding of the economic environment, identify potential opportunities and risks in strategy recommendations, thereby improving the adaptability and overall effectiveness of asset allocation.

[0099] S302. Obtain the user's historical investment behavior data within the current configuration period, and determine the corresponding investment characteristics based on the historical investment behavior data.

[0100] While acquiring macroeconomic data and macroeconomic factor characteristics, it is also possible to simultaneously collect historical investment behavior data of users within the same allocation period. By extracting features from this behavioral data, corresponding investment characteristics are generated, serving as another important basis for asset allocation strategy recommendations. In this way, allocation needs can be comprehensively grasped from both market and user perspectives, improving the accuracy and effectiveness of strategy recommendations.

[0101] In this embodiment, historical investment behavior data may include, but is not limited to, user transaction data, investment questionnaire data, and investment preference data. To adapt to its dynamic nature over time, investment behavior data can be divided and managed according to configuration periods to ensure the timeliness and relevance of feature extraction.

[0102] Specifically, for each configuration cycle, multi-dimensional user behavior data is collected from multiple platforms via API calls or data synchronization, including Customer Relationship Management (CRM) systems, transaction databases, fund clearing platforms, and other investment and wealth management applications. For example: user profile data, including risk assessment results, investment preference tags, and demographic information, is obtained from the CRM system; detailed transaction records for the current cycle, such as transaction targets, time, quantity, and price, are extracted from the transaction database; data on fund inflows and outflows, portfolio valuations, and profit settlements are obtained from the fund clearing platform; furthermore, preference signals can be collected from user behavior logs of various wealth management applications and investment platforms, such as users' reading records of financial information, product page browsing time, and content they are interested in.

[0103] Furthermore, the acquired raw data is cleaned, integrated, and periodically aligned to filter out valid data belonging to the current configuration period, forming a complete set of historical investment behavior data for the user within this configuration period.

[0104] Based on this dataset, user investment characteristics can be further extracted using feature extraction techniques. Here, investment characteristics are a comprehensive quantitative representation of a user's investment behavior, risk tolerance, and preferences, and may include multiple dimensions such as risk preference score, trading activity, liquidity preference, and areas of interest.

[0105] By introducing these dynamic and personalized investment features, asset allocation strategy recommendations can be better aligned with users' actual needs, providing more accurate and adaptable strategy recommendations in a complex and ever-changing market environment, thereby improving the personalization level and decision-making quality of investment recommendation services.

[0106] S303. Based on macroeconomic factors and investment characteristics, generate asset allocation strategy recommendations suitable for the next allocation cycle.

[0107] In this embodiment, based on the macroeconomic factor characteristics and user investment characteristics extracted in the preceding steps, a preset asset allocation strategy recommendation method is used for calculation and analysis to generate specific asset allocation strategy recommendations suitable for the next allocation cycle. These recommendations include, but are not limited to, target asset classes, detailed allocation ratios, and corresponding weight adjustment suggestions.

[0108] Optionally, the recommended method can be implemented using various technical approaches, including but not limited to the following categories: First, machine learning models, such as decision trees and support vector machines, which are suitable for mining potential patterns from structured features; second, neural network models, which are particularly capable of effectively capturing complex interactions and nonlinear relationships between high-dimensional features by leveraging multi-layer nonlinear transformation capabilities; and third, optimization algorithms based on traditional mathematical finance, such as the Markowitz mean-variance model, which can achieve the optimal solution of portfolio theory under a given risk preference.

[0109] This implementation method does not limit the specific model and algorithm. In practical applications, a single method or a combination of multiple methods can be flexibly selected based on factors such as data characteristics, risk control requirements, and real-time requirements, in order to achieve a more robust asset allocation recommendation effect that is more adaptable to the market environment and meets user needs.

[0110] The asset allocation strategy recommendation method provided in this embodiment effectively improves the adaptability of the strategy to dynamic changes in the market environment by introducing a macroeconomic factor prediction mechanism, overcoming the lag defect of traditional methods that rely on historical data. At the same time, by integrating individual and group user behavior data to construct personalized investment characteristics, it achieves accurate characterization of user risk preferences, enhancing the customization capability and risk control precision of the strategy. On this basis, relying on automated modeling and multi-objective optimization, it generates asset allocation strategies that combine market adaptability and user matching while reducing human bias and transaction costs, thereby improving investment efficiency and user satisfaction.

[0111] The following section provides a detailed description of the entire process of the asset allocation strategy recommendation method provided in this embodiment. It should be understood that the following description is not intended to limit the technical solution of this embodiment.

[0112] Optionally, when macroeconomic data for the current configuration period is available, the process of obtaining the corresponding macroeconomic factor features based on the macroeconomic data may include: preprocessing and constructing factors from the macroeconomic data to generate a historical macroeconomic factor feature sequence for the current configuration period; inputting the historical macroeconomic factor feature sequence into a pre-trained macroeconomic factor prediction model to obtain the initial macroeconomic factor feature set for the next configuration period; and performing significance analysis and screening on the initial macroeconomic factor feature set to obtain the macroeconomic factor features for the next configuration period.

[0113] Specifically, the acquired raw macroeconomic data undergoes preprocessing, including data cleaning, missing value imputation, outlier handling, and standardization, to eliminate noise and unify dimensions. Subsequently, a preliminary factor pool is constructed based on economic theory and statistical methods to generate a historical macroeconomic factor feature sequence with temporal consistency within the current allocation period.

[0114] Based on this, the historical sequence is input into a pre-trained macroeconomic factor prediction model. For example, the macroeconomic factor prediction model includes, but is not limited to, time series models, machine learning regression models, or deep learning networks, to predict the initial set of macroeconomic factor features for the next configuration period.

[0115] Statistical testing methods, such as t-tests, ANOVA, or feature selection based on L1 regularization, are used to perform significance analysis on the initial set and screen out macroeconomic factors that have a significant impact on market expected returns or risks, ultimately forming macroeconomic factor features for generating asset allocation strategies.

[0116] In the above implementation process, by introducing macroeconomic factor prediction and significance screening, the adaptability of asset allocation strategies to future market trends has been improved. At the same time, based on multi-dimensional dynamic data fusion and automated modeling, subjective bias and manual costs can be reduced while enhancing the accuracy of personalized recommendations and the level of risk control, thereby achieving higher risk-adjusted returns and user satisfaction.

[0117] During this period, the acquired historical investment behavior data will also be processed to obtain the user's investment characteristics.

[0118] In some scenarios, if a single user has limited historical data, the investment features extracted from that data may be insufficient to support accurate asset allocation strategy recommendations. To address this, user profile matching and investment behavior similarity calculations can be used to identify user groups with high similarity to the current user. The historical investment behavior data of this user group can then be added to the individual user's dataset as an extension. The integrated dataset is then cleaned, normalized, and its features extracted to create more comprehensive investment characteristics. This improves the quality of input data for subsequent asset allocation strategy recommendation models, thereby optimizing the effectiveness of the recommendations.

[0119] Based on this, the method for obtaining historical investment behavior data of users in the current period provided in this embodiment may include: obtaining historical investment behavior data of multiple users in the financial system; cleaning and processing the historical investment behavior data of multiple users to form a historical investment behavior dataset; extracting investment behavior indicators from the historical investment behavior dataset according to a preset period to obtain corresponding multi-dimensional investment indicators; using multi-dimensional investment indicators to reflect investment preferences and risk attributes; clustering investment behavior characteristics of multiple users according to the multi-dimensional investment indicators to identify user groups with similar investment behavior characteristics; and fusing the aggregated investment behavior data of the user group to which the user belongs with the user's historical investment behavior data to generate the user's historical investment behavior data.

[0120] Specifically, the raw historical investment behavior data of multiple users is obtained in batches from financial systems (such as CRM, trading systems, fund clearing platforms, etc.) through data interfaces. The data is then cleaned, including removing abnormal transaction records, filling in missing fields, and unifying the data format to form a standardized historical investment behavior dataset.

[0121] Subsequently, multidimensional indicators that reflect investment preferences and risk attributes are extracted from the dataset according to a preset period (such as monthly or quarterly), such as portfolio volatility, average holding period, industry concentration, proportion of risky assets, turnover rate and return stability.

[0122] Furthermore, clustering algorithms such as K-means or hierarchical clustering are used to segment users based on these multidimensional indicators, identifying user groups with similar investment behavior characteristics.

[0123] Based on this, aggregated investment behavior data of the target user group is integrated with their personal historical data. For example, through weighted averaging or feature splicing, enhanced version of user historical investment behavior data is generated, providing a more comprehensive and robust data foundation for subsequent personalized asset allocation strategies.

[0124] This effectively enriches the amount and dimensions of historical investment behavior data for individual users. Especially when individual user data is insufficient, the comprehensiveness and accuracy of investment feature extraction can be improved by leveraging common group data, providing more reliable data support for subsequent asset allocation strategy recommendation decisions, thereby optimizing the rationality and applicability of asset allocation strategy recommendation decisions.

[0125] In this embodiment, the obtained historical investment behavior data includes at least one of fund transaction data, investment questionnaire data, and investment preference data. Based on this, the method of determining the corresponding investment characteristics according to the historical investment behavior data may include: determining the dynamic investment behavior indicators of users in the next configuration period based on the fund transaction data of user groups; determining the static risk preference indicators of users in the next configuration period based on investment questionnaire data and investment preference data; and inputting the dynamic investment behavior indicators and the static risk preference indicators into a preset feature fusion model to obtain the corresponding investment characteristics.

[0126] In this embodiment, fund transaction data includes, but is not limited to, transaction frequency, changes in holdings, and asset conversion records; investment questionnaire data includes risk assessment scores and investment objective statements; and investment preference data includes product browsing history, sectors of interest, and historical allocation selections.

[0127] Based on this, after obtaining the fund transaction data of the user's group, information is extracted from the above data to obtain dynamic investment behavior indicators of the user in the next allocation cycle, such as expected turnover rate, industry rotation tendency, momentum sensitivity and other time-series indicators that reflect their behavior patterns.

[0128] Meanwhile, based on users' investment questionnaire data and investment preference data, static risk preference indicators for the next period are extracted, such as relatively stable preference parameters like risk tolerance level, expected return, liquidity requirements, and investment period.

[0129] Based on this, dynamic investment behavior indicators and static risk preference indicators are input into a preset feature fusion model. For example, they can be input into a multilayer perceptron or factor decomposition machine based on an attention mechanism for fusion calculation. This enables the model to dynamically weigh behavioral changes and preference constraints, and output a comprehensive multidimensional investment feature vector that represents the user's investment tendency and risk preference in the next cycle, which is used to generate subsequent asset allocation recommendation strategies.

[0130] In the above implementation, by integrating dynamic investment behavior indicators and static risk preference indicators, the changes in users' short-term behavior and long-term investment preferences are unified, improving the comprehensiveness and timeliness of investment feature extraction. Furthermore, by leveraging the multi-dimensional information integration capabilities of the feature fusion model, the adaptability to changes in market conditions is enhanced while accurately depicting user profiles. This provides a reliable basis for generating more robust and personalized asset allocation strategies, improving the accuracy of investment decisions and user satisfaction.

[0131] In this application, investment characteristics include return preference characteristics, risk tolerance characteristics, and liquidity demand characteristics; the feature fusion model includes a return preference feature extraction sub-model, a risk tolerance feature extraction sub-model, and a liquidity demand feature extraction sub-model.

[0132] Based on this, one implementation method for inputting dynamic investment behavior indicators and static risk preference indicators into a preset feature fusion model to obtain corresponding investment features may include: inputting dynamic investment behavior indicators and static risk preference indicators into a return preference feature extraction sub-model to obtain the user's return preference features; inputting dynamic investment behavior indicators and static risk preference indicators into a risk tolerance feature extraction sub-model to obtain the user's risk tolerance features; inputting dynamic investment behavior indicators and static risk preference indicators into a liquidity demand feature extraction sub-model to obtain the user's liquidity demand features; and vectorizing and concatenating the return preference features, risk tolerance features, and liquidity demand features to generate investment features.

[0133] In this embodiment, dynamic investment behavior indicators and static risk preference indicators are input into three dedicated sub-models in parallel.

[0134] Specifically, the profit preference feature extraction sub-model analyzes the profit performance in dynamic indicators and the profit expectation in static indicators to output profit preference features that reflect users' profit goals and profit preferences; the risk tolerance feature extraction sub-model combines risk exposure and drawdown in dynamic indicators and risk tolerance level in static indicators to output risk tolerance features that characterize users' risk aversion and risk control preferences; and the liquidity demand feature extraction sub-model outputs liquidity demand features that reflect users' capital flexibility and liquidity preferences based on capital turnover rate and asset realization speed in dynamic indicators and liquidity requirements in static indicators.

[0135] Based on this, the three types of features are standardized and vectorized, and then integrated into a high-dimensional structured investment feature vector through splicing or weighted fusion to comprehensively describe users' overall preferences in the three dimensions of return, risk and liquidity.

[0136] In the above implementation, by adopting a multi-sub-model parallel processing structure and specialized feature extraction, the ability to deeply characterize users' returns, risks, and liquidity preferences is enhanced. Furthermore, by fusing multi-source information based on dynamic behavior and static questionnaire data, the dimensional richness and indicator targeting of investment feature construction are improved, laying a reliable data foundation for generating asset allocation strategies that are highly matched with users' real needs. This results in improved personalization of investment strategy recommendations and enhanced portfolio performance.

[0137] Based on the macroeconomic factor characteristics and investment characteristics obtained from the above implementation methods, they can be calculated using a preset asset allocation strategy recommendation method, and the weight adjustment actions of various assets in the next allocation cycle can be output to form a specific asset allocation strategy recommendation result.

[0138] In one optional implementation, the process of recommending asset allocation strategies may include: obtaining a pre-trained asset allocation strategy recommendation model; inputting macroeconomic factor features and investment features into the asset allocation strategy recommendation model; and outputting asset allocation strategy recommendation results adapted to the next allocation cycle.

[0139] Specifically, a pre-defined asset allocation strategy recommendation model is obtained; optionally, machine learning models, neural network models, or mathematical algorithms may be used, depending on the specific needs of the actual scenario.

[0140] The macroeconomic factor features and investment features obtained after processing are standardized according to the format required by the model to remove data noise and unify data dimensions, and then input into the asset allocation strategy recommendation model. Subsequently, the model performs fusion analysis and calculation on the input macroeconomic factor features and investment features, and based on the built-in algorithm logic, outputs the specific allocation ratios and adjustment suggestions for various assets in the next allocation cycle, forming the asset allocation strategy recommendation results.

[0141] The above implementation method generates asset allocation strategy recommendations for the next period by inputting macroeconomic factor characteristics and investment characteristics into a preset asset allocation strategy recommendation model. This provides users with more accurate and personalized asset allocation strategy recommendations, which not only improves the synchronization between asset allocation strategy recommendations and market changes, but also enhances the adaptability to investors' risk preferences and return objectives.

[0142] In one optional implementation, the asset allocation strategy recommendation model is a reinforcement learning-based strategy generation model, including a state space, an action space, and a reward / penalty function. The process of inputting macroeconomic factor features and investment feature vectors into the asset allocation strategy recommendation model, and outputting personalized asset allocation strategy recommendation results through multi-dimensional fusion and strategy reasoning by the model may include: concatenating macroeconomic factor features and investment feature vectors to form a comprehensive state representation, which is used as the input to the state space; defining the generated asset allocation actions as the action space, and the actions include asset class allocation ratio adjustment operations.

[0143] A reward function is constructed, comprising a basic reward and a penalty. The basic reward is calculated based on risk-adjusted return metrics, including the Sharpe ratio and the Calmar ratio. The penalty includes transaction cost penalties, turnover rate penalties, and risk deviation penalties. The transaction cost penalty is calculated based on the fees and transaction amounts incurred from asset transactions. The turnover rate penalty is calculated based on the ratio of the total buy and sell transaction amount of the portfolio within a preset period to the average asset size. The risk deviation penalty is calculated based on the difference between the volatility and value at risk of the actual portfolio and the user's target risk level. An asset allocation strategy recommendation model perceives the environmental state in the state space, explores allocation strategies in the action space, and optimizes strategy selection based on the feedback from the reward function to obtain the asset allocation strategy recommendation result for the next allocation period.

[0144] Specifically, macroeconomic factor features and investment features are concatenated, and the resulting fused features are defined as the state space of the reinforcement learning model. Asset allocation strategy recommendations are also generated, such as the weight adjustment ratio of various assets and specific buying and selling operations, and these asset allocation strategy recommendations are defined as the action space of the model.

[0145] Subsequently, the base reward is determined based on the risk-adjusted return, and the penalty items are determined based on transaction cost penalties, turnover rate penalties, and customer risk deviation penalties.

[0146] Specifically, the basic reward can be obtained by calculating the ratio of portfolio excess return to volatility; for example, the Sharpe ratio, the Calmar ratio obtained by the ratio of return to maximum drawdown, etc.

[0147] Specifically, transaction cost penalties can be calculated by statistically analyzing fees such as handling fees and taxes incurred in asset transactions; turnover rate penalties can be calculated by the ratio of the amount of buying and selling of assets in a portfolio to the total market value within a specific period; and risk deviation penalties can be calculated by comparing the difference between the risk level recommended by the actual asset allocation strategy and the client's target risk level. The risk level recommended by the actual asset allocation strategy can be obtained by calculating the volatility of the portfolio return and the Value at Risk (VaR).

[0148] It should be noted that during reinforcement learning model training, the model takes the state space as input, selects asset allocation strategies from the action space to recommend actions to be executed, and calculates the comprehensive reward value by combining basic rewards and penalty terms. Through continuous iterative learning, the action selection strategy is optimized so that the model can select actions that bring higher comprehensive rewards in different states.

[0149] Therefore, in practical applications, namely the process of generating asset allocation strategy recommendation results, based on the generated state space data, the trained allocation strategy is used to output the optimal asset allocation strategy recommendation action from the action space, forming the final asset allocation strategy recommendation result.

[0150] By introducing an asset allocation strategy recommendation model based on reinforcement learning, dynamic response and collaborative optimization to macro market conditions and personalized user needs are achieved. With the help of a multi-dimensional reward function, the model can simultaneously pursue maximum returns and risk control, effectively avoiding problems such as overtrading, style drift, and violation of user risk preferences. Thus, it outputs asset allocation strategies that are adaptive, robust, and executable, thereby improving the long-term performance of the investment portfolio and user satisfaction.

[0151] It is worth noting that the asset allocation strategy recommendations generated before the current allocation cycle switch can be transmitted to individual investors, institutional investors, and fund managers to ensure that the results can be applied in real time to the asset allocation strategy recommendations for the next cycle.

[0152] Specifically, real-time connection with exchanges, brokers, and internal trading systems can be achieved through the FIX protocol or API interface. The instructions are automatically distributed to the corresponding systems before each configuration cycle according to the preset format, so as to facilitate the asset allocation strategy recommendation operation for investors' assets.

[0153] Figure 4 This is a schematic diagram of an asset allocation strategy recommendation device provided in an embodiment of this application. See also... Figure 4 The asset allocation strategy recommendation device 40 includes: a macroeconomic factor feature acquisition module 401, an investment feature acquisition module 402, and a recommendation result generation module 403; wherein,

[0154] The macroeconomic factor feature acquisition module 401 is used to acquire macroeconomic data within the current allocation period and determine the corresponding macroeconomic factor features based on the macroeconomic data; wherein, the macroeconomic factor features are used to reflect the market environment of the next allocation period;

[0155] The investment feature acquisition module 402 is used to acquire the user's historical investment behavior data within the current configuration period, and determine the corresponding investment features based on the historical investment behavior data; wherein, the investment features are used to reflect the user's preferences and risk tolerance.

[0156] The recommendation result generation module 403 is used to generate asset allocation strategy recommendations suitable for the next allocation cycle based on macroeconomic factor characteristics and investment characteristics.

[0157] In one optional implementation, the macroeconomic factor feature acquisition module 401 includes:

[0158] The historical feature acquisition submodule is used to preprocess macroeconomic data and construct factors to generate a historical macroeconomic factor feature sequence within the current configuration period.

[0159] The initial factor feature acquisition submodule is used to input the historical macroeconomic factor feature sequence into the pre-trained macroeconomic factor prediction model to obtain the corresponding initial macroeconomic factor feature set for the next configuration period.

[0160] The factor feature acquisition submodule is used to perform significance analysis and screening on the initial set of macroeconomic factor features to obtain the macroeconomic factor features corresponding to the next configuration period.

[0161] In one optional implementation, the investment feature acquisition module 402 includes:

[0162] The dataset acquisition submodule is used to acquire historical investment behavior data of multiple users in the financial system, clean and process the historical investment behavior data of multiple users, and form a historical investment behavior dataset.

[0163] The investment indicator acquisition submodule is used to extract investment behavior indicators from historical investment behavior datasets according to preset periods to obtain corresponding multi-dimensional investment indicators; the multi-dimensional investment indicators are used to reflect investment preferences and risk attributes.

[0164] The user group identification submodule is used to cluster the investment behavior characteristics of multiple users based on multi-dimensional investment indicators, and identify user groups with similar investment behavior characteristics.

[0165] The historical investment behavior data acquisition submodule is used to merge and process the aggregated investment behavior data of the user's user group with the user's historical investment behavior data to generate the user's historical investment behavior data.

[0166] In one optional implementation, historical investment behavior data includes at least one of fund transaction data, investment questionnaire data, and investment preference data;

[0167] Investment feature acquisition module 402 includes:

[0168] The investment indicator determination submodule is used to determine the dynamic investment behavior indicators of users in the next configuration cycle based on the capital transaction data of user groups.

[0169] The investment characteristics determination submodule is used to determine the user's static risk preference index in the next configuration cycle based on investment questionnaire data and investment preference data.

[0170] In one optional implementation, the investment characteristics include return preference characteristics, risk tolerance characteristics, and liquidity demand characteristics; the feature fusion model includes a return preference characteristic extraction sub-model, a risk tolerance characteristic extraction sub-model, and a liquidity demand characteristic extraction sub-model.

[0171] The investment characteristics determination submodule includes:

[0172] The first feature extraction unit is used to input dynamic investment behavior indicators and static risk preference indicators into the return preference feature extraction sub-model to obtain the return preference features corresponding to the user.

[0173] The second feature extraction unit is used to input dynamic investment behavior indicators and static risk preference indicators into the risk tolerance feature extraction sub-model to obtain the user's corresponding risk tolerance features.

[0174] The third feature extraction unit is used to input dynamic investment behavior indicators and static risk preference indicators into the liquidity demand feature extraction sub-model to obtain the liquidity demand features corresponding to the user.

[0175] The investment characteristic determination unit is used to vectorize and concatenate return preference characteristics, risk tolerance characteristics, and liquidity demand characteristics to generate investment characteristics.

[0176] In one alternative implementation, the recommendation result generation module 403 includes:

[0177] The module obtains a submodule, which is used to obtain a preset asset allocation strategy recommendation model;

[0178] The asset allocation strategy recommendation result generation submodule is used to input macroeconomic factor characteristics and investment characteristics into the asset allocation strategy recommendation model and output asset allocation strategy recommendation results adapted to the next allocation cycle.

[0179] In one alternative implementation, the asset allocation strategy recommendation model is a strategy generation model based on reinforcement learning, including a state space, an action space, and a reward / penalty function.

[0180] The asset allocation strategy recommendation result generation submodule includes:

[0181] The state space definition unit is used to concatenate macroeconomic factor characteristics with investment characteristic vectors to form a comprehensive state representation, which serves as the input to the state space.

[0182] The action space definition unit is used to define the generated asset allocation actions as the action space. The actions include the asset class allocation ratio adjustment operation.

[0183] A reward building block, used to construct a reward function, which includes a basic reward term and a penalty term: where,

[0184] The base reward is calculated based on risk-adjusted return metrics, including the Sharpe ratio and the Calmar ratio.

[0185] The penalties include transaction cost penalties, turnover rate penalties, and risk deviation penalties; among them,

[0186] Transaction cost penalties are calculated based on the fees and transaction amount incurred during the asset transaction.

[0187] The turnover rate penalty is calculated based on the ratio of the total trading volume of the portfolio to the average asset size within a preset period.

[0188] Risk deviation penalty is calculated based on the degree of difference between the volatility and value at risk of the actual portfolio and the user's target risk level;

[0189] The recommendation result generation unit is used to perceive the environmental state in the state space, explore the allocation strategy in the action space, and optimize the strategy selection based on the reward function feedback to obtain the asset allocation strategy recommendation result for the next allocation cycle.

[0190] Figure 5 This is a block diagram of an electronic device provided in an embodiment of this application. See also... Figure 5 The device 800 may include one or more of the following components: processing component 802, memory 804, power supply component 806, multimedia component 808, audio component 810, input / output interface 812, sensor component 814, and communication component 816.

[0191] Processing component 802 typically controls the overall operation of device 800, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 802 may include one or more modules to facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802.

[0192] Memory 804 is configured to store various types of data to support the operation of device 800. Examples of this data include instructions for any application or method operating on device 800, contact data, phonebook data, messages, pictures, videos, etc. Memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0193] Power supply component 806 provides power to various components of device 800. Power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 800.

[0194] Multimedia component 808 includes a screen that provides an output interface between device 800 and the user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a Touch Panel, the screen may be implemented as a touchscreen to receive input signals from the user. The Touch Panel includes one or more touch sensors to sense touches, swipes, and gestures on the Touch Panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 808 includes a front-facing camera and / or a rear-facing camera. When device 800 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0195] Audio component 810 is configured to output and / or input audio signals. For example, audio component 810 includes a microphone (MIC) configured to receive external audio signals when device 800 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 804 or transmitted via communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

[0196] Input / output interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0197] Sensor assembly 814 includes one or more sensors for providing state assessments of various aspects of device 800. For example, sensor assembly 814 may detect the on / off state of device 800, the relative positioning of components such as the display and keypad of device 800, changes in the position of device 800 or a component of device 800, the presence or absence of user contact with device 800, the orientation or acceleration / deceleration of device 800, and temperature changes of device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 814 may also include optical sensors, such as complementary metal-oxide-semiconductor (CMOS) sensors or charge-coupled device (CCD) sensors, for use in imaging applications. In some embodiments, sensor assembly 814 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0198] Communication component 816 is configured to facilitate wired or wireless communication between device 800 and other devices. Device 800 can access wireless networks based on communication standards, such as WiFi, 4G, or 5G, or combinations thereof. In one exemplary embodiment, communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 816 also includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra Wide Band (UWB), Bluetooth (BT), and other technologies.

[0199] In an exemplary embodiment, device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processors (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0200] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, which can be executed by a processor 820 of the device 800 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0201] A non-transitory computer-readable storage medium, wherein instructions in the storage medium, when executed by a server's processor, enable the server to perform the aforementioned asset allocation strategy recommendation method.

[0202] This application also provides a chip for executing instructions, which is used to execute the technical solution of the asset allocation strategy recommendation method in the above embodiments.

[0203] This application also provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are run on a computer, the computer executes the asset allocation strategy recommendation method described in the above embodiments.

[0204] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium, and when the at least one processor executes the computer program, it can implement the technical solution of the asset allocation strategy recommendation method in the above embodiments.

[0205] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0206] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

[0207] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0208] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for recommending asset allocation strategies, characterized in that, include: Obtain macroeconomic data for the current configuration period, and determine the corresponding macroeconomic factor characteristics based on the macroeconomic data; wherein, the macroeconomic factor characteristics are used to reflect the market environment for the next configuration period; Obtain the user's historical investment behavior data within the current configuration period, and determine the corresponding investment characteristics based on the historical investment behavior data; wherein, the investment characteristics are used to reflect the user's preferences and risk tolerance; Based on the macroeconomic factors and investment characteristics, an asset allocation strategy recommendation result suitable for the next allocation cycle is generated; wherein, the asset allocation strategy recommendation result includes asset class, allocation ratio and adjustment suggestions.

2. The method according to claim 1, characterized in that, Based on the aforementioned macroeconomic data, the corresponding macroeconomic factor characteristics are obtained, including: The macroeconomic data is preprocessed and factor constructed to generate a historical macroeconomic factor feature sequence within the current allocation period; The historical macroeconomic factor feature sequence is input into a pre-trained macroeconomic factor prediction model to obtain the initial macroeconomic factor feature set corresponding to the next configuration period. Significance analysis and screening are performed on the initial set of macroeconomic factor features to obtain the macroeconomic factor features corresponding to the next configuration period.

3. The method according to claim 1, characterized in that, Obtain user's historical investment behavior data within the current period, including: Obtain historical investment behavior data from multiple users in the financial system, clean and process the historical investment behavior data of the multiple users, and form a historical investment behavior dataset; Investment behavior indicators are extracted from the historical investment behavior dataset according to a preset period to obtain corresponding multidimensional investment indicators; the multidimensional investment indicators are used to reflect investment preferences and risk attributes. Based on the multidimensional investment indicators, the investment behavior characteristics of the multiple users are clustered to identify user groups with similar investment behavior characteristics; The aggregated investment behavior data of the user's user group and the user's historical investment behavior data are merged to generate the user's historical investment behavior data.

4. The method according to claim 1, characterized in that, The historical investment behavior data includes at least one of the following: fund transaction data, investment questionnaire data, and investment preference data; Based on the aforementioned historical investment behavior data, the corresponding investment characteristics are determined, including: Based on the fund transaction data of the user group, determine the dynamic investment behavior indicators of the user in the next configuration cycle; Based on the investment questionnaire data and the investment preference data, determine the user's static risk preference index in the next configuration cycle; The dynamic investment behavior indicators and the static risk preference indicators are input into a preset feature fusion model to obtain the corresponding investment characteristics.

5. The method according to claim 4, characterized in that, The investment characteristics include return preference characteristics, risk tolerance characteristics, and liquidity demand characteristics; the feature fusion model includes a return preference characteristic extraction sub-model, a risk tolerance characteristic extraction sub-model, and a liquidity demand characteristic extraction sub-model. The dynamic investment behavior indicators and the static risk preference indicators are input into a preset feature fusion model to obtain the corresponding investment characteristics, including: The dynamic investment behavior indicators and the static risk preference indicators are input into the return preference feature extraction sub-model to obtain the return preference features corresponding to the user. The dynamic investment behavior indicators and the static risk preference indicators are input into the risk tolerance feature extraction sub-model to obtain the risk tolerance features corresponding to the user. The dynamic investment behavior indicators and the static risk preference indicators are input into the liquidity demand feature extraction sub-model to obtain the liquidity demand features corresponding to the user. The investment characteristics are generated by vectorizing and concatenating the aforementioned return preference characteristics, risk tolerance characteristics, and liquidity demand characteristics.

6. The method according to any one of claims 1-5, characterized in that, Based on the aforementioned macroeconomic factors and investment characteristics, asset allocation strategy recommendations suitable for the next allocation cycle are generated, including: Obtain a pre-trained asset allocation strategy recommendation model; The macroeconomic factors and investment characteristics are input into the asset allocation strategy recommendation model, and the resulting asset allocation strategy recommendation is adapted to the next allocation cycle.

7. The method according to claim 6, characterized in that, The asset allocation strategy recommendation model is a strategy generation model based on reinforcement learning, which includes a state space, an action space, and a reward / penalty function. The macroeconomic factor features and the investment feature vector are jointly input into the asset allocation strategy recommendation model. The model performs multi-dimensional fusion and strategy inference, outputting personalized asset allocation strategy recommendation results, including: The macroeconomic factor features are concatenated with the investment feature vector to form a comprehensive state representation, which is then used as the input to the state space. The generated asset allocation actions are defined as the action space, and the actions include asset class allocation ratio adjustment operations. Construct a reward function that includes a basic reward term and a penalty term: where, The basic reward is calculated based on risk-adjusted return metrics; these metrics include the Sharpe ratio and the Calmar ratio. The penalties include transaction cost penalties, turnover rate penalties, and risk deviation penalties; among which... The transaction cost penalty is calculated based on the fee rate and transaction amount generated by the asset transaction. The turnover rate penalty is calculated based on the ratio of the total trading volume of the portfolio to the average asset size within a preset period. The risk deviation penalty is calculated based on the degree of difference between the volatility and value at risk of the actual investment portfolio and the user's target risk level. The asset allocation strategy recommendation model perceives the environmental state in the state space, explores allocation strategies in the action space, and optimizes strategy selection based on the reward function feedback to obtain the asset allocation strategy recommendation result for the next allocation cycle.

8. An asset allocation strategy recommendation device, characterized in that, include: The macroeconomic factor feature acquisition module is used to acquire macroeconomic data within the current allocation period and determine the corresponding macroeconomic factor features based on the macroeconomic data; wherein, the macroeconomic factor features are used to reflect the market environment of the next allocation period; The investment characteristic acquisition module is used to acquire the user's historical investment behavior data within the current configuration period, and determine the corresponding investment characteristics based on the historical investment behavior data; wherein, the investment characteristics are used to reflect the user's preferences and risk tolerance. The recommendation result generation module is used to generate asset allocation strategy recommendation results suitable for the next allocation cycle based on the macroeconomic factor characteristics and the investment characteristics.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images