Fund product recommendation method and computer readable storage medium

Abstract

The invention relates to the technical field of data processing, and discloses a fund product recommendation method and a computer readable storage medium. After current market factor exposure, current scale factor exposure, current value factor exposure, historical average market factor exposure, historical average scale factor exposure and historical average value factor exposure of the fund product are acquired, a flexibility score of the fund product is calculated and acquired in combination with a preset factor weight; calculating a scene adjustment factor based on the market emotion index of the current market cycle type, optimizing the initial feature weight combination, and calculating a comprehensive fund score of the fund product with the yield, the volatility and the flexibility score; and classifying fund products and investor risk preferences, obtaining comprehensive fund scores of all fund products of corresponding types based on the investor risk preferences, and selecting first k corresponding fund products from large to small as recommendation results.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a fund product recommendation and a computer-readable storage medium. Background Art

[0002] In the financial sector, the use of algorithms to develop investment recommendation tools for investors has become an important research and development direction. The development of this field can be traced back to the 1990s, when some financial institutions began to explore the use of computer algorithms to execute trading instructions, which was one of the earliest applications of algorithms in the financial sector. Over time, information technology has experienced rapid development, and the ability to obtain data in the financial sector has continued to increase. The application of algorithms in the financial sector has gradually increased, especially in investment recommendations. In the 21st century, quantitative investment began to rise. This investment method emphasizes the use of mathematical models and statistical methods to guide investment decisions. Successful cases of quantitative investment, such as Renaissance Technologies' Medallion Fund, have demonstrated the strong potential of algorithms in investment decisions.

[0003] In recent years, with the advancement of machine learning technology, especially the rise of deep learning, the application of algorithms in investment recommendation tools has reached new heights. Machine learning algorithms can automatically learn and discover complex patterns from large amounts of data, which is essential for tasks such as predicting market trends, assessing investment risks, and recommending investment portfolios.

[0004] In the field of financial technology, especially in the process of fund management and portfolio construction, existing technologies mainly rely on static features in historical fund data (such as historical returns, fund categories) or simple similarity calculations to make fund recommendations; however, historical data may lose its effectiveness in predicting the future due to changes in the market environment, especially when there are structural changes in the market or changes in the economic cycle. Predictions based on historical performance data are not accurate, which leads to recommended products that do not conform to the market cycle and cannot meet the needs of investors. Summary of the invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the problem in the prior art that the cyclical changes in the market are not taken into consideration, resulting in the recommended products not meeting the market cycle and customer needs.

[0006] In order to solve the above technical problems, the present invention provides a fund product recommendation method, comprising: Using the Fama-French three-factor model, we calculated the average values ​​of the fund products’ market factor exposure, scale factor exposure, and value factor exposure in multiple time periods in history, and obtained the historical average market factor exposure, historical average scale factor exposure, and historical average value factor exposure. Use the Fama-French three-factor model to obtain the fund product's current market factor exposure, current scale factor exposure, and current value factor exposure in the current time period; Based on the fund product's current market factor exposure, current scale factor exposure, current value factor exposure and the corresponding historical average market factor exposure, historical average scale factor exposure, historical average value factor exposure, combined with the preset factor weights, calculate and obtain the fund product's corresponding flexibility score; Obtain the current market cycle type based on market volatility, economic growth rate and interest rate level; Calculate the scenario adjustment factor based on the market sentiment index corresponding to the current market cycle type and the initial feature weight combination; the feature weight combination includes yield weight, volatility weight and flexibility weight; Optimize the initial feature weight combination based on the scenario adjustment factor to obtain the optimized feature weight combination; Calculate the comprehensive fund score corresponding to the fund product based on the optimized feature weight combination and the fund product's rate of return, volatility and flexibility score; Based on the comprehensive fund scores corresponding to the fund products, all fund products in the fund pool are clustered and divided into conservative fund sets and risky fund sets; Based on the investor's historical transaction records, the investor's risk preference is obtained, including conservative and risky types; the historical transaction records include transaction frequency, investment fund types, profit and loss records, holding time and asset allocation; Based on the risk preference of investors, the comprehensive fund scores of all fund products in the corresponding type of fund set are obtained, and the comprehensive fund scores are sorted from large to small to obtain the fund products corresponding to the top k as the recommendation results.

[0007] Preferably, the Fama-French three-factor model is used to obtain the market factor exposure, scale factor exposure, and value factor exposure of the fund product within a preset time period, expressed as: ; in, It indicates the expected rate of return of the fund product within the preset time period. It indicates the risk-free rate of return of the fund product within the preset time period. represents the yield benchmark, , and They represent market factor exposure, size factor exposure, and value factor exposure, respectively. represents the market factor, represents the scale factor, represents the value factor, Represents the preset model error term.

[0008] Preferably, the flexibility score of a fund product is expressed as: ; in, Indicates the flexibility score of the fund product. Indicates the number of factor categories, the value is 3, ; Indicates The preset factor weights of the factors, Indicates Factor exposure to factors, Indicates The historical average factor exposure of the factor.

[0009] Preferably, based on the market volatility, economic growth rate and interest rate level of the market, the current market cycle type is obtained, including: When the economic growth rate is greater than 2% and the market volatility is less than 5%, the current market cycle type is determined to be an expansion period; When the economic growth rate is no more than 2% and the market volatility is no less than 5%, the current market cycle type is determined to be a contraction period; When the economic growth rate turns from negative to positive and the market volatility gradually decreases, the current market cycle type is determined to be a recovery period; When the economic growth rate is less than 0 and the market volatility is greater than 10%, the current market cycle type is determined to be a recession.

[0010] Preferably, based on the market sentiment index corresponding to the current market cycle type, combined with the initial feature weight, the scenario adjustment factor is calculated, expressed as: ; in, Indicates the market cycle type is The scenario adjustment factor is Indicates the market cycle type is The market sentiment index, ; Indicates The initial feature weights of the class features, Represents the total number of feature categories in the initial feature weight combination.

[0011] Preferably, the initial feature weight combination is optimized based on the scenario adjustment factor to obtain the optimized feature weight combination, which is expressed as: ; in, Represents the first Optimized feature weights for class features.

[0012] Preferably, based on the optimized feature weight combination and the rate of return, volatility and flexibility score of the fund product, the comprehensive fund score corresponding to the fund product is calculated, which is expressed as: ; in, Indicates the comprehensive fund score corresponding to the fund product. Represents the first Optimized feature weights of class features; Indicates Class features, .

[0013] Preferably, if the fund product is a new fund product without historical data, it includes: All fund products in the fund pool, their corresponding fund managers, investment themes, and market sentiment indexes are used as nodes, and the relationships between nodes are used as edges to construct a knowledge graph; Based on the fund managers and investment themes of the new fund products, in the knowledge graph, using cosine similarity, we search for multiple fund products with similarities higher than the preset similarity to the new fund products to form a similar fund set; The historical data of all fund products in the similar fund set are averaged as the historical data of the new fund product; The historical data includes yield and volatility.

[0014] Preferably, if the investor is a new investor with no historical transaction records, it includes: Based on the basic information of the new investor, among the investors with historical transaction records, multiple investors with similarity higher than a preset threshold are searched to form a similar user set; Calculate the average risk preference of all investors in the similar user set as the risk preference of the new investor; The basic information of the new investor includes gender, age and education level.

[0015] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the fund product recommendation method as described above are implemented.

[0016] The above technical solution of the present invention has the following beneficial effects compared with the prior art: The fund product recommendation method described in the present invention uses the Fama-French three-factor model to calculate the market factor exposure, scale factor exposure and value factor exposure of the fund product to obtain the flexibility score of the fund product; based on the rate of return, volatility and flexibility score of the fund product, a weighted sum is performed to obtain the corresponding comprehensive fund score for fund recommendation, and the market preference, scale and valuation of the fund product are taken into consideration, so that the calculated comprehensive fund score is more accurate; at the same time, the present invention calculates the scenario adjustment factor according to the current market scenario, and uses the scenario adjustment factor to optimize the initial feature weight combination to obtain the optimized feature weight combination that is more in line with the current market cycle, and performs weighted summation on the rate of return, volatility and flexibility score of the fund product, further optimizing the calculation of the comprehensive fund score, so that the recommendation result is more in line with the market cycle, so as to better achieve the expected investment goals of investors.

[0017] When faced with new fund products and new investors, the present invention calculates the cosine similarity between the new fund products and historical fund products, and the similarity between the new investors and historical investors, and infers the feature data corresponding to the new fund products and the new investors, thereby avoiding the defect of relying on historical data in traditional cold start methods. When faced with funds and investors without historical data, the present invention can still have a high recommendation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein: Figure 1 It is a flow chart of the steps of the fund product recommendation method provided by the present invention. DETAILED DESCRIPTION

[0019] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0020] Reference Figure 1 As shown, the flow chart of the fund product recommendation method of the present invention includes the following specific steps: S101: Using the Fama-French three-factor model, calculate the average values ​​of the market factor exposure, scale factor exposure and value factor exposure of the fund products in multiple time periods in history, and obtain the historical average market factor exposure, historical average scale factor exposure and historical average value factor exposure; S102: Using the Fama-French three-factor model, obtain the current market factor exposure, current scale factor exposure, and current value factor exposure of the fund product in the current time period; S103: Based on the current market factor exposure, current scale factor exposure, current value factor exposure of the fund product and the corresponding historical average market factor exposure, historical average scale factor exposure, historical average value factor exposure, combined with the preset factor weights, calculate and obtain the flexibility score corresponding to the fund product; S104: Based on the market volatility, economic growth rate and interest rate level, obtain the current market cycle type; S105: Calculate the scenario adjustment factor based on the market sentiment index corresponding to the current market cycle type and the initial feature weight combination; the feature weight combination includes the yield weight, volatility weight and flexibility weight; S106: Optimizing the initial feature weight combination based on the scenario adjustment factor to obtain an optimized feature weight combination; S107: Calculate the comprehensive fund score corresponding to the fund product based on the optimized feature weight combination and the rate of return, volatility and flexibility score of the fund product; S108: clustering all fund products in the fund pool based on the comprehensive fund scores corresponding to the fund products, and dividing them into a conservative fund set and a risky fund set; S109: Based on the investor's historical transaction records, the investor's risk preference is obtained, including conservative and risky types; the historical transaction records include transaction frequency, investment fund types, profit and loss records, holding time and asset allocation; S110: Based on the risk preference of the investor, the comprehensive fund scores of all fund products in the fund set of the corresponding type are obtained, and the comprehensive fund scores are sorted from large to small to obtain the fund products corresponding to the first k products as the recommendation results.

[0021] In the embodiment of the present invention, the Fama-French three-factor model is used to obtain the market factor exposure, scale factor exposure, and value factor exposure of the fund product within a preset time period, which is expressed as: ; in, It indicates the expected rate of return of the fund product within the preset time period. It indicates the risk-free rate of return of the fund product within the preset time period. represents the yield benchmark, , and They represent market factor exposure, size factor exposure, and value factor exposure, respectively. represents the market factor, represents the scale factor, represents the value factor, Represents the preset model error term.

[0022] Specifically, the historical average market factor exposure, the historical average scale factor exposure, and the historical average value factor exposure are obtained by calculating the market factor exposure, scale factor exposure, and value factor exposure in multiple time periods in history and then finding the average value.

[0023] In the embodiment of the present invention, the flexibility score of the fund product is expressed as: ; in, Indicates the flexibility score of the fund product. Indicates the number of factor categories, the value is 3, ; Indicates The preset factor weights of the factors, Indicates Factor exposure to factors, Indicates The historical average factor exposure of the factor.

[0024] Specifically, in step S104, based on the market volatility, economic growth rate and interest rate level, the current market cycle type is obtained, including: When the economic growth rate is greater than 2% and the market volatility is less than 5%, the current market cycle type is determined to be an expansion period; When the economic growth rate is no more than 2% and the market volatility is no less than 5%, the current market cycle type is determined to be a contraction period; When the economic growth rate turns from negative to positive and the market volatility gradually decreases, the current market cycle type is determined to be a recovery period; When the economic growth rate is less than 0 and the market volatility is greater than 10%, the current market cycle type is determined to be a recession.

[0025] Specifically, during the expansion period, the economy grows rapidly and the market returns are high; during the contraction period, the economic growth slows down, the market returns are low or the volatility increases; during the recovery period, the economy gradually recovers from the recession and the market sentiment is positive; during the recession period, the economy declines and the market has strong risk aversion. The setting of the preset weight factor in the flexibility score of the fund product is related to the current market cycle type. The higher the flexibility score, the more flexible the fund manager of the fund product is in adjusting the factors during the market cycle changes.

[0026] Specifically, in step S105, based on the market sentiment index corresponding to the current market cycle type and combined with the initial feature weight, the scenario adjustment factor is calculated, which is expressed as: ; in, Indicates the market cycle type is The scenario adjustment factor is Indicates the market cycle type is The market sentiment index, ; Indicates The initial feature weights of the class features, Represents the total number of feature categories in the initial feature weight combination.

[0027] Specifically, in step S106, the initial feature weight combination is optimized based on the scenario adjustment factor to obtain an optimized feature weight combination, which is expressed as: ; in, Represents the first Optimized feature weights for class features.

[0028] Specifically, in step S107, based on the optimized feature weight combination and the rate of return, volatility and flexibility score of the fund product, the comprehensive fund score corresponding to the fund product is calculated, which is expressed as: ; in, Indicates the comprehensive fund score corresponding to the fund product. Represents the first Optimized feature weights of class features; Indicates Class features, .

[0029] Specifically, scenario adjustment factors are used to dynamically adjust feature weights to adapt to different market scenarios, thereby improving the flexibility and accuracy of recommendation results.

[0030] Based on the above embodiment, when the current market scenario is an expansion period, the method provided by the present invention is used to calculate the comprehensive fund score corresponding to the fund product, specifically including: ① The yield, volatility and flexibility scores are 8.5, 5.0 and 7.0 respectively; ② The initial weights of the initial feature weight combination corresponding to the yield, volatility and flexibility scores are 0.4, 0.3 and 0.3 respectively; ③ The corresponding scenario adjustment factors are calculated as 1.2, 0.8 and 1.0 respectively; ④ Use the scenario adjustment factor to optimize the initial weights and obtain the corresponding optimized feature weights, which are expressed as: 0.4×1.2=0.48, 0.3×0.8=0.24, 0.3×1.0=0.3; ⑤Normalize the weights of the optimized features, expressed as: 0.48 / (0.48+0.24+0.3)=0.44, 0.24 / (0.48+0.24+0.3)=0.22, 0.3 / (0.48+0.24+0.3)=0.28; ⑥ Calculate the comprehensive fund score corresponding to the fund product, expressed as: ; This embodiment can flexibly respond to different market conditions by dynamically adjusting feature weights to adapt to market scenarios, and the comprehensive scoring method can ensure the scientificity and applicability of the recommendation results.

[0031] Based on the above embodiment, in the embodiment of the present invention, multi-dimensional data of funds and investors are first collected to generate standardized feature vectors that can be used for analysis and matching for subsequent analysis, which specifically includes: ①Data collection: Collect relevant data of fund products, including: historical returns, volatility, maximum drawdown, factor exposure data, fund manager description, and investment theme information; Collect investor data, including questionnaires, transaction records, and obtain information such as investment goals, risk preferences, and topics of interest; Collect market data, including news sentiment index, economic cycle information, etc. ②Text feature extraction: Use pre-trained language models, such as the BERT model, to perform semantic vectorization on fund manager descriptions and investor needs to generate text feature vectors V 文本 .

[0032] ③Numerical feature extraction: Normalize the numerical data such as yield and volatility to obtain the numerical feature vector V 数值 , ensuring comparability of features.

[0033] ④Image feature extraction: Use CNN to extract feature information from asset allocation graphs and income trend graphs and generate image feature vectors V 图像 .

[0034] ⑤Multimodal feature fusion: The text features, numerical features and image features are directly concatenated and fused into a unified feature vector, including: Fund product feature vector: V 基金 =V 文本 +V 数值 +V 图像 ; Investor characteristic vector: V 投资者 =V' 文本 +V' 数值 +V'图像 ; Specifically, if the importance of different types of feature vectors needs to be considered, corresponding weights are set to perform weighted concatenation.

[0035] Based on the above embodiment, if the fund product is a new fund product without historical data, it includes: All fund products in the fund pool, their corresponding fund managers, investment themes, and market sentiment indexes are used as nodes, and the relationships between nodes are used as edges to construct a knowledge graph; Based on the fund managers and investment themes of the new fund products, in the knowledge graph, using cosine similarity, we search for multiple fund products with similarities higher than the preset similarity to the new fund products to form a similar fund set; The historical data of all fund products in the similar fund set are averaged as the historical data of the new fund product; The historical data includes yield and volatility.

[0036] Based on the above embodiment, if the investor is a new investor with no historical transaction records, it includes: Based on the basic information of the new investor, among the investors with historical transaction records, multiple investors with similarity higher than a preset threshold are searched to form a similar user set; Calculate the average risk preference of all investors in the similar user set as the risk preference of the new investor; The basic information of the new investor includes gender, age and education level.

[0037] Specifically, when constructing the knowledge graph in this embodiment, the nodes include funds, fund managers, investment themes and market sentiments, and the edges between the nodes are defined as funds belonging to a certain investment theme, fund managers managing a certain fund, and market sentiments affecting a certain fund; and the strength of the relationship between the nodes can be reflected by assigning weights to the edges. At the same time, the feature vector corresponding to each node can be embedded into the corresponding node using the knowledge graph embedding method.

[0038] When faced with new fund products and new investors, the present invention calculates the cosine similarity between the new fund products and historical fund products, and the similarity between the new investors and historical investors, and infers the feature data corresponding to the new fund products and the new investors, thereby avoiding the defect of relying on historical data in traditional cold start methods. When faced with funds and investors without historical data, the present invention can still have a high recommendation accuracy.

[0039] Based on the above embodiments, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the fund product recommendation method as described above are implemented.

[0040] The fund product recommendation method described in the present invention uses the Fama-French three-factor model to calculate the market factor exposure, scale factor exposure and value factor exposure of the fund product to obtain the flexibility score of the fund product; based on the rate of return, volatility and flexibility score of the fund product, a weighted sum is performed to obtain the corresponding comprehensive fund score for fund recommendation, and the market preference, scale and valuation of the fund product are taken into consideration, so that the calculated comprehensive fund score is more accurate; at the same time, the present invention calculates the scenario adjustment factor according to the current market scenario, and uses the scenario adjustment factor to optimize the initial feature weight combination to obtain the optimized feature weight combination that is more in line with the current market cycle, and performs weighted summation on the rate of return, volatility and flexibility score of the fund product, further optimizing the calculation of the comprehensive fund score, so that the recommendation result is more in line with the market cycle, so as to better achieve the expected investment goals of investors.

[0041] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt 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 codes.

[0042] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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 device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0044] 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 for the functions specified in one or more boxes.

[0045] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A fund product recommendation method, characterized in that: include: Using the Fama-French three-factor model, we calculated the average values ​​of the fund products’ market factor exposure, scale factor exposure, and value factor exposure in multiple time periods in history, and obtained the historical average market factor exposure, historical average scale factor exposure, and historical average value factor exposure. Use the Fama-French three-factor model to obtain the fund product's current market factor exposure, current scale factor exposure, and current value factor exposure in the current time period; Based on the fund product's current market factor exposure, current scale factor exposure, current value factor exposure and the corresponding historical average market factor exposure, historical average scale factor exposure, historical average value factor exposure, combined with the preset factor weights, calculate and obtain the fund product's corresponding flexibility score; Obtain the current market cycle type based on market volatility, economic growth rate and interest rate level; Calculate the scenario adjustment factor based on the market sentiment index corresponding to the current market cycle type and the initial feature weight combination; The characteristic weight combination includes yield weight, volatility weight and flexibility weight; Optimize the initial feature weight combination based on the scenario adjustment factor to obtain the optimized feature weight combination; Calculate the comprehensive fund score corresponding to the fund product based on the optimized feature weight combination and the fund product's rate of return, volatility and flexibility score; Based on the comprehensive fund scores corresponding to the fund products, all fund products in the fund pool are clustered and divided into conservative fund sets and risky fund sets; Based on the investor's historical transaction records, the investor's risk preference is obtained, including conservative and risky types; the historical transaction records include transaction frequency, investment fund types, profit and loss records, holding time and asset allocation; Based on the risk preference of investors, the comprehensive fund scores of all fund products in the corresponding type of fund set are obtained, and the comprehensive fund scores are sorted from large to small to obtain the fund products corresponding to the top k as the recommendation results.

2. The fund product recommendation method according to claim 1, characterized in that: Using the Fama-French three-factor model, we can obtain the market factor exposure, scale factor exposure, and value factor exposure of fund products within a preset time period, expressed as: ； in, It indicates the expected rate of return of the fund product within the preset time period. It indicates the risk-free rate of return of the fund product within the preset time period. represents the yield benchmark, , and They represent market factor exposure, size factor exposure, and value factor exposure, respectively. represents the market factor, represents the scale factor, represents the value factor, Represents the preset model error term.

3. The fund product recommendation method according to claim 2, characterized in that: The flexibility score of a fund product is expressed as: ； in, Indicates the flexibility score of the fund product. Indicates the number of factor categories, the value is 3, ; Indicates The preset factor weights of the factors, Indicates Factor exposure to factors, Indicates The historical average factor exposure of the factor.

4. The fund product recommendation method according to claim 1, characterized in that: Based on the market volatility, economic growth rate and interest rate level, obtain the current market cycle type, including: When the economic growth rate is greater than 2% and the market volatility is less than 5%, the current market cycle type is determined to be an expansion period; When the economic growth rate is no more than 2% and the market volatility is no less than 5%, the current market cycle type is determined to be a contraction period; When the economic growth rate turns from negative to positive and the market volatility gradually decreases, the current market cycle type is determined to be a recovery period; When the economic growth rate is less than 0 and the market volatility is greater than 10%, the current market cycle type is determined to be a recession.

5. The fund product recommendation method according to claim 1, characterized in that: Based on the market sentiment index corresponding to the current market cycle type and combined with the initial feature weight, the scenario adjustment factor is calculated, expressed as: ； in, Indicates the market cycle type is The scenario adjustment factor is Indicates the market cycle type is The market sentiment index, ; Indicates The initial feature weights of the class features, Represents the total number of feature categories in the initial feature weight combination.

6. The fund product recommendation method according to claim 5, characterized in that: The initial feature weight combination is optimized based on the scenario adjustment factor to obtain the optimized feature weight combination, which is expressed as: ； in, Indicates the optimization of feature weight combination Optimized feature weights for class features.

7. The fund product recommendation method according to claim 6, characterized in that: Based on the optimized feature weight combination and the yield, volatility and flexibility scores of the fund product, the comprehensive fund score corresponding to the fund product is calculated, which is expressed as: ； in, Indicates the comprehensive fund score corresponding to the fund product. Indicates the optimization of feature weight combination Optimized feature weights of class features; Indicates Class features, .

8. The fund product recommendation method according to claim 1, characterized in that: If the fund product is a new fund product with no historical data, it includes: All fund products in the fund pool, their corresponding fund managers, investment themes, and market sentiment indexes are used as nodes, and the relationships between nodes are used as edges to construct a knowledge graph; Based on the fund managers and investment themes of the new fund products, in the knowledge graph, using cosine similarity, we search for multiple fund products with similarities higher than the preset similarity to the new fund products to form a similar fund set; The historical data of all fund products in the similar fund set are averaged as the historical data of the new fund product; The historical data includes yield and volatility.

9. The fund product recommendation method according to claim 1, characterized in that: If the investor is a new investor with no historical trading records, including: Based on the basic information of the new investor, among the investors with historical transaction records, multiple investors with similarity higher than a preset threshold are searched to form a similar user set; Calculate the average risk preference of all investors in the similar user set as the risk preference of the new investor; The basic information of the new investor includes gender, age and education level.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the fund product recommendation method according to any one of claims 1 to 9 are implemented.

Images