Financial product intelligent risk assessment and matching system

Abstract

The invention relates to the technical field of financial science and technology, in particular to an intelligent risk assessment and matching system for financial products. Comprising a multi-source data acquisition module which is used for acquiring financial product data, market data, macroeconomic data and investor personal data; the risk assessment model construction module is used for carrying out risk assessment on financial products based on the data acquired by the multi-source data acquisition module; the investor portrait construction module is used for constructing an investor portrait; and the intelligent matching module is used for matching the risk assessment result of the risk assessment model construction module with the investor portrait constructed by the investor portrait construction module, and recommending financial products meeting the risk tolerance and investment requirements of the investor to the investor. Through the multi-source data acquisition module, the risk assessment model construction module, the investor portrait construction module and the intelligent matching module, accurate risk assessment and personalized matching are realized, and the system has the functions of risk early warning, data updating and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of financial technology, and in particular to a financial wealth management product intelligent risk assessment and matching system. BACKGROUND

[0002] In today's financial market, there are various types of wealth management products, and the demand of investors also shows a diversification trend. The traditional financial wealth management product risk assessment and matching method has many shortcomings. On the one hand, the existing risk assessment relies on a simple index system and artificial experience judgment, which is difficult to comprehensively and accurately measure the complex risk characteristics of wealth management products. For example, the common risk assessment only considers the historical yield fluctuation of the wealth management product, ignoring the influence of multi-dimensional factors such as market environment changes, industry competition situation, and macroeconomic policy adjustments on risk. When the market appears sudden conditions such as major policy adjustments or global economic crisis, the risk assessment based on simple indicators cannot timely and accurately reflect the dramatic changes in the risk of wealth management products, resulting in insufficient understanding of potential risks by investors.

[0003] On the other hand, in the matching link of wealth management products and investors, the traditional method lacks precision and individualization. In most cases, only the simple risk preference classification (such as conservative, stable, and aggressive) of the investor is used to recommend products, without fully considering factors such as the dynamic changes of the detailed financial situation, investment goals, investment period, and risk tolerance of the investor. For example, for two investors who are both classified as stable, one is an elderly person who is close to retirement and whose financial reserves are mainly used for old-age support, and the other is a middle-aged person who is in the rising period of his career, has some idle funds, and has no major financial needs in the short term. They have significant differences in actual risk tolerance and investment needs, but the traditional matching method may recommend the same type of wealth management products for them, which cannot meet their individualized investment needs.

[0004] Therefore, there is an urgent need for a financial wealth management product intelligent risk assessment and matching system that can comprehensively assess risks in multiple dimensions and intelligently and individually match according to the detailed situation of investors, in order to solve the problems of incomplete risk assessment and inaccurate matching in the prior art. SUMMARY

[0005] In order to overcome the technical defects of incomplete risk assessment and inaccurate matching in the prior art, the purpose of the present application is to provide a financial wealth management product intelligent risk assessment and matching system. The present application realizes precise risk assessment and individualized matching by combining multiple algorithms and technologies through a multi-source data acquisition module, a risk assessment model construction module, an investor portrait construction module, and an intelligent matching module, and has functions such as risk warning and data updating.

[0006] The present application discloses a financial wealth management product intelligent risk assessment and matching system, comprising:

[0007] Multi-source data collection module: used for collecting financial wealth management product data, market data, macroeconomic data and investor personal data from databases within financial institutions, financial market data providers, macroeconomic data publishing platforms and investor terminals;

[0008] Risk assessment model construction module: based on the data collected by the multi-source data collection module, using machine learning algorithms and risk assessment models to assess the risk of financial wealth management products;

[0009] Investor portrait construction module: used to construct an investor portrait based on collected investor personal data, the investor portrait including investor risk tolerance characteristics;

[0010] Intelligent matching module: matching the risk assessment results of the risk assessment model construction module with the investor portrait constructed by the investor portrait construction module, and recommending financial wealth management products that meet the risk tolerance and investment needs of investors.

[0011] Preferably, the financial wealth management product data collected by the multi-source data collection module includes basic information of the wealth management product, investment portfolio composition, historical performance data, wherein the historical performance data at least includes annualized yield and volatility in the past three years.

[0012] Preferably, when the risk assessment model assesses the risk of financial wealth management products, it considers the multi-dimensional risks of market risk, credit risk, liquidity risk and operational risk.

[0013] When the risk assessment model construction module constructs the risk assessment model, it uses the analytic hierarchy process to determine the weight of each risk dimension, and the calculation formula is:

[0014]

[0015] Wherein, is the weight of the ith risk dimension, is the importance index of the ith risk dimension relative to the jth judgment criterion, m is the number of risk dimensions, and n is the number of judgment criteria.

[0016] Preferably, when the investor portrait construction module constructs the investor risk tolerance characteristics, it calculates the investor risk tolerance score based on the following calculation formula:

[0017]

[0018] Wherein, S is the investor risk tolerance score, I is the investor annual income, X is the total assets of the investor, E is the investment experience of the investor, and T is the investment period of the investor, , , 、 is a corresponding parameter, and .

[0019] Preferably, the intelligent matching module calculates the matching degree of the risk of the financial product and the risk tolerance of the investor by using a cosine similarity algorithm, and the calculation formula is:

[0020]

[0021] wherein, is the matching degree of the risk vector A of the financial product and the risk tolerance B of the investor, is the score of the financial product in the i-th risk dimension, is the tolerance score of the investor in the i-th risk dimension, and n is the number of risk dimensions.

[0022] Preferably, it further comprises a risk warning module, which monitors the risk indicators of the financial products in real time, and sends warning information to the investor and / or the financial institution when the risk indicators exceed the preset threshold.

[0023] Preferably, the risk warning module provides risk adjustment suggestions when sending the warning information, and the risk adjustment suggestions include adjusting the portfolio proportion and replacing the investment product type.

[0024] Preferably, it further comprises a data updating module, which is used to update the financial product data, market data, macroeconomic data and investor personal data collected by the multi-source data collection module periodically, and the updating period is less than 1 week.

[0025] Preferably, the system is provided with a user feedback interface, and the investor feeds back the recommended financial product through the feedback interface, and the system optimizes the risk assessment model and / or the intelligent matching module according to the feedback information.

[0026] After adopting the above technical scheme, compared with the prior art, the following beneficial effects are obtained:

[0027] 1. The prior art relies on single data and simple indicators, and the present scheme integrates multi-dimensional information such as financial product data, market data, macroeconomic data, etc. through the multi-source data collection module, to provide a more comprehensive data basis for risk assessment, and to realize more comprehensive assessment of the risk of the financial product;

[0028] 2. The prior art relies on artificial experience, and the present scheme uses machine learning algorithms through the risk assessment model construction module, and combines the analytic hierarchy process to determine the weight of each risk dimension, to comprehensively consider market, credit and other types of risks, so that the risk assessment result is more accurate;

[0029] 3. The prior art only classifies investors simply by risk preference, and the present scheme collects detailed data such as financial status and investment goals of investors through an investor portrait construction module, and calculates a risk bearing capacity score by formula, so that the constructed investor portrait is more accurate;

[0030] 4. The prior art is a rough match, and the present scheme matches the comprehensive risk assessment result with the accurate investor portrait by using a cosine similarity algorithm through an intelligent matching module, so that the product recommendation is more in line with the individualized needs of investors;

[0031] 5. The prior art risk warning is lagging or has no suggestion, and the present scheme monitors risk indicators in real time through a risk warning module, sends a warning and provides adjustment suggestions when the threshold is exceeded, helping investors and institutions to respond to risks in a timely manner;

[0032] 6. The prior art data is not updated in time, and the present scheme updates data every week through a data updating module to ensure that the system evaluates and matches based on the latest data, adapting to market dynamic changes;

[0033] 7. The prior art is difficult to optimize according to feedback, and the present scheme collects investor opinions through a user feedback interface for optimizing the risk assessment model and matching strategy, and improving the system performance. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 is a flowchart of the financial wealth management product intelligent risk assessment and matching system. DETAILED DESCRIPTION

[0035] The advantages of the present application will be further described below in combination with the drawings and specific embodiments.

[0036] Hereinafter, exemplary embodiments will be described in detail with reference to the accompanying drawings. In the following description, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments are not representative of all embodiments consistent with the present disclosure. Rather, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0037] The terms used in the present disclosure are merely for the purpose of describing specific embodiments, and are not intended to limit the present disclosure. The singular forms "a", "an" and "the" used in the present disclosure and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein means and includes any or all possible combinations of one or more associated listed items.

[0038] It should be understood that, although the terms first, second, third, etc. can be employed in this disclosure to describe various information, the information is not to be limited to these terms. These terms are only used to distinguish one category of information from another. For example, without departing from the scope of the present disclosure, first information could also be referred to as second information, and, similarly, second information can also be referred to as first information. Depending on the context, the word "if' as used herein can be interpreted to mean "when" or "in response to determining".

[0039] In the description of the present application, it should be understood that the orientation or positional relationship indicated by the terms "longitudinal", "transverse", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0040] In the description of the present application, unless otherwise specified and limited, it should be noted that the terms "mounting", "connection", "connection" should be understood broadly, for example, it can be mechanical connection or electrical connection, it can be the communication between two elements, it can be direct connection or indirect connection through intermediate medium, and the specific meaning of the above terms can be understood by those skilled in the art according to the specific circumstances.

[0041] In the subsequent description, the suffix such as "module", "component" or "unit" used to represent elements is only for the convenience of the description of the present application, and has no specific meaning. Therefore, "module" and "component" can be used interchangeably.

[0042] Reference Figure 1 In the present embodiment, a financial wealth management product intelligent risk assessment and matching system will be described in detail by way of example 1 and example 2.

[0043] Example 1

[0044] Scenario description: investor Zhang, 35 years old, a manager of an internet company, annual income 800,000 yuan, total assets 5,000,000 yuan (including a house worth 3,000,000 yuan, no mortgage), 5 years of stock investment experience, plans to reserve funds for children's overseas study in 10 years, investment period 10 years, risk tolerance medium. The scheme needs to match suitable financial products. DETAILED DESCRIPTION

[0045] The multi-source data acquisition module obtains data based on the following channels:

[0046] Database within the financial institution: Collect data of 3 candidate wealth management products, including Product A (hybrid fund, investment portfolio: stocks 60% + bonds 30% + cash 10%, past three years annual yield: 8%, 10%, 9% respectively, volatility: 15%); Product B (bond fund, investment portfolio: bonds 90% + cash 10%, past three years annual yield: 4%, 5%, 4.5% respectively, volatility: 3%); Product C (stock fund, investment portfolio: stocks 90% + cash 10%, past three years annual yield: 12%, 15%, -2% respectively, volatility: 25%).

[0047] Financial market data provider: Obtain current Shanghai-Shenzhen 300 Index volatility 20%, 10-year Treasury yield 3.2%.

[0048] Macro data release platform: Obtain current GDP growth rate 5.5%, inflation rate 2.1%.

[0049] Investor terminal: Collect data of Zhang's age, income, assets, investment experience, investment goals, and time limit.

[0050] Risk assessment model construction:

[0051] The risk assessment model construction module uses the analytic hierarchy process to determine the weights of the risk dimensions (market risk, credit risk, liquidity risk, and operational risk), sets the judgment criteria as "impact on yield", "occurrence probability", and "impact range" (n=3), and expert scoring obtains the importance scale matrix as shown in Table 1:

[0052] Table 1

[0053] Risk dimensions Impact on earnings Probability of occurrence Scope of impact Market risk 3 2 3 Credit risk 2 3 2 Liquidity risk 1 1 1 Operational risk 1 1 1

[0054] According to the calculation formula Calculate the weight:

[0055] Numerator: Market risk =3+2+3=8; Credit risk =2+3+2=7; Liquidity risk =1+1+1=3; Operational risk =1+1+1=3.

[0056] Denominator: Total sum 8+7+3+3=21.

[0057] Weight: Market risk =8 / 21≈0.38; Credit risk =7 / 21≈0.33; Liquidity risk =3 / 21≈0.14; Operational risk =3 / 21≈0.15.

[0058] Risk score of 3 products by random forest algorithm (0~10, the higher the score, the higher the risk):

[0059] Product A: market risk 6 points, credit risk 4 points, liquidity risk 3 points, operational risk 2 points, comprehensive risk 6x0.38+4x0.33+3x0.14+2x0.15=4.44 points.

[0060] Product B: market risk 2 points, credit risk 3 points, liquidity risk 2 points, operational risk 2 points, comprehensive risk 2x0.38+3x0.33+2x0.14+2x0.15=2.35 points.

[0061] Product C: market risk 9 points, credit risk 5 points, liquidity risk 4 points, operational risk 2 points, comprehensive risk 9x0.38+5x0.33+4x0.14+2x0.15=6.49 points.

[0062] Investor portrait construction:

[0063] According to the calculation formula , wherein, =0.2, =0.3, =0.2, =0.3; Substituting the data:

[0064] Annual income I=80 million (standardized as 0.8), total assets X=500 million (standardized as 0.6), investment experience E=5 years (standardized as 0.5), investment period T=10 years (standardized as 0.8). Then according to the calculation formula, S=0.2x0.8+0.3x0.6+0.2x0.5+0.3x0.8=0.16+0.18+0.1+0.24=0.68. Then 0.68 is medium to high.

[0065] Investor portrait: risk tolerance is medium to high, investment goal is long-term (10 years) education reserve, and prefers medium-risk, medium-yield products.

[0066] Intelligent matching:

[0067] Cosine similarity algorithm is adopted , the risk vector of financial products takes the comprehensive risk and the score of each dimension, and the risk tolerance vector of investors takes the corresponding dimension of tolerance (according to the score 0.68, it is mapped to market risk 0., 0.84, credit risk 0.6, liquidity risk 0.6, and operational risk 0.5).

[0068] Matching degree of product A and Zhang: = 0.6 * 0.7 + 0.4 * 0.6 + 0.3 * 0.6 + 0.2 * 0.5 = 0.42 + 0.24 + 0.18 + 0.1 = 0.94; ; ; then (high degree of matching).

[0069] According to the above content, the matching degrees of product B and product C with Zhang are calculated respectively, the matching degree of product B is 0.72, and the matching degree of product C is 0.85. Then the system finally recommends product A, and simultaneously attaches risk description and expected return analysis.

[0070] User feedback and system optimization:

[0071] Zhang uses the product for 3 months, and then feeds back through the user feedback interface set by the system, and thinks that the risk assessment of product A is slightly lower than the actual feeling, and the recommended product does not fully meet the expectation in terms of return stability. After the system collects the feedback, it is used as the basis for optimization. For the risk assessment model, the weight parameter of market risk in the model is adjusted again, and the factor of short-term market fluctuation is considered; for the intelligent matching module, the weight of the return stability index in the cosine similarity algorithm is increased, and after optimization, when the system recommends products for Zhang and similar investors in the future, the risk assessment is more in line with the actual situation, and the product matching is more in line with the demand of investors for return stability.

[0072] It should be noted that in the risk assessment model construction module, the quantization standard of the analytic hierarchy process judgment criteria: "the degree of influence on return", "occurrence probability" and "influence range" are all 1-9 scale method, 1 represents equal importance, 3 represents slightly important, 5 represents obviously important, 7 represents strongly important, 9 represents extremely important, and 2, 4, 6 and 8 are intermediate values. In this embodiment 1, the values of 3 and 2 in the expert scoring matrix are based on this standard, for example, the market risk has a degree of influence on return of 3, which means it is slightly more important than return stability. At the same time, the expert scoring process needs to go through 3 rounds of back-to-back scoring, and when the standard deviation of the scoring result is less than 1, the average value is taken as the final result to ensure the objectivity of the scoring.

[0073] It should be noted that, , , , The values of the above are determined by historical data regression analysis. For example, the behavior risk tolerance of 1000 investors in the past 5 years is fitted to obtain the above = 0.2, = 0.3, = 0.2, =0.3. Specifically, the portfolio adjustment of these investors under different market conditions, loss tolerance, and other actual risk performance data are collected as dependent variables, and four indicators such as annual income are used as independent variables to perform multiple linear regression. The significant influencing factors are screened out through F test and t test. The finally determined parameters pass the significance test within a 95% confidence interval, ensuring that the deviation between the calculation results and the actual risk bearing capacity is less than 5%.

[0074] It should be noted that in the intelligent matching module, the risk dimension score adopts a 0-10 point system, wherein 0-3 is low risk, 4-6 is medium risk, and 7-10 points is high risk. The score is determined by quantitatively assigning the influence degree of the historical risk events of the financial product, for example, a product has experienced a slight loss event once, deducting 1 point for its market risk; has experienced a more serious default event once, deducting 3 points for its credit risk, and so on. The risk bearing capacity score of the investor 0-1 corresponds to the risk dimension bearing capacity mapping rule: market risk bearing capacity = 0.5 + 0.5 x S, credit risk bearing capacity = 0.4 + 0.6 x S, liquidity risk bearing capacity = 0.5 + 0.5 x S, and operational risk bearing capacity = 0.3 + 0.7 x S. In this embodiment 1, S = 0.68, therefore market risk bearing capacity = 0.5 + 0.5 x 0.68 = 0.84.

[0075] It should be noted that the information of the user feedback interface mobile phone includes satisfaction (1-5 points) for the recommended product, risk assessment accuracy evaluation (“high”, “low”, “moderate”), and improvement suggestions for the matching result. These feedback information will be quantitatively processed, for example, 4-5 points for satisfaction is positive feedback, and 1-2 points is negative feedback. For the risk assessment model, when the proportion of negative feedback of a certain type of product exceeds 30%, the weight of the risk dimension in the risk assessment model of this type of product is adjusted; for the intelligent matching module, the weight coefficient of each risk dimension in the cosine similarity algorithm is adjusted according to the user's suggestions for the matching result, such as the user repeatedly feedbacks that he pays more attention to liquidity risk, then the weight of liquidity risk in the matching calculation is increased. For example, in this embodiment 1, after Zhang feedbacks that the risk assessment of product A is slightly low, the system adjusts the weight of market risk in the risk assessment model and calculates the risk score of the product.

[0076] Embodiment 2

[0077] Scenario description: a small technology enterprise (hereinafter referred to as “enterprise A”), with a registered capital of 5 million yuan and an annual net profit of 1 million yuan, plans to invest 2 million yuan in short-term (1 year) investment, and needs to evaluate and match a bank financial product with conservative risk preference. DETAILED DESCRIPTION

[0078] Multi-source data collection module obtains data based on the following channels:

[0079] Database within the financial institution: Collect bank wealth management product D (structured deposit, linked to national debt, expected annual yield of 3.5%~4.5%), investment period of 1 year, liquidity rating of "T+1 redemption".

[0080] Financial market data provider: National debt yield curve, interbank lending rate.

[0081] Macro data release platform: The current monetary policy is prudent, and the inflation rate is 2.1%.

[0082] Investor terminal: A enterprise financial statements, investment decision documents (clear risk conservative, period of 1 year).

[0083] Risk assessment model construction:

[0084] Risk dimension weight (analytic hierarchy process calculation): Market risk 0.2, credit risk 0.4 (mainly bank credit), liquidity risk 0.3, operational risk 0.1.

[0085] Product D risk score: Market risk 2 points, credit risk 1 point (bank credit rating AAA), liquidity risk 1 point, operational risk 1 point, comprehensive risk 2×0.2+1×0.4+1×0.3+1×0.1=0.4+0.4+0.3+0.1=1.2 points (low risk).

[0086] Investor portrait construction:

[0087] Enterprise risk tolerance score, according to the calculation formula , wherein =0.3, =0.3, =0.2, =0.2. Net profit I=100 million yuan (standardized 0.5), total assets X=500 million yuan (standardized 0.5), investment experience E=3 years (standardized 0.3), period T=1 year (standardized 0.2), then according to the above calculation formula, the enterprise risk tolerance score is: 0.3×0.5+0.3×0.5+0.2×0.3+0.2×0.2=0.40 (conservative type).

[0088] Intelligent matching:

[0089] Matching degree calculation: The cosine similarity of product D risk vector and enterprise bearing vector is 0.91 (high matching), and the system recommends product D.

[0090] Risk warning: The risk indicator of product D is updated by the data updating module every week. In the third month, the yield fluctuation of the linked government bond exceeds the preset threshold (±0.5%), and the risk warning module sends a warning message, prompting "the yield fluctuation of the government bond may cause the actual yield to be lower than expected", and suggests "if the risk tolerance has not changed, you can continue to hold; if you need to preserve the principal, you can replace it with a fixed-income deposit".

[0091] The enterprise adjusts according to the warning and chooses to continue to hold, and the final yield at maturity is 4.2%, which meets the expectation.

[0092] After product D expires, enterprise A feedbacks through the user feedback interface that the overall product D recommended by the system meets the needs, but the timeliness of the risk warning can be further improved, and hopes to receive a prompt when the risk indicator approaches the threshold. After receiving this feedback, the system optimizes the parameters of the risk warning module, adjusting the trigger condition of the risk warning from the risk indicator exceeding the preset threshold to sending a prompt when the risk indicator reaches 90% of the preset threshold. At the same time, in the intelligent matching module, in view of the high demand of enterprise investors for the timeliness of the warning, the prediction weight of the risk change trend of the product is increased. The optimized system is more timely in risk warning and more in line with the actual needs of enterprises when serving other enterprise investors.

[0093] It should be noted that when the risk assessment model construction module uses the random forest algorithm, the number of decision trees is set to 50, the maximum depth is set to 10 layers, and the minimum leaf node sample size is set to 5. These parameters are determined by grid search method, and the risk assessment accuracy is used as the index to optimize in the range of 20-100 decision trees and 5-15 maximum depth. The selected parameter combination makes the model accuracy the highest. The training data selects the weekly data of 200 bank wealth management products in the past 3 years, covering products of different types (such as structured deposits, fixed-income products, etc.) and different issuers, and the input features include 15 indicators such as government bond yield, interbank borrowing rate, and issuer credit rating. The model is optimized through 5-fold cross-validation, i.e. the data set is divided into 5 parts, 4 parts are used as the training set and 1 part is used as the validation set, and the average accuracy rate is taken after repeating 5 times, so that the risk assessment accuracy reaches 89%.

[0094] It should be noted that the basis for the threshold of ±0.5% of the yield volatility of Chinese bonds in this embodiment 2 is that the standard deviation of the daily yield of the bond linked to the product in the past year is 0.25%, and 2 times the standard deviation (0.5%) is taken as the early warning threshold, which meets the abnormal value judgment standard of 95% confidence interval in statistics, that is, the probability of yield volatility exceeding the range is only 5% under normal circumstances. At the same time, it is clear that the threshold adjustment method for different types of products, such as the threshold of ±2% for stock-type products, is based on 2 times the historical volatility rate of 1%; the threshold of mixed-type products is calculated by weighting the investment proportion of stocks and bonds, such as stock proportion = 60% x 2% + 40% x 0.5% = 1.4%. When major unexpected events occur in the market (such as significant adjustment of macroeconomic policy), the system will automatically temporarily lower the threshold by 20% to improve the early warning sensitivity.

[0095] It should be noted that in the data labeling of the investor portrait construction, min-max normalization is used, and the formula is: standardized value = (actual value - minimum value) / (maximum value - minimum value). For the net profit indicator of corporate investors, the minimum value is 0 and the maximum value is 200,000 yuan, which covers the net profit interval of similar size (registered capital of about 500,000 yuan) technology enterprises in the past 3 years, and in embodiment 2, 100,000 yuan standardized value = (100-0) / (200-0) = 0.5; the minimum value of total assets is 0 and the maximum value is 1,000,000 yuan, and the 500,000 yuan standardized value = (500-0) / (1000-0) = 0.5. For individual investors, the minimum value of annual income is 0 and the maximum value is 500,000 yuan, the investment experience is 0 years as the minimum value and 20 years as the maximum value, and the investment period is 1 month as the minimum value and 30 years as the maximum value, all of which are standardized according to the above formula to ensure the comparability of data of different types of investors.

[0096] It should be noted that for bank wealth management product D, a structural deposit, in the construction of the risk assessment model, in addition to considering market risk, credit risk, etc. through the risk dimension, a "derivative-linked risk" dimension can also be added, with a weight of 0.15. This weight is determined by analyzing the risk frequency and impact degree of structural products caused by abnormal fluctuations of the linked target (such as government bonds) in history. In calculating the score of this risk dimension, based on the volatility of the linked target (such as government bonds) and the correlation between the product yield structure and the target, a scoring card form (0-10 points) is used for evaluation. In this embodiment 2, product D has a low volatility of the linked government bonds, so the score of this dimension is 2, and the final comprehensive risk score = original comprehensive risk score x (1-0.15) + 2 x 0.15 = 1.2 x 0.85 + 0.3 = 1.32 points.

[0097] It should be noted that the embodiments of the present application have better implementation, and do not limit the present application in any form, any skilled person in the art can change or modify the above disclosed technical content into equivalent effective embodiments, as long as it does not deviate from the content of the technical scheme of the present application, any modification or equivalent change and modification of the above embodiments according to the technical essence of the present application, still belongs to the scope of the technical scheme of the present application.

Claims

1. An intelligent risk assessment and matching system for financial products, characterized by: include: Multi-source data collection module: used to collect financial product data, market data, macroeconomic data and investor personal data from the internal databases of financial institutions, financial market data providers, macroeconomic data publishing platforms and investor terminals; Risk assessment model building module: Based on the data collected by the multi-source data acquisition module, the module uses machine learning algorithms and risk assessment models to conduct risk assessment on the financial product; Investor profile building module: used to build an investor profile based on the collected investor personal data, the investor profile including the investor's risk tolerance characteristics; Intelligent matching module: matches the risk assessment results of the risk assessment model construction module with the investor portrait constructed by the investor portrait construction module, and recommends financial management products that meet the investor's risk tolerance and investment needs.

2. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: The financial product data collected by the multi-source data collection module includes basic information of the financial product, investment portfolio composition, and historical performance data, wherein the historical performance data at least includes the annualized rate of return and volatility over the past three years.

3. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: When performing risk assessment on the financial management product, the risk assessment model comprehensively considers the multi-dimensional risks of market risk, credit risk, liquidity risk, and operational risk. When constructing the risk assessment model, the risk assessment model construction module uses the hierarchical analysis method to determine the weight of each risk dimension, and the calculation formula is: in, is the weight of the i-th risk dimension, is the importance index of the i-th risk dimension relative to the j-th judgment criterion, m is the number of risk dimensions, and n is the number of judgment criteria.

4. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: When the investor profile building module is building the investor risk tolerance characteristics, the investor risk tolerance score is calculated based on the following calculation formula: Among them, S is the investor's risk tolerance score, I is the investor's annual income, X is the investor's total assets, E is the investor's investment experience, and T is the investor's investment period. 、 、 、 are the corresponding parameters, and .

5. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: The intelligent matching module uses the cosine similarity algorithm to calculate the matching degree between the risk of the financial product and the investor's risk tolerance. The calculation formula is: in, is the matching degree between the risk vector A of the financial product and the investor’s risk tolerance B, is the score of the financial product on the i-th risk dimension, is the investor's tolerance score on the i-th risk dimension, and n is the number of risk dimensions.

6. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: It also includes a risk warning module, which monitors the risk indicators of the financial management products in real time and sends warning information to the investors and / or financial institutions when the risk indicators exceed a preset threshold.

7. The intelligent risk assessment and matching system for financial products according to claim 6, characterized in that: When sending warning information, the risk warning module also provides risk adjustment suggestions, which include adjusting the investment portfolio ratio and changing the investment product type.

8. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: It also includes a data update module, which is used to regularly update the financial product data, market data, macroeconomic data and investor personal data collected by the multi-source data collection module, with an update cycle of less than 1 week.

9. The intelligent risk assessment and matching system for financial products according to claim 1, characterized in that: The system is provided with a user feedback interface, through which the investor provides feedback on the recommended financial products, and the system optimizes the risk assessment model and / or the intelligent matching module based on the feedback information.

Images