Product combination analysis method and system based on large model agent

Abstract

The invention relates to a product position combination analysis method and system based on a large model agent, and the method comprises the steps: defining an agent based on a large model, and obtaining a large model agent; obtaining position data and transaction data by using a large model agent, and preprocessing the position data and the transaction data; inputting the preprocessed position data and transaction data into a large model agent for analysis; generating a report according to an analysis result. Data cleaning and preprocessing are automatically completed by using a large model agent technology, so that the data processing efficiency is improved; the agent processes the complex unstructured data by using deep learning and natural language processing technologies, and identifies a complex mode in the user position data; by analyzing the historical data of the user, the agent can generate personalized suggestions, so that the user experience and satisfaction are improved; large-scale data are processed and analyzed in real time, and timely investment suggestions and combination optimization schemes are provided; and through deep analysis and real-time monitoring of the agent, product combination income and risk characteristics are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a product portfolio analysis method and system based on a large model agent. Background Art

[0002] In the field of financial technology, there are a variety of technologies and systems used to analyze and manage clients' assets and investment portfolios. For example, traditional asset management systems can help analyze clients' positions and risk tolerance and provide basic investment advice to clients. These systems usually rely on manual analysis and fixed rules to identify high-quality investment opportunities and build a robust investment portfolio. At the same time, some modern systems have also begun to use data analysis and machine learning technologies to provide users with more accurate and personalized investment advice by analyzing large amounts of market data and user behavior data.

[0003] In addition, some professional asset management software and services can also provide basic position analysis functions, such as displaying the client's asset distribution, investment return rate, risk level, etc. These systems can usually provide some basic charts and reports to help clients understand their investment status and market trends.

[0004] Traditional asset management and position analysis systems often rely too much on manual analysis and fixed rules, resulting in an inefficient and time-consuming analysis process, and are easily affected by the subjective judgment of analysts; existing systems often lack sufficient personalization, and they are usually unable to fully understand and reflect the unique needs, risk preferences and long-term investment goals of each customer; traditional systems have limited data processing capabilities and find it difficult to handle large amounts of real-time market data and customer data, resulting in inaccurate and in-time analysis results; existing technologies and systems often lack intelligent algorithm optimization and are unable to automatically learn and optimize from historical data and market dynamics, resulting in investment advice that is not scientific and accurate enough; existing systems often lack a friendly user interface and interactive design, making it difficult for customers to understand and use the system, reducing user satisfaction and the practical value of the system. Summary of the invention

[0005] In order to address the deficiencies of the prior art, the present invention uses agent technology to provide a more efficient, accurate and personalized product portfolio analysis method to solve the problems existing in traditional methods, such as complex data processing, poor model adaptability, insufficient personalization, and low real-time performance.

[0006] To achieve the above objectives, the technical solutions adopted by the present invention include:

[0007] A product holding portfolio analysis method based on a large model agent, characterized by comprising:

[0008] S1. Define an intelligent agent based on a large model and obtain a large model agent;

[0009] S2. Using the large model agent to obtain the position data and transaction data and pre-process the position data and transaction data;

[0010] S3. Input the preprocessed position data and transaction data into the large model agent for analysis;

[0011] S4. Generate a report based on the analysis results, the report including an analysis report, an evaluation report, a diagnosis report, optimization suggestions, and adjustment reasons.

[0012] Furthermore, the training of the large model agent includes:

[0013] Data collection and preprocessing: collecting financial text data, cleaning, deduplication and formatting the financial text data, including research reports, news and regulatory documents;

[0014] Design pre-training tasks to perform initial training on the large model, wherein the pre-training tasks include a masked language model task, a next sentence prediction task, and a document-level coherence prediction task;

[0015] Multi-task fine-tuning: build sub-task datasets based on each sub-task of product analysis and fine-tune the sub-tasks;

[0016] Reinforcement learning optimization, build simulation environment, design comprehensive reward function, and introduce advanced reinforcement learning algorithms;

[0017] Continuous learning mechanism, design incremental learning algorithm, establish feedback loop, and strengthen the decision-making ability of the agent.

[0018] Furthermore, the method of using the large model agent to obtain the position data and transaction data and preprocessing the position data and transaction data includes:

[0019] Collect data through API interface or data export function;

[0020] The outliers in the data are identified and processed using the anomaly detection algorithm of the large model agent, and the outliers are corrected or deleted.

[0021] Furthermore, the pre-processed position data and transaction data are input into the large model agent for analysis, including:

[0022] Obtain asset feature data from the position data, and use the large model agent to identify, classify, and calculate the proportion of the asset feature data;

[0023] Use a large model agent to perform risk assessment based on position data, including volatility analysis and stress testing;

[0024] Based on the questionnaire and transaction data, the risk tolerance is assessed using a large model agent to obtain the risk level;

[0025] Use the large model agent to analyze the holding data, obtain the proportion and quantity of various funds, and recommend the number of funds to be held. The types of funds include stock funds, bond funds, mixed funds, and index funds;

[0026] Use the large model agent to analyze the stock holding sector and optimize the portfolio configuration. The stock holdings include directly held stocks and stock funds;

[0027] Use the large model agent to analyze and diagnose the funds held in the holdings data.

[0028] Furthermore, the asset feature data in the position data is obtained, and the asset feature data is identified, classified and proportion calculated by the large model agent, including:

[0029] Use the big model agent to analyze asset feature data, identify and classify various types of assets, and obtain asset category identification;

[0030] Obtain the quantity and value of each type of asset in the position data, use the large model agent to calculate the proportion of each type of asset based on the quantity and value of each type of asset, and obtain the asset proportion result;

[0031] Calculate the proportion of equity assets, where the equity assets include stocks and stock funds;

[0032] The use of a large model agent to perform risk assessment based on position data includes:

[0033] Use the big model agent to calculate the volatility of assets using statistical methods to obtain volatility indicators;

[0034] The large model agent simulates extreme market conditions, uses Monte Carlo simulation to generate multiple market scenarios and evaluates portfolio performance;

[0035] The risk tolerance is assessed using a large model agent based on questionnaire surveys and transaction data to obtain a risk level, including:

[0036] Based on the questionnaire survey, a preliminary score was obtained using a large model agent;

[0037] Get behavior scores based on transaction data using a large model agent;

[0038] The big model agent analyzes the current product portfolio and risk preference match to obtain a portfolio score;

[0039] Based on the large model agent, the weight ratio is adjusted, and the preliminary score, behavior score, and combined score are weighted and calculated to obtain a comprehensive score result;

[0040] Mapping the comprehensive scoring result to a specific risk level, wherein the risk level includes conservative, robust, and aggressive;

[0041] The large model agent is used to analyze the holding data, obtain the proportion and quantity of various funds, and recommend the number of funds to be held, including:

[0042] Obtain fund data from holdings data, use the large model agent to analyze the type distribution of various funds, and calculate the proportion of various funds;

[0043] Based on the investment scale, the proportion of various funds and the user's risk tolerance, the large model agent calculates the number of recommended funds;

[0044] The large model agent evaluates the diversification of the investment portfolio based on the proportion of each type of fund and calculates the degree of diversification;

[0045] The large model agent calculates the recommended number of fund holdings and gives dynamic optimization suggestions;

[0046] The use of the large model agent to analyze the stock holding sector and optimize the holding portfolio configuration includes:

[0047] Obtain data on held stocks, including directly held stocks and stock funds, and identify the sector to which the stock holdings belong;

[0048] Using a large model agent to analyze the performance and market characteristics of each sector, the performance and market characteristics of each sector include the investment risk and return characteristics of the sector;

[0049] Use large model agents to optimize portfolio configuration;

[0050] The use of the large model agent to analyze and diagnose the funds held in the holding data includes:

[0051] Get detailed information about the funds held by the user;

[0052] The large model agent evaluates the risk and return characteristics of the funds held.

[0053] Furthermore, the method further comprises:

[0054] Collect new and feedback data;

[0055] Train and fine-tune large model agents;

[0056] Update the knowledge base and analytical models.

[0057] The present invention also relates to a product holding combination analysis system based on a large model agent, which is characterized by comprising:

[0058] The agent definition module is used to define the intelligent agent based on the large model and obtain the large model agent;

[0059] A data preprocessing module, used to obtain position data and transaction data using a large model agent and preprocess the position data and transaction data;

[0060] The agent analysis module is used to input the pre-processed position data and transaction data into the large model agent for analysis;

[0061] The report generation module is used to generate reports based on the analysis results, and the reports include analysis reports, evaluation reports, diagnosis reports, optimization suggestions, and adjustment reasons.

[0062] The present invention also relates to a computer-readable storage medium, characterized in that a computer program is stored on the storage medium, and the computer program implements the above method when executed by a processor.

[0063] The present invention also relates to an electronic device, characterized in that it comprises a processor and a memory;

[0064] The memory is used to store program instructions;

[0065] The processor is used to execute the above method by calling program instructions.

[0066] The present invention also relates to a computer program product, comprising a computer program and / or instructions, characterized in that the computer program and / or instructions implement the steps of the above method when executed by a processor.

[0067] The beneficial effects of the present invention are:

[0068] The product portfolio analysis method and system based on a large model agent described in the present invention are adopted to automatically complete data cleaning and preprocessing using agent intelligent body technology, greatly reducing manual operation time and improving data processing efficiency; the agent uses deep learning and natural language processing technology to process complex unstructured data and identify complex patterns in user position data; by analyzing the user's historical data, the agent can generate personalized suggestions to improve user experience and satisfaction; the agent can process and analyze large-scale data in real time, and provide timely investment suggestions and portfolio optimization solutions; through the agent's deep analysis and real-time monitoring, the product portfolio return and risk characteristics are optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0070] Figure 1 A schematic diagram of a product portfolio analysis method based on a large model agent according to the present invention;

[0071] Figure 2 This is a structural schematic diagram of a product portfolio analysis system based on a large model agent according to the present invention;

[0072] Figure 3 The figure is a schematic structural diagram of an electronic device of the present invention. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present application and the corresponding drawings. It should be understood that the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments, and these embodiments are only used to illustrate the present application and are not used to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0074] like Figure 1 As shown, the embodiment of the present invention discloses a product portfolio analysis method based on a large model agent, and the steps of the process include:

[0075] Step S1. Define an intelligent agent based on a large model and obtain a large model agent.

[0076] In the present invention, the large model agent is an intelligent agent based on a large-scale language model, which is specially designed for product portfolio analysis.

[0077] Specifically, the training of the large model agent includes:

[0078] Data collection and preprocessing: collecting financial text data, cleaning, deduplication and formatting the financial text data, including research reports, news and regulatory documents;

[0079] Design pre-training tasks to perform initial training on the large model. Pre-training tasks may include masked language model (MLM) tasks, next sentence prediction (NSP) tasks, and document-level coherence prediction tasks.

[0080] Multi-task fine-tuning: build sub-task datasets based on each sub-task of product analysis and fine-tune the sub-tasks;

[0081] Reinforcement learning optimization, build simulation environment, design comprehensive reward function, and introduce advanced reinforcement learning algorithms;

[0082] Continuous learning mechanism, design incremental learning algorithm, establish feedback loop, and strengthen the decision-making ability of the agent.

[0083] In this embodiment, the specific implementation process of building and training a large model agent includes the following key steps:

[0084] (1) The GPT3.5 large-scale language model is selected as the basic model. This model performs well in the field of natural language processing and can understand and generate natural language text, providing a strong foundation for subsequent customized training in the financial field.

[0085] (2) Pre-training in the financial field: Integrate a large amount of text data such as financial research reports, news, and regulatory documents to pre-process the data; design pre-training tasks, including masked language model (MLM) tasks, next sentence prediction (NSP) tasks, and document-level coherence prediction tasks, to enhance the model's ability to understand financial texts; use distributed training during training to use multiple GPUs or multiple clusters for efficient training to improve training efficiency.

[0086] (3) Collect and annotate data for each subtask of product analysis, and optimize multiple related tasks at the same time, such as product classification, risk assessment, etc., optimize the performance of the model on multiple related tasks, and use progressive fine-tuning or adapter methods to accelerate the training of specific tasks so that the model can quickly adapt to new tasks.

[0087] (4) Construct a simulated financial market environment to provide a relatively safe environment to test and optimize the decision-making ability of the model; design a comprehensive reward function based on multiple dimensions such as investment return, risk control, and user satisfaction to train the model to make better decisions; use proximal policy optimization (PPO) to implement advanced reinforcement learning algorithms to improve the model's adaptability and decision-making quality in complex financial markets.

[0088] (5) Design an incremental learning algorithm to continuously update the knowledge base so that the large model agent can learn from new data to adapt to changes in the financial market; establish a feedback loop so that the large model agent can self-adjust and optimize according to actual performance and continuously optimize model performance.

[0089] (6) Use historical data for backtesting to verify the performance of the model on historical data; conduct A / B testing in simulated markets and actual market environments to compare the performance of different models or strategies and determine the best model or strategy; establish a continuous monitoring mechanism to evaluate the decision quality of the large model agent in real time and make timely adjustments as needed.

[0090] Through the above steps, we can obtain a large model agent with deep financial knowledge, strong analytical capabilities, and the ability to adapt and operate reliably in the actual market environment. This large model agent is the core of product portfolio analysis.

[0091] In the embodiment of the present invention, the large model agent constructed according to the above steps has the following characteristics:

[0092] (1) Multimodal input processing capability: Ability to process multiple data types such as text, numbers, and images;

[0093] (2) In-depth understanding of financial knowledge: Have a deep understanding of financial markets, fund operations and investment strategies;

[0094] (3) Personalized analysis capabilities: Ability to provide customized analysis and suggestions based on the user's specific situation;

[0095] (4) Real-time learning and adaptation: Able to continuously learn and optimize from the latest market data and user feedback.

[0096] Step S2: using the large model agent to obtain the position data and transaction data and pre-process the position data and transaction data.

[0097] Obtain relevant holdings and transaction data from the user's bank account, brokerage account and fund platform. Through the API interface or data export function, collect the user's product holding portfolio information, including product name, asset classification and proportion, asset volume, historical transactions, etc., to ensure that comprehensive and accurate user holding information is obtained, providing basic data for subsequent analysis.

[0098] The data is preprocessed using the large model agent, and the anomaly detection algorithm of the large model agent is used to identify and process the outliers in the acquired data, and the outliers are corrected or deleted. The anomaly detection algorithm is used to identify the outliers obtained from the user's position data, and the identified outliers are corrected (such as filling missing values) or deleted to ensure data quality and data reliability. The data is cleaned, feature engineered, and standardized to form a structured data set, and the preprocessed user position data is obtained.

[0099] Step S3: Input the preprocessed position data and transaction data into the large model agent for analysis.

[0100] In the embodiment of the present invention, the analysis process mainly includes the following steps:

[0101] (1) Obtain asset feature data from the position data, and use the large model agent to identify, classify, and calculate the proportion of the asset feature data.

[0102] Obtain asset feature data from user holdings, analyze these features through the big model agent using a pre-trained classification model, automatically identify and classify various types of assets, and obtain the processed asset category identification. The big model agent can quickly and accurately classify assets into different categories, such as stocks, bonds, funds, cash, etc., providing a basis for subsequent proportion calculations.

[0103] Obtain the quantity and value data of various assets in the user's holdings, and use the large model agent to calculate the proportion of various assets based on these data to obtain the processed asset proportion results. Help users intuitively understand their asset allocation, so as to assess investment risks and profit potential, and thus support more reasonable investment decisions. For example, input a user's holdings data and output:

[0104] [Total assets: 1,035,000 yuan

[0105] Cash ratio: 48.31%

[0106] Stock share: 15.94%

[0107] Fund share: 16.43%

[0108] Other investment proportion: 19.32%]

[0109] Calculate the proportion of equity assets, which include stocks and stock funds. Obtain the stock component data of the stocks and funds directly held by the user. The big model agent integrates and calculates the above data to obtain the total value of the user's equity assets and the proportion of equity assets, thereby helping the user to evaluate the proportion of their investment in equity assets, understand the risk-return characteristics, and assist in decision-making. For example, based on a user's holdings data, use the big model agent to output:

[0110] [Total value of equity assets: 225,000 yuan

[0111] Equity assets: 21.74%]

[0112] (2) Using a large model agent to perform risk assessment based on position data, the risk assessment includes volatility analysis and stress testing.

[0113] The agent obtains the historical price data of the assets held by the user, and uses statistical methods to calculate the volatility of the above assets, obtains the volatility index of various assets, and provides users with quantitative analysis of the risk of asset price changes to help them understand potential market risks.

[0114] By obtaining historical market data and user position data, the agent simulates extreme market conditions (such as a stock market crash or economic recession) and uses Monte Carlo simulation to generate multiple market scenarios and evaluate portfolio performance, helping users evaluate the stability and risk resistance of their portfolios in different market environments, thereby optimizing investment strategies. For example, in a simulated market decline of 20%, a user's portfolio is expected to lose 12%.

[0115] (3) Use the big model agent to assess risk tolerance based on the questionnaire and transaction data to obtain the risk level. User risk tolerance assessment is a key step in formulating personalized investment strategies for investors. Combining traditional questionnaires with big model agent behavior analysis can obtain more accurate assessment results.

[0116] Risk tolerance assessment includes multi-dimensional data collection, behavioral pattern analysis, portfolio assessment, risk tolerance scoring, risk level classification, and personalized risk tolerance improvement suggestions.

[0117] Collect questionnaires and historical transaction data, and analyze the user's behavior patterns. The user fills out a risk tolerance questionnaire, which includes questions about financial status, investment experience, risk preference, etc. The large model agent is used to organize and analyze the data filled out by the user, quantify the user's risk preference and investment goals, and obtain a preliminary score. The user's risk tolerance is evaluated in a quantitative and qualitative way to provide basic information for subsequent analysis; the user's historical transaction data is obtained, including transaction frequency, loss tolerance, profit and loss performance, etc. The agent analyzes the historical transaction data, identifies the user's past investment behavior and decision-making pattern, and obtains a behavior score. Based on the user's transaction pattern and risk tolerance characteristics, the user's real risk tolerance is extracted from the historical transaction records, and the accuracy of risk assessment is further improved to provide support for personalized risk assessment. In this embodiment, the transaction frequency, maximum single loss, profit and loss ratio, etc. in the past year can be obtained through historical transaction data. In this embodiment, the full score can be set to 100 points. The user's preliminary score and behavior score can be obtained by analyzing the user's behavior.

[0118] The large model agent analyzes the matching degree between the user's current product portfolio and risk preference to obtain a portfolio score. It evaluates the risk and return characteristics of the current portfolio and provides the user with portfolio optimization suggestions. In this embodiment, the full score of the portfolio score is set to 100 points.

[0119] Obtain various scoring data (including preliminary scoring, behavioral scoring and combined scoring), the large model agent determines the weight ratio, and uses the weighted algorithm to calculate the comprehensive scoring result, providing users with a comprehensive risk tolerance assessment to help them better understand their own risk tolerance in investment decisions; regularly re-evaluate the user's risk tolerance. In this embodiment, the re-evaluation mechanism is triggered every quarter or after a major market event.

[0120] The comprehensive scoring result is mapped to a specific risk level. In this embodiment, the risk levels may include conservative (score < 30), robust (score >= 30 and score < 60), aggressive (score >= 60), etc. The risk level can provide users with intuitive risk awareness and facilitate them to make corresponding investment decisions.

[0121] By obtaining the user's risk level and related data, the agent generates personalized risk capacity improvement suggestions based on the user's specific situation. It helps users identify their own risk management shortcomings and provides corresponding education and resources to improve their risk tolerance and the effectiveness of their investment strategies.

[0122] (4) Use the large model agent to analyze the holding data, obtain the proportion and quantity of each type of fund, and recommend the number of funds to be held. This step mainly uses the data of held funds. The analysis of the number of held funds aims to evaluate the diversification of the user's investment portfolio and provide optimization suggestions. This process considers multiple influencing factors to balance diversification and management efficiency.

[0123] The fund data held by the user is obtained from the holding data. The large model agent integrates the data based on the type, size, historical performance, etc. of the fund to obtain the integrated fund data. The agent performs distribution analysis on different types of funds and counts the investment proportion of each type of fund. The fund types and proportions invested by the user are identified to help the user understand the diversity and risk characteristics of the investment portfolio. In this embodiment, the fund types may include stock type, bond type, hybrid type, index type, etc.

[0124] Based on the investment scale, the proportion of various funds and the risk tolerance of users, the large model agent calculates the number of recommended funds, provides users with fund selection suggestions that suit their risk preferences, and optimizes investment portfolio allocation. In this embodiment, for primary investors, it is recommended to hold 5-10 funds; for medium-sized investors, it is recommended to hold 10-20 funds; for professional investors, it is recommended to hold 15-25 funds.

[0125] The investment proportion of each type of fund held by the user is obtained, and the agent evaluates the diversification of the investment portfolio, calculates the degree of diversification, quantifies the risk diversification of the user's investment portfolio, and reminds the user to avoid over-concentrated investment and reduce potential risks. In this embodiment, indicators such as the Herfindahl-Hirschman Index (HHI) can be used to evaluate concentration and dispersion.

[0126] Based on the degree of diversification, market conditions and user characteristics, the agent provides users with dynamic optimization suggestions, adjusts investment strategies in a timely manner according to changes in market conditions and users' investment situations, and improves the overall performance of the investment portfolio and risk management capabilities.

[0127] (5) Use the large model agent to analyze the stock holding sectors and optimize the portfolio configuration. Stock holding sector analysis is a key step in evaluating the diversification and potential risks of the investment portfolio. It involves a comprehensive analysis of stocks held directly and indirectly through funds to ensure that the investment portfolio is appropriately diversified across different industries and sectors.

[0128] The agent obtains the fund data held by the user from the holdings data, classifies and analyzes various types of assets, collects and classifies stock holdings information, analyzes fund quarterly reports, and extracts the top ten holdings. It identifies the main stocks invested by the user and the sectors they belong to, laying the foundation for subsequent in-depth analysis.

[0129] Based on the analysis results of the above-mentioned agent, the performance of each sector and its market characteristics are obtained, with special attention paid to the investment risk and return characteristics of each sector.

[0130] Obtain market trend data (such as industry indexes, market news, etc.), and the agent combines it with the user's position data to identify the impact of current market trends on user positions and provide a more timely analysis basis.

[0131] Based on user characteristics, sector analysis and market trends, the agent provides users with personalized portfolio adjustment suggestions to help users optimize their position allocation according to market changes and investment goals, and improve overall investment returns and risk management capabilities.

[0132] (6) Use the large model agent to analyze and diagnose the funds held in the holdings data and analyze the user's funds one by one.

[0133] Obtain detailed information about the funds held by the user, and use the agent to query the latest data of each fund, including historical net value, holding information, fund manager information, performance of similar funds, and other data.

[0134] The big model agent conducts a comprehensive analysis of the above data based on the fund's historical performance, risk indicators and fee structure, evaluates the risk and return characteristics of each fund, and provides users with a comprehensive investment perspective.

[0135] By obtaining market data and macroeconomic indicators, the agent uses predictive models to evaluate the future performance of funds, helping users understand potential risks and returns and supporting more scientific investment decisions.

[0136] Combined with the analysis results, the agent generates personalized investment advice for users, guiding them to optimize the configuration of their held funds to better meet their investment goals and risk tolerance.

[0137] Step S4. Generate a report based on the analysis results, the report including analysis report, evaluation report, diagnosis report, optimization suggestions, and adjustment reasons.

[0138] In the embodiment of the present invention, the large model agent can generate investment suggestions, optimization plans and risk warnings, and present the final results through the user interaction layer.

[0139] In the present invention, a deep learning model is used to replace the traditional rule-based system or expert system for financial analysis, in-depth behavioral analysis is used to replace standardized questionnaires for risk assessment, weight distribution is dynamically adjusted in position analysis instead of a fixed industry weight distribution model, multi-dimensional analysis and prediction are used for fund diagnosis instead of focusing only on the fund's historical performance and fundamental indicators, and the use of advanced AI technology and comprehensive data analysis can realize the processing of complex market dynamics, personalized analysis, dynamic adjustment and comprehensive evaluation.

[0140] The present invention also provides an expansion method for continuous learning and updating, including: collecting new data and feedback data; training and fine-tuning a large model agent; and updating a knowledge base and an analysis model.

[0141] like Figure 2 As shown, the embodiment of the present invention discloses a product holding combination analysis system based on a large model agent, and its structure includes:

[0142] The agent definition module is used to define the intelligent agent based on the large model and obtain the large model agent;

[0143] A data preprocessing module, used to obtain position data and transaction data using a large model agent and preprocess the position data and transaction data;

[0144] The agent analysis module is used to input the pre-processed position data and transaction data into the large model agent for analysis;

[0145] The report generation module is used to generate reports based on the analysis results, and the reports include analysis reports, evaluation reports, diagnosis reports, optimization suggestions, and adjustment reasons.

[0146] By using this system, the above-mentioned operation processing method can be executed and the corresponding technical effects can be achieved.

[0147] Example:

[0148] This paper describes a specific application scenario of a product portfolio analysis system based on a large model agent. An investor user has a portfolio of multiple funds and stocks. The system can be used to help the user analyze and optimize his portfolio through the following steps.

[0149] 1.Agent training and acquisition

[0150] enter:

[0151] Pre-trained model: GPT-3 (175B parameters);

[0152] Financial data set: 10TB of text data, including financial textbooks, research reports, market analysis, etc.

[0153] Market data: daily data for stocks, bonds, and foreign exchange for the past 20 years.

[0154] 1.1 Pre-training in the financial field

[0155] Continue pre-training GPT-3 with 10TB of financial data

[0156] Training duration: 2 weeks

[0157] Result: The perplexity is reduced from the original 20 to 12

[0158] 1.2 Multi-task fine-tuning

[0159] Design 5 tasks:

[0160] a) Market trend forecast

[0161] b) Risk Assessment

[0162] c) Asset Allocation

[0163] d) Financial statement analysis

[0164] e) Macroeconomic analysis

[0165] Each task uses 100,000 labeled data.

[0166] Training time: 5 days

[0167] result:

[0168] The average accuracy increased from 65% to 88%, and the F1 score increased from 0.60 to 0.85.

[0169] 1.3 Reinforcement Learning Optimization

[0170] Build a stock trading simulation environment and use the PPO algorithm for training.

[0171] Training steps: 1 million steps

[0172] Training duration: 1 week

[0173] result:

[0174] In the simulation environment, the Agent's annualized rate of return increased from 5% to 15%, and the maximum drawdown decreased from 30% to 20%.

[0175] 1.4 Implementation of continuous learning mechanism

[0176] Design online learning algorithms and update the models with the latest 1,000 market data every day.

[0177] result:

[0178] The model’s response time to new events was shortened from 24 hours to 2 hours, and the prediction accuracy under abnormal market conditions was increased from 50% to 70%.

[0179] Output:

[0180] The financial agent finally trained has the following capabilities:

[0181] The average accuracy on 5 financial tasks reached 88%;

[0182] Achieve 15% annualized return in a simulated trading environment;

[0183] Ability to adapt to new market conditions within 2 hours;

[0184] Maintain 70% forecast accuracy under abnormal market conditions.

[0185] 2. Analysis of overall account equity position

[0186] Input data:

[0187] User ID: USER_12345

[0188] Account Information:

[0189] Bank deposit: 500,000 yuan

[0190] Stock Account:

[0191] Stock A: 1000 shares, current price 90 yuan / share

[0192] Stock B: 500 shares, current price 150 yuan / share

[0193] Fund Account:

[0194] C Fund (Stock): 50,000 shares, net value 1.2 yuan / share

[0195] D Fund (bond type): 100,000 shares, net value 1.1 yuan / share

[0196] Other investments:

[0197] E financial product: 200,000 yuan, expected annualized return 4.5%

[0198] The processing process is as follows:

[0199] 2.1 Data Collection and Preprocessing

[0200] Agent calls the API to obtain real-time market data and update the current value of each asset.

[0201] Enter the Agent query prompt sentence: "Collect and preprocess all account data, clean up abnormal data, and update the current market value of each asset."

[0202] Result output:

[0203] Total value of stock A: 90,000 yuan

[0204] Total value of B stock: 75,000 yuan

[0205] Total value of C Fund: 60,000 yuan

[0206] Total value of D Fund: 110,000 yuan

[0207] 2.2 Asset Classification and Proportion Calculation

[0208] Divide assets into cash, stocks, funds and other investments, and calculate the proportion of each type of asset.

[0209] Enter the Agent query prompt sentence: "Classify the assets and calculate the proportion of each type of asset."

[0210] Result output:

[0211] Total assets: 1,035,000 yuan

[0212] Cash ratio: 48.31%

[0213] Stock share: 15.94%

[0214] Fund share: 16.43%

[0215] Other investment proportion: 19.32%

[0216] 2.3 Equity Position Analysis

[0217] Calculate the proportion of equity assets (stocks + stock funds).

[0218] Enter the Agent query prompt sentence: "Analyze the total proportion of equity assets."

[0219] Result output:

[0220] Total value of equity assets: 225,000 yuan

[0221] Equity assets accounted for: 21.74%

[0222] 2.4 Risk Assessment

[0223] Use historical volatility and VaR (Value at Risk) models to assess risk.

[0224] Enter the Agent query prompt sentence: "Assess the overall risk of the portfolio, including calculating historical volatility and using Monte Carlo to calculate VaR."

[0225] Result output:

[0226] Annualized portfolio volatility: 8.5%

[0227] 95% VaR (10 days): 2.3%

[0228] 2.5 Report Generation

[0229] Generate analysis reports and recommendations:

[0230] Enter the Agent query prompt sentence: "Generate a detailed analysis report, including an overview of the current situation, risk assessment, and personalized recommendations.

[0231] Please ensure that your analysis is thorough, accurate, and that the results are presented in an understandable manner.”

[0232] Output:

[0233] "Dear user USER_12345:

[0234] Your total account assets are RMB 1,035,000, and the current equity asset ratio is 21.74%, which is at a low level. Considering your risk tolerance assessment results (assuming it is medium), it is recommended to moderately increase the equity asset ratio to 30-40% to increase long-term profit potential.

[0235] The annualized volatility of your investment portfolio is 8.5%, and the maximum possible loss in 10 days at a 95% confidence level is 2.3%. The overall risk is controllable.

[0236] Specific suggestions:

[0237] Consider transferring part of the cash (about 100,000 to 150,000 yuan) into stocks or stock funds to increase the equity ratio.

[0238] You can choose to increase your holdings of C Fund or purchase new index funds to diversify the risks of single stocks.

[0239] Pay close attention to the maturity of E financial products and consider transferring part of the funds to investment products with a higher risk-return ratio.

[0240] Please carefully consider the above suggestions based on your specific situation and market changes. If you have any questions, please feel free to consult us.

[0241] 3. User risk tolerance assessment

[0242] 3.1 Multi-dimensional data collection

[0243] The data were obtained through online questionnaire:

[0244] For example:

[0245] User ID: USER_12345

[0246] Questionnaire Results:

[0247] Age: 35

[0248] Annual income: 500,000 yuan

[0249] Investment experience: 5 years

[0250] Financial goals: retirement savings and children's education fund

[0251] Loss tolerance: Able to accept 20% short-term fluctuations

[0252] Based on the user data obtained previously, historical transaction data is queried and given to the Agent, and then the Agent query prompt is entered: "Analyze the user's historical transaction data and calculate the transaction frequency, maximum single loss, and profit and loss ratio in the past year."

[0253] Result output:

[0254] Transaction frequency in the past year: 3 times per month on average

[0255] Maximum single loss: 8%

[0256] Profit-loss ratio: 1.2

[0257] Risk Tolerance Assessment Process:

[0258] 3.2 Behavior Pattern Analysis

[0259] a) Questionnaire analysis

[0260] Analyze the questionnaire results and assign weights to each factor.

[0261] Enter the Agent query prompt sentence: "Analyze the user's questionnaire results, consider factors such as age, income, and investment experience, and give a preliminary score."

[0262] Result output: Preliminary score 75 points (out of 100 points)

[0263] b) Trading behavior analysis

[0264] Evaluate historical trading data.

[0265] Enter the Agent query prompt sentence: "Evaluate the user's historical trading behavior, including trading frequency, loss tolerance, and profit and loss performance, and give a behavior score."

[0266] Result output: Behavior score 70 points (out of 100 points)

[0267] 3.3 Portfolio Evaluation

[0268] Analyze the match between current investment portfolio and risk appetite.

[0269] Enter the Agent query prompt sentence: "Combined with the current investment portfolio status, evaluate the user's actual risk preference and give a portfolio score."

[0270] Result output: combined score 80 points (out of 100 points)

[0271] 3.4 Comprehensive score calculation

[0272] The final score was calculated using a weighted average method.

[0273] Enter the Agent query prompt sentence: "Use the multi-factor model to calculate the user's risk tolerance score"

[0274] Result output: Comprehensive score: (75*0.4)+(70*0.3)+(80*0.3)=75 points

[0275] 3.5 Risk type determination

[0276] Users are grouped into categories based on their scores.

[0277] Enter the Agent query prompt sentence: "Classify users into appropriate risk tolerance types based on their scores"

[0278] Result output: Conservative investors (65-80 points)

[0279] 3.6 Personalized risk capability improvement suggestions

[0280] Enter the Agent query prompt sentence: "Generate a detailed assessment report, including analysis results, risk type determination, and personalized recommendations. Please ensure that the assessment is comprehensive and objective, and provide clear explanations and recommendations."

[0281] Agent output:

[0282] "Dear user USER_12345:

[0283] Based on the information you provided and our comprehensive analysis, your risk tolerance is assessed as follows:

[0284] Comprehensive risk score: 75 points (out of 100 points)

[0285] Risk tolerance type: Conservative investor

[0286] Analysis Notes:

[0287] Your age of 35 and 5 years of investment experience indicate that you are able to take moderate risks.

[0288] Your trading behavior shows a rational attitude towards risk and an acceptance of short-term fluctuations.

[0289] The current investment portfolio basically matches your risk appetite, but there is still room for optimization.

[0290] suggestion:

[0291] Consider increasing the proportion of equity assets moderately to 30-40% to increase long-term return potential.

[0292] You can explore some medium-risk investment products, such as balanced funds or high-quality blue-chip stocks.

[0293] Establish a regular investment plan and use dollar-cost averaging strategies to diversify risks.

[0294] Reassess your risk tolerance annually to ensure your investment strategy stays consistent with changes in your life.

[0295] Please note that this assessment is based on the information you provide and historical data. Changes in market conditions and personal circumstances may affect your actual risk tolerance. It is recommended that you review and adjust your investment strategy regularly.

[0296] If you have any questions, please feel free to consult us."

[0297] 4. Analysis of the number of funds held

[0298] Input the user's position data obtained previously into the Agent.

[0299] Analysis and processing process:

[0300] 4.1 Analysis of current fund holdings

[0301] Analyze the 4 funds currently held.

[0302] Enter the Agent query prompt sentence: "Integrate all fund holdings information of user USER_12345, including the type, size, historical performance, etc. of each fund."

[0303] result:

[0304] "User ID: USER_12345

[0305] Current fund holdings:

[0306] C Fund (Stock): 50,000 shares, net value 1.2 yuan / share

[0307] D Fund (bond type): 100,000 shares, net value 1.1 yuan / share

[0308] F Fund (mixed): 30,000 shares, net value 1.5 yuan / share

[0309] G Fund (index type): 20,000 shares, net value 1.0 yuan / share

[0310] Total user assets: 1,035,000 yuan (obtained from previous analysis)

[0311] Risk tolerance: Moderate (obtained from previous assessment)

[0312] Total fund value = $253,000, accounting for 24.44% of total assets"

[0313] 4.2 Fund Type Distribution Analysis

[0314] Evaluate the distribution of fund types.

[0315] Enter the Agent query prompt sentence: "Analyze the distribution of fund types"

[0316] Results: Stock type 23.72%, bond type 47.43%, mixed type 19.37%, index type 9.48%

[0317] 4.3 Calculation of the number of recommended funds

[0318] The number of recommendations is calculated based on total asset size and risk tolerance.

[0319] Enter the Agent query prompt sentence: "Considering the user's total asset size and risk tolerance, recommend a suitable range of fund quantities"

[0320] Calculation: Number of recommended funds = 5 + (total assets / 500,000) = 5 + (1,035,000 / 500,000) ≈ 7

[0321] 4.4 Dispersion Evaluation

[0322] Concentration was assessed using the Herfindahl-Hirschman Index (HHI).

[0323] Enter the Agent query prompt sentence: "Based on the position data, use the Herfindahl-Hirschman Index to evaluate the concentration"

[0324] Calculation: HHI = 0.2372^2 + 0.4743^2 + 0.1937^2 + 0.0948^2 = 0.3374

[0325] Results: Moderate dispersion (HHI between 0.15 and 0.45)

[0326] 4.5 Optimization Suggestion Generation

[0327] Generate optimization recommendations based on the analysis results.

[0328] Enter the Agent query prompt sentence: "Generate a detailed analysis report including the current status, recommended fund quantity and type, and reasons for adjustment.

[0329] Please ensure that your analysis is comprehensive, your recommendations are actionable, and that they take into account your specific circumstances.”

[0330] "Dear user USER_12345:

[0331] Based on your fund holdings and overall financial situation, our analysis results are as follows:

[0332] Current fund holdings:

[0333] Number of funds held: 4

[0334] Total fund value: $253,000, accounting for 24.44% of total assets

[0335] Analysis results:

[0336] Number of funds: You currently hold 4 funds, which is below the recommended range.

[0337] Type distribution: Bond funds account for a high proportion, and you can consider increasing the proportion of equity funds.

[0338] Recommended number of funds: Considering your asset size and risk tolerance, it is recommended to hold 6-8 funds.

[0339] Diversification: The current concentration index of the portfolio is 0.3374, which is a medium level of dispersion and there is still room for improvement.

[0340] Optimization suggestions:

[0341] Add 2-3 funds, giving priority to the following types:

[0342] 1 large-cap stock fund

[0343] 1 industry-themed fund (such as technology or healthcare)

[0344] 1 QDII fund (for overseas market allocation)

[0345] Moderately reduce the proportion of bond funds held and allocate funds to newly added funds.

[0346] Consider adding a money market fund for cash management and short-term liquidity needs.

[0347] Reason for adjustment:

[0348] Increasing the number of funds will help diversify risks and improve the portfolio's ability to withstand volatility.

[0349] Joining different types of funds can optimize asset allocation and seize more market opportunities.

[0350] A moderate increase in equity funds is in line with your conservative risk tolerance and will help improve your long-term return potential.

[0351] Please note that these recommendations are based on your current financial situation and market environment. Before making any investment decision, it is recommended that you carefully evaluate the specific situation of each fund and consider your personal investment objectives and market expectations. If you need more detailed advice or have any questions, please feel free to ask. "

[0352] 5. Stock holding sector analysis

[0353] Input the user's position data obtained previously to the Agent.

[0354] Input data:

[0355] User ID: USER_12345

[0356] 5.1 Fund holdings data analysis

[0357] Enter the Agent query prompt sentence: "Based on the fund holdings information of user USER_12345, calculate the stock holdings, including direct holdings and indirect holdings through funds, and classify them by industry sector."

[0358] Agent output:

[0359] "The stock holding analysis results of user USER_12345 are as follows:

[0360] Direct shareholding:

[0361] Technology sector: 90,000 yuan (8.70%)

[0362] Financial sector: 75,000 yuan (7.25%)

[0363] Indirect holdings through funds:

[0364] Internet sector: 9,000 yuan (0.87%)

[0365] Consumption sector: 6,000 yuan (0.58%)

[0366] New energy sector: 4,800 yuan (0.46%)

[0367] Financial sector: 7,200 yuan (0.70%)

[0368] Medical sector: 3,600 yuan (0.35%)

[0369] The distribution of the merged sectors:

[0370] Technology / Internet: $99,000 (9.57%)

[0371] Finance: 82,200 yuan (7.94%)

[0372] Consumption: 6,000 yuan (0.58%)

[0373] New energy: 4,800 yuan (0.46%)

[0374] Medical treatment: $3,600 (0.35%)

[0375] Total stock investment: 195,600 yuan (18.90% of total assets)"

[0376] 5.2 In-depth sector analysis

[0377] Enter the Agent query prompt sentence: "Analyze the sector distribution of stock holdings and evaluate each sector's risk exposure, deviation from the market, and future potential."

[0378] Agent output:

[0379] "In-depth plate analysis results:

[0380] Risk exposure:

[0381] The technology / internet sector accounts for the largest proportion and has a greater risk exposure. The volatility of this sector is usually higher than the market average.

[0382] The financial sector is the second largest holding and also has higher systemic risk.

[0383] The other sectors account for a smaller proportion, providing a certain diversification effect.

[0384] Market Deviation:

[0385] Technology / Internet: +3.57% (assuming a 6% market average)

[0386] Financials: +1.94% (assuming a 6% market average)

[0387] Consumption: -2.42% (assuming a market average of 3%)

[0388] New energy: -1.54% (assuming a market average of 2%)

[0389] Healthcare: -1.65% (assuming a market average of 2%)

[0390] Future Potential:

[0391] Technology / Internet: Great long-term growth potential, but may face regulatory and valuation pressures in the short term.

[0392] Finance: Affected by the macro-economy, it may face challenges in the medium term.

[0393] Consumption: has good defensiveness and stable long-term growth.

[0394] New energy: in line with national strategy, with strong long-term growth potential.

[0395] Medical care: With the aging population, it has long-term growth potential.

[0396] 5.3 Market Trend Integration

[0397] Enter the Agent query prompt sentence: "Combining current market trends and industry development prospects, evaluate the future performance potential of each sector."

[0398] Agent output:

[0399] "Market Trend Integrated Analysis:

[0400] Technology / Internet:

[0401] Current trends: Valuations are high, but innovation momentum is strong.

[0402] Future potential: Driven by new technologies such as 5G, AI, and cloud computing, it has good long-term growth prospects.

[0403] finance:

[0404] Current Trend: Affected by economic recovery and interest rate environment.

[0405] Future potential: Fintech innovation may bring new growth points.

[0406] Consumption:

[0407] Current trends: Consumption upgrading continues and demand for high-end consumer goods increases.

[0408] Future potential: Under the domestic circulation strategy, the consumer sector is expected to continue to benefit.

[0409] New Energy:

[0410] Current trends: strong policy support and rapid development of the industrial chain.

[0411] Future potential: Under the carbon neutrality goal, there is huge room for long-term growth.

[0412] Medical:

[0413] Current trends: Medical investment has increased after the epidemic, and the development of innovative drugs has accelerated.

[0414] Future potential: Aging population and increasing medical needs support long-term growth. "

[0415] 5.4 Personalized adjustment suggestions

[0416] Enter the Agent query prompt sentence: "Based on the user's stock holding ratio, risk preference, investment goals and current market trends, provide personalized sector adjustment suggestions."

[0417] Agent output:

[0418] "Personalized adjustment suggestions:

[0419] Considering your conservative risk appetite and current market trends, the following suggestions are made:

[0420] Technology / Internet sector:

[0421] Recommendation: Reduce holdings slightly from 9.57% to around 8%.

[0422] Rationale: Reduce concentration risk but retain sufficient exposure to capture long-term growth.

[0423] Financial sector:

[0424] Recommendation: Keep the existing configuration.

[0425] Rationale: Current allocation is roughly in line with the market and is well positioned to benefit from an economic recovery.

[0426] Consumption sector:

[0427] Recommendation: Increase holdings to 2-3%.

[0428] Reason: To improve the defensiveness of the portfolio and share the dividends of consumption upgrading.

[0429] New Energy Sector:

[0430] Recommendation: Increase holdings to around 2%.

[0431] Reason: Seize the opportunity of rapid growth in the industry and improve the growth potential of the portfolio.

[0432] Medical sector:

[0433] Recommendation: Increase holdings to around 2%.

[0434] Reason: Diversify risks and capture the industry's long-term growth potential.

[0435] Consider adding:

[0436] High-end manufacturing sector: 1-2%

[0437] Semiconductor sector: 1-2%

[0438] Reason: Further diversify risks and seize opportunities in emerging industries.

[0439] Action Suggestions:

[0440] Reduce holdings of some technology / Internet stocks or funds and transfer funds to consumer, new energy and medical sectors.

[0441] Consider buying industry ETFs or thematic funds focused on high-end manufacturing or semiconductors.

[0442] Implement adjustments in steps to avoid major changes in the portfolio structure in the short term.

[0443] Please note that the above suggestions are based on the current market situation and your risk appetite. When actually operating, please combine your specific situation and market judgment. It is recommended that you review and adjust the sector configuration regularly.

[0444] 6. Diagnosis of individual holdings

[0445] Based on the user holdings data obtained previously, analyze the user's funds one by one.

[0446] For example:

[0447] User ID: USER_12345

[0448] Funds to be diagnosed: Fund C (stock type)

[0449] Fund holdings: 50,000 shares

[0450] Current net value: 1.2 yuan / share

[0451] Input to Agent

[0452] 6.1 Fund Data Query

[0453] Enter the Agent query prompt sentence: "Query the historical net value, holding information, fund manager information, performance of similar funds and other data of Fund C."

[0454] Agent output:

[0455] "C Fund data query completed:

[0456] Historical net worth data: daily net worth in the last three years

[0457] Current top ten holdings: [Stock list]

[0458] Fund manager: Zhang San, 5 years of experience

[0459] Average performance of similar funds: annualized return 8%, maximum drawdown 15%

[0460] 6.2 Multi-dimensional Analysis

[0461] Enter the query prompt sentence of Agent: "Based on the collected data, analyze the fund from multiple dimensions such as risk, return, management ability, market adaptability and future potential, and calculate indicators such as maximum drawdown, volatility, Sharpe ratio, annualized rate of return, excess return, and relative benchmark win rate."

[0462] Agent output:

[0463] "C Fund multi-dimensional analysis results:

[0464] Risk indicators:

[0465] Maximum drawdown: 18%

[0466] Volatility: 15% (annualized)

[0467] Revenue indicators:

[0468] Annualized rate of return: 10.5%

[0469] Sharpe Ratio: 0.7

[0470] Relative performance:

[0471] Excess return: 2.5% (relative to the same category average)

[0472] Relative Benchmark Win Rate: 65%

[0473] Management Capacity Assessment:

[0474] Fund manager Zhang San's historical performance is better than the average of his peers

[0475] Stable investment style, good at stock picking

[0476] Market adaptability:

[0477] Balanced performance in different market environments

[0478] Strong ability to control downside risks

[0479] Future potential:

[0480] The current holding structure is reasonable, and most of the heavily-weighted stocks are industry leaders

[0481] The investment theme is in line with current market hot spots and long-term trends"

[0482] 6.3 Predictive Analysis

[0483] Enter the Agent query prompt sentence: "Based on historical data and market forecasts, estimate the possible performance of the fund in the next 6-12 months."

[0484] Agent output:

[0485] "C Fund's forecast for the next 6-12 months:

[0486] Baseline scenario: expected annualized return of 9-11%

[0487] Optimistic scenario: expected annualized return of 12-15%

[0488] Pessimistic scenario: expected annualized return of 5-7%

[0489] Main influencing factors:

[0490] Macroeconomic trends

[0491] Industry policy changes

[0492] Market style shift

[0493] Risks to watch out for:

[0494] The concentration of heavily weighted stocks is high, which may increase volatility

[0495] If the market style shifts to value, it may affect short-term performance."

[0496] 6.4 Personalized Recommendation Generation

[0497] Enter the Agent query prompt sentence: "Consider the user's investment goals, risk tolerance and current market views, and generate personalized investment recommendations based on fund analysis results."

[0498] Agent output:

[0499] "Personalized suggestions for user USER_12345:

[0500] Considering your conservative risk tolerance and investment goal of long-term wealth appreciation, the following are the recommendations for Fund C:

[0501] Holding recommendation: Maintain current position

[0502] Reason: The fund's risk-return characteristics match your investment objectives and it has excellent long-term performance

[0503] Allocation ratio: You can consider increasing your holdings slightly, but it is recommended to control it within 10-15% of total assets

[0504] Reason: Maintain moderate diversification and avoid excessive proportion of a single fund

[0505] Fixed investment strategy: It is recommended to adopt a regular fixed amount investment strategy

[0506] Reason: It can smooth market fluctuations and reduce timing risks

[0507] Focus:

[0508] Check fund reports regularly and pay attention to changes in heavily-weighted stocks

[0509] Pay attention to changes in fund managers

[0510] Pay attention to the impact of market style changes on fund performance

[0511] Rebalancing: Rebalance every six months or annually based on overall asset allocation Reason: Ensure that the portfolio always meets your risk-return objectives"

[0512] 6.5 Report Generation

[0513] Enter the Agent query prompt sentence: "Based on the fund analysis results and recommendations, generate a detailed but easy-to-understand report that explains the analysis logic and reasons for the recommendations."

[0514] Agent output:

[0515] "C Fund Diagnostic Report

[0516] Dear user USER_12345,

[0517] Based on a comprehensive analysis of C Fund, we have generated the following diagnostic report for you:

[0518] Fund Overview

[0519] Fund C is a stock fund established in [year] and managed by experienced manager Zhang San. The fund mainly invests in [main investment direction].

[0520] Performance Analysis

[0521] Annualized rate of return: 10.5%, higher than the average of 8% in the same category

[0522] Risk control: Maximum drawdown is 18%, slightly higher than the average of the same category, but within an acceptable range. Risk-adjusted return: Sharpe ratio is 0.7, which is a good performance.

[0523] Investment style

[0524] The fund adopts [describe investment style] and performs evenly in different market environments. The fund manager is good at stock selection, and most of the heavily invested stocks are industry leaders.

[0525] Future Outlook

[0526] Based on current market trends and the fund's holdings structure, we expect the fund to continue to perform well in the next 6-12 months, with an expected annualized return of between 9-11%.

[0527] Personalized recommendations

[0528] Considering your conservative risk appetite and long-term investment goals, we recommend:

[0529] Maintain current position and consider a small increase

[0530] Adopt a regular fixed investment strategy

[0531] Control that the fund's share of total assets does not exceed 15%

[0532] Pay attention to fund reports and market changes regularly

[0533] Risk Warning

[0534] The stock market is subject to volatility risks, and the fund's net value may fluctuate with the market

[0535] Past performance is not indicative of future performance

[0536] Please make investment decisions based on your own risk tolerance

[0537] in conclusion:

[0538] Fund C has excellent performance and matches your investment goals. With reasonable allocation and regular monitoring, it is expected to bring long-term and stable returns to your investment portfolio.

[0539] If you have any questions about the report, please feel free to consult. I wish you a smooth investment! "

[0540] An embodiment of the present invention also provides a computer-readable storage medium capable of implementing all the steps in the method in the above embodiment, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all the steps in the method in the above embodiment are implemented.

[0541] The embodiment of the present invention further provides an electronic device for executing the above method, as a device for implementing the method, and its structure is as follows: Figure 3 As shown, the electronic device 300 includes: a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, such as through a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in actual applications, the transceiver 304 is not limited to one. It is understandable that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on the specific structure of the electronic device 300. In other embodiments of the present application, the electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware. Optionally, the electronic device may also include a display screen 305 for displaying images, or receiving user operation instructions when necessary.

[0542] The processor 301 is applied in the embodiment of the present application to implement the method shown in the above method embodiment. The transceiver 304 may include a receiver and a transmitter. The transceiver 304 is applied in the embodiment of the present application to implement the function of the electronic device of the embodiment of the present application communicating with other devices when executed.

[0543] Processor 301 may be a CPU (Central Processing Unit), a general purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0544] The processor 301 may also include one or more processing units, for example, the processor 301 may include an application processor (AP), a modem processor, a graphics processor (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural network processor (NPU). Among them, different processing units may be independent devices or integrated into one or more processors. Among them, the controller may be the nerve center and command center of the electronic device 300. The controller may generate an operation control signal according to the instruction opcode and the timing signal to complete the control of fetching and executing instructions. A memory may also be set in the processor 301 for storing instructions and data. In some embodiments, the memory in the processor 301 is a high-speed cache memory. The memory may store instructions or data that the processor 301 has just used or circulated. If the processor 301 needs to use the instruction or data again, it may be directly called from the memory. Repeated access is avoided, the waiting time of the processor 301 is reduced, and the efficiency of the system is improved.

[0545] The processor 301 can run the method provided in the embodiment of the present application. The processor 301 can include different devices. For example, when a CPU and a GPU are integrated, the CPU and the GPU can cooperate to execute the method provided in the embodiment of the present application, for example, part of the algorithm in the method is executed by the CPU, and another part of the algorithm is executed by the GPU to obtain faster processing efficiency.

[0546] The bus 302 may include a path to transmit information between the above components. The bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 302 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0547] The memory 303 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory), or may include a high-speed random access memory, or may include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this.

[0548] Optionally, the memory 303 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the method provided in any embodiment of the present application.

[0549] The memory 303 can be used to store computer executable program codes, which include instructions. The processor 301 executes various functional applications and data processing of the electronic device 300 by running the instructions stored in the memory 303. The memory 303 may include a program storage area and a data storage area. Among them, the program storage area may store the code of the operating system, application program, etc. The data storage area may store data created during the use of the electronic device 300 (such as images, videos, etc. collected by the camera application), etc.

[0550] The memory 303 may also store one or more computer programs corresponding to the method provided in the embodiment of the present application. The one or more computer programs are stored in the above-mentioned memory 303 and are configured to be executed by the one or more processors 301. The one or more computer programs include instructions, and the above-mentioned instructions can be used to execute each step in the above-mentioned corresponding embodiment.

[0551] Of course, the code of the method provided in the embodiment of the present application can also be stored in an external memory. In this case, the processor 301 can run the code of the method stored in the external memory through the external memory interface, and the processor 301 can control the running process.

[0552] The display screen 305 includes a display panel. The display panel can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode or an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), Miniled, MicroLed, Micro-oLed, a quantum dot light-emitting diode (QLED), etc. In some embodiments, the electronic device 300 may include 1 or N display screens 305, where N is a positive integer greater than 1. The display screen 305 can be used to display information input by a user or information provided to a user and various graphical user interfaces (GUI). For example, the display screen 305 can display photos, videos, web pages, or files.

[0553] The electronic device provided in the embodiment of the present application is applicable to any embodiment of the above method. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, which will not be repeated here.

[0554] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0555] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or boxes.

[0556] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0557] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps of the functions specified in a box or multiple boxes. Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the attached claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0558] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A product holding portfolio analysis method based on a large model agent, characterized in that: include: S1. Define an intelligent agent based on a large model and obtain a large model agent; S2. Using the large model agent to obtain the position data and transaction data and pre-process the position data and transaction data; S3. Input the preprocessed position data and transaction data into the large model agent for analysis; S4. Generate a report based on the analysis results, the report including an analysis report, an evaluation report, a diagnosis report, optimization suggestions, and adjustment reasons.

2. The method according to claim 1, characterized in that The training of the large model agent includes: Data collection and preprocessing: collecting financial text data, cleaning, deduplication and formatting the financial text data, including research reports, news and regulatory documents; Design pre-training tasks to perform initial training on the large model, wherein the pre-training tasks include a masked language model task, a next sentence prediction task, and a document-level coherence prediction task; Multi-task fine-tuning: build sub-task datasets based on each sub-task of product analysis and fine-tune the sub-tasks; Reinforcement learning optimization, build simulation environment, design comprehensive reward function, and introduce advanced reinforcement learning algorithms; Continuous learning mechanism, design incremental learning algorithm, establish feedback loop, and strengthen the decision-making ability of the agent.

3. The method according to claim 1, characterized in that The method of using the large model agent to obtain the position data and transaction data and preprocessing the position data and transaction data includes: Collect data through API interface or data export function; The outliers in the data are identified and processed using the anomaly detection algorithm of the large model agent, and the outliers are corrected or deleted.

4. The method according to claim 1, characterized in that The pre-processed position data and transaction data are input into the large model agent for analysis, including: Obtain asset feature data from the position data, and use the large model agent to identify, classify, and calculate the proportion of the asset feature data; Use a large model agent to perform risk assessment based on position data, including volatility analysis and stress testing; Based on the questionnaire and transaction data, the risk tolerance is assessed using a large model agent to obtain the risk level; Use the large model agent to analyze the holding data, obtain the proportion and quantity of various funds, and recommend the number of funds to be held. The types of funds include stock funds, bond funds, mixed funds, and index funds; Use the large model agent to analyze the stock holding sector and optimize the portfolio configuration. The stock holdings include directly held stocks and stock funds; Use the large model agent to analyze and diagnose the funds held in the holdings data.

5. The method according to claim 4, characterized in that The step of obtaining asset feature data from the position data and identifying, classifying and calculating the proportion of the asset feature data through a large model agent includes: Use the big model agent to analyze asset feature data, identify and classify various types of assets, and obtain asset category identification; Obtain the quantity and value of each type of asset in the position data, use the large model agent to calculate the proportion of each type of asset based on the quantity and value of each type of asset, and obtain the asset proportion result; Calculate the proportion of equity assets, where the equity assets include stocks and stock funds; The use of a large model agent to perform risk assessment based on position data includes: Use the big model agent to calculate the volatility of assets using statistical methods to obtain volatility indicators; The large model agent simulates extreme market conditions, uses Monte Carlo simulation to generate multiple market scenarios and evaluates portfolio performance; The risk tolerance is assessed using a large model agent based on questionnaire surveys and transaction data to obtain a risk level, including: Based on the questionnaire survey, a preliminary score was obtained using a large model agent; Get behavior scores based on transaction data using a large model agent; The big model agent analyzes the current product portfolio and risk preference match to obtain a portfolio score; Based on the large model agent, the weight ratio is adjusted, and the preliminary score, behavior score, and combined score are weighted and calculated to obtain a comprehensive score result; Mapping the comprehensive scoring result to a specific risk level, wherein the risk level includes conservative, robust, and aggressive; The large model agent is used to analyze the holding data, obtain the proportion and quantity of various funds, and recommend the number of funds to be held, including: Obtain fund data from holdings data, use the large model agent to analyze the type distribution of various funds, and calculate the proportion of various funds; Based on the investment scale, the proportion of various funds and the user's risk tolerance, the large model agent calculates the number of recommended funds; The large model agent evaluates the diversification of the investment portfolio based on the proportion of each type of fund and calculates the degree of diversification; The large model agent calculates the recommended number of fund holdings and gives dynamic optimization suggestions; The use of the large model agent to analyze the stock holding sector and optimize the holding portfolio configuration includes: Obtain data on held stocks, including directly held stocks and stock funds, and identify the sector to which the stock holdings belong; Using a large model agent to analyze the performance and market characteristics of each sector, the performance and market characteristics of each sector include the investment risk and return characteristics of the sector; Use large model agents to optimize portfolio configuration; The use of the large model agent to analyze and diagnose the funds held in the holding data includes: Get detailed information about the funds held by the user; The large model agent evaluates the risk and return characteristics of the funds held.

6. The method according to claim 1, characterized in that The method further comprises: Collect new and feedback data; Train and fine-tune large model agents; Update the knowledge base and analytical models.

7. A product holding portfolio analysis system based on a large model agent, characterized in that: include: The agent definition module is used to define the intelligent agent based on the large model and obtain the large model agent; A data preprocessing module, used to obtain position data and transaction data using a large model agent and preprocess the position data and transaction data; The agent analysis module is used to input the pre-processed position data and transaction data into the large model agent for analysis; The report generation module is used to generate reports based on the analysis results, and the reports include analysis reports, evaluation reports, diagnosis reports, optimization suggestions, and adjustment reasons.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store program instructions; The processor is configured to execute the method according to any one of claims 1 to 6 by calling program instructions.

10. A computer program product comprising a computer program and / or instructions, characterized in that: When the computer program and / or the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images