Online loan portfolio recommendation method and system, terminal and medium
By constructing a list of loan records and combining it with risk and return prediction models, along with modern portfolio theory, this approach solves the problem that traditional models in online lending platforms cannot balance risk and return, and provides a scientific and reasonable portfolio recommendation method.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-11
- Publication Date
- 2026-03-31
AI Technical Summary
Existing credit scoring and profit scoring models of online lending platforms cannot simultaneously take into account both risk and return, and cannot provide multi-dimensional investment advice and optimized fund allocation strategies.
A list of loan records is constructed, and risk and return prediction models are established based on the cleared loan records. An optimization model is then constructed by combining modern portfolio theory to make portfolio recommendations.
It achieves the goal of maximizing returns while controlling risks, and minimizing risks while maintaining a certain level of returns, providing investors with more accurate and effective investment strategy advice.
Smart Images
Figure CN121767084A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the financial field, and more particularly to an online lending portfolio recommendation method, system, terminal, and medium. Background Technology
[0002] In the field of internet finance, online lending is a new market finance model that directly connects investors and borrowers through online platforms. This model mainly involves two types of participants: borrowers and investors. Borrowers submit detailed personal information, which is reviewed by the online lending platform, and then they can publish loan applications on the platform; investors decide whether to invest and how much to invest based on the information displayed. Investors on online lending platforms pursue two main goals: reducing portfolio risk and increasing returns. To this end, scholars have conducted extensive research, including exploring factors affecting the probability of borrower default, constructing credit scoring models to predict default probability, and developing profit scoring models to predict returns. However, existing methods have some limitations: traditional credit scoring models only focus on the probability of default, ignoring the high returns that may result from a high probability of default; traditional profit scoring models only focus on returns, without fully considering the high risks associated with high returns. These two methods can only provide single-dimensional investment advice, that is, recommending loans with the lowest expected probability of default or the highest expected return, but they cannot provide investors with guidance on how to rationally allocate funds to achieve an optimal investment strategy. In addition, although some studies have attempted to take a portfolio perspective, most of them focus on the goal of maximizing returns, failing to consider the two important aspects of maximizing returns and minimizing risks simultaneously.
[0003] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention
[0004] The technical problem to be solved by this invention is to provide an online lending portfolio recommendation method, system, terminal and medium that addresses the above-mentioned deficiencies of the prior art. It aims to solve the problem that traditional credit scoring and profit scoring models in online lending platforms cannot simultaneously take into account risk and return, provide multi-dimensional investment advice and optimize fund allocation strategies.
[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides an online lending portfolio recommendation method, wherein the method includes: Using the database of an online lending platform, a list of loan records is constructed, which includes settled loan records and investable loan records; Based on the settled loan records, obtain the first revenue data and the first risk data of the settled loan records, and establish a risk prediction model and a revenue prediction model based on the first risk data and the information of the settled loan records. Based on the risk prediction model and the return prediction model, calculate the second risk data and the second return data of the investable loan record, and construct an optimization model based on the second risk data, the second return data and modern portfolio theory. Based on the optimization model, portfolio recommendations are made.
[0006] In one implementation, constructing a list of loan records using the database of an online lending platform includes: Using the database of an online lending platform, collect all loan records of the online lending platform since its establishment, and record the collection time; Based on the loan records and the collection time, obtain the records of settled loans and the records of investable loans; A list of loan records is constructed based on the settled loan records and the investable loan records.
[0007] In one implementation, obtaining the settled loan records and investable loan records based on the loan records and the collection time includes: Based on the loan records, obtain each loan in the loan records; If the collection time is greater than the sum of the bidding end date and repayment time of the loan, then the loan is a settled loan, and a settled loan record is obtained based on the settled loan. If the data collection time falls within the active bidding phase of the loan, then the loan is an investable loan, and based on the investable loan, investable loan records are obtained. The active phase refers to the time period from when the loan begins to accept investment until when the loan stops accepting investment.
[0008] In one implementation, obtaining the first revenue data and first risk data of the settled loan record based on the settled loan record includes: Based on the settled loan records, obtain each loan in the settled loan records and the corresponding loan information for each loan; Based on the loan information, calculate the revenue data for each loan, and based on the revenue data for each loan, obtain the first revenue data for the settled loan records. Based on the first revenue data, the credit rating of each loan, and the number of loans with the same credit rating as each credit rating in the settled loan records, risk data of the loans belonging to each credit rating in the settled loan records is obtained, and based on the risk data of the loans of each credit rating, the first risk data of the settled loan records is obtained.
[0009] In one implementation, establishing a risk prediction model and a return prediction model based on the first risk data and the information on the settled loan records includes: Based on the first risk data, a risk prediction model is established; Based on the information of the settled loan records, obtain the relevant attributes of the loan and the borrower of the loan, and establish a revenue prediction model based on the relevant attributes.
[0010] In one implementation, obtaining the second risk data and second return data of the investable loan record based on the risk prediction model and the return prediction model includes: Based on the risk prediction model and the credit rating of each loan in the investable loan record, the risk data of each loan in the investable loan record is calculated, and based on the risk data of each loan in the investable loan record, the second risk data of the investable loan record is obtained. Based on the revenue prediction model, each loan in the investable loan record, and the relevant attributes of the borrower of each loan, the revenue data of each loan in the investable loan record is calculated, and based on the revenue data of each loan in the investable loan record, the second revenue data of the investable loan record is obtained.
[0011] In one implementation, the optimization model comprises a first model and a second model. The first model is used to maximize the second return data given a certain second risk data, and the second model is used to minimize the second risk data given a certain second return data.
[0012] Secondly, embodiments of the present invention also provide an online lending portfolio recommendation system, wherein the system includes: A loan record list construction module is used to construct a loan record list using the database of an online lending platform. The loan record list includes settled loan records and investable loan records. The prediction model building module is used to obtain the first income data and the first risk data of the settled loan records based on the settled loan records, and to build a risk prediction model and an income prediction model based on the first risk data and the information of the settled loan records. The optimization model building module is used to calculate the second risk data and the second return data of the investable loan record according to the risk prediction model and the return prediction model, and to build an optimization model based on the second risk data, the second return data and modern portfolio theory. The portfolio recommendation module is used to recommend portfolios based on the optimization model.
[0013] Thirdly, embodiments of the present invention also provide a terminal, wherein the terminal includes a memory, a processor, and an online lending portfolio recommendation program stored in the memory and capable of running on the processor. When the processor executes the online lending portfolio recommendation program, it implements the steps of the online lending portfolio recommendation method of any of the above solutions.
[0014] Fourthly, embodiments of the present invention also provide a computer-readable storage medium, wherein an online lending portfolio recommendation program is stored on the computer-readable storage medium, and when the online lending portfolio recommendation program is executed by a processor, it implements the steps of the online lending portfolio recommendation method described in any of the above schemes.
[0015] Beneficial Effects: This invention provides an online lending portfolio recommendation method. Compared with existing technologies, this invention first utilizes the database of an online lending platform to construct a loan record list. This list includes settled loan records and investable loan records. Settled loan records are those whose repayment period has ended, while investable loan records are those still eligible for investment. This step lays a data foundation for subsequent analysis, enabling model construction and optimization based on real historical data, thus enhancing the accuracy and practicality of the model. Next, based on the settled loan records, first return data and first risk data are obtained. Based on the first risk data and the information from the settled loan records, a risk prediction model and a return prediction model are established. This step facilitates a deeper understanding of... This invention addresses the impact of various factors on loan returns and risks, thereby improving the accuracy of risk and return predictions for investable loans. Then, based on the risk and return prediction models, it calculates second risk and second return data for the investable loan records. Based on these data and modern portfolio theory, an optimization model is constructed. This step considers both maximizing returns and minimizing risks, overcoming the shortcomings of traditional credit scoring and profit scoring models that focus on only a single objective (i.e., minimizing default probability or maximizing return). Finally, based on the optimization model, portfolio recommendations are made. This process accurately guides investors on how to rationally allocate their funds to achieve optimal investment goals, improving the scientific nature of investment decisions and effectively balancing the relationship between returns and risks. This invention provides a more scientific and rational portfolio recommendation method by comprehensively integrating historical loan data, constructing and applying risk and return prediction models, and introducing modern portfolio theory. This method not only overcomes the limitations of existing technologies that rely solely on a single indicator to evaluate loan projects but also achieves the goals of maximizing returns while controlling risks and minimizing risks while maintaining a certain level of returns, providing investors with more accurate and effective investment strategy suggestions. Attached Figure Description
[0016] Figure 1 A flowchart illustrating a specific implementation of the online lending portfolio recommendation method provided in this embodiment of the invention.
[0017] Figure 2 A flowchart of a preferred embodiment of the online lending portfolio recommendation method provided in this invention.
[0018] Figure 3 The diagram shows the optimal investment loan portfolio result of the online lending portfolio recommendation method provided in this embodiment of the invention.
[0019] Figure 4This is a schematic diagram of the online lending portfolio recommendation system provided in an embodiment of the present invention.
[0020] Figure 5 This is a block diagram illustrating the internal structure of a terminal provided in an embodiment of the present invention. Detailed Implementation
[0021] To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0022] In the field of internet finance, online lending is a new market finance model that directly connects investors and borrowers through online platforms. This model mainly involves two types of participants: borrowers and investors. Borrowers submit detailed personal information, which is reviewed by the online lending platform, and then they can publish loan applications on the platform; investors decide whether to invest and how much to invest based on the information displayed. Investors on online lending platforms pursue two main goals: reducing portfolio risk and increasing returns. To this end, scholars have conducted extensive research, including exploring factors affecting the probability of borrower default, constructing credit scoring models to predict default probability, and developing profit scoring models to predict returns. However, existing methods have some limitations: traditional credit scoring models only focus on the probability of default, ignoring the high returns that may result from a high probability of default; traditional profit scoring models only focus on returns, without fully considering the high risks associated with high returns. These two methods can only provide single-dimensional investment advice, that is, recommending loans with the lowest expected probability of default or the highest expected return, but they cannot provide investors with guidance on how to rationally allocate funds to achieve an optimal investment strategy. In addition, although some studies have attempted to take a portfolio perspective, most of them focus on the goal of maximizing returns, failing to consider the two important aspects of maximizing returns and minimizing risks simultaneously.
[0023] To address the aforementioned issues, this embodiment provides an online lending portfolio recommendation method. Specifically, this embodiment first utilizes the database of an online lending platform to construct a loan record list. This list includes settled loan records and investable loan records. Settled loan records are those whose repayment period has ended, while investable loan records are those still eligible for investment. This step lays the data foundation for subsequent analysis, enabling model construction and optimization based on real historical data, thus enhancing the model's accuracy and practicality. Next, based on the settled loan records, first return data and first risk data are obtained. Based on the first risk data and the information from the settled loan records, a risk prediction model and a return prediction model are established. This step facilitates in-depth analysis. This invention aims to improve the accuracy of risk and return predictions for investable loans by comprehensively integrating historical loan data, constructing and applying risk and return prediction models, and introducing modern portfolio theory. It then calculates second risk and second return data for the investable loan records based on the risk and return prediction models, and constructs an optimization model based on these data and modern portfolio theory. This step considers both maximizing returns and minimizing risks, overcoming the shortcomings of traditional credit scoring and profit scoring models that focus on only a single objective (i.e., minimizing default probability or maximizing return). Finally, based on the optimization model, portfolio recommendations are made. This process accurately guides investors on how to rationally allocate their funds to achieve optimal investment goals, improving the scientific nature of investment decisions and effectively balancing the relationship between returns and risks. This invention provides a more scientific and reasonable portfolio recommendation method by comprehensively integrating historical loan data, constructing and applying risk and return prediction models, and introducing modern portfolio theory. This method not only overcomes the limitations of existing technologies that rely on a single indicator to evaluate loan projects, but also achieves the goals of maximizing returns while controlling risks and minimizing risks while maintaining certain returns, providing investors with more accurate and effective investment strategy suggestions.
[0024] For example, suppose an investor wants to invest through an online lending platform and hopes to build an optimized portfolio to maximize returns while controlling risk, or minimize risk while maintaining a certain return. In this case, the investor can use the method proposed in this invention to make investment decisions. First, a list of loan records is retrieved from the online lending platform's database. This list includes settled loan records and investable loan records. This step lays the data foundation for subsequent analysis, enabling model building and optimization based on real historical data, enhancing the model's accuracy and practicality. Next, based on the settled loan records, the first return data and the first risk data are calculated, and risk prediction models and return prediction models are established. For example, by analyzing various factors in the settled loan records (such as the borrower's credit score, loan term, etc.), it is determined which factors have a significant impact on loan returns and risks. This step helps to deeply understand the impact of different factors on loan returns and risks, thereby improving the accuracy of risk and return predictions for future investable loans. Then, based on the previously established risk and return prediction models, second-level risk and return data for investable loan records are calculated. Using modern portfolio theory, and combining this second-level risk and return data, an optimization model is constructed. This process considers not only maximizing the overall return of the portfolio but also minimizing overall risk. Finally, based on the constructed optimization model, specific portfolio recommendations are made. This process accurately guides investors on how to rationally allocate their funds to achieve optimal investment goals, improving the scientific nature of investment decisions and effectively balancing the relationship between return and risk. The entire method, by comprehensively integrating historical loan data, constructing and applying risk and return prediction models, and introducing modern portfolio theory, provides a more scientific and rational portfolio recommendation method. This method overcomes the limitations of existing technologies that rely solely on a single indicator to evaluate loan projects, achieving the goals of maximizing returns while controlling risk and minimizing risk while maintaining a certain return, providing investors with more accurate and effective investment strategy suggestions.
[0025] It should be noted that the online lending portfolio recommendation method, system, terminal, and medium proposed in this invention are specifically designed for application scenarios of legal and compliant online lending platforms. The core of this invention lies in providing investors with optimized portfolio recommendations through scientific algorithms and data analysis techniques, based on the premise that the lending platform complies with legal and regulatory requirements. This lending platform must comply with the laws and regulations of its country or region, fulfill its information disclosure obligations as required, protect the rights and interests of both lenders and borrowers, and strictly adhere to relevant financial regulatory policies, including but not limited to holding necessary financial licenses, complying with anti-money laundering regulations, implementing user identity authentication, and ensuring the transparency and traceability of fund flows. Therefore, the scope of application of this invention is explicitly limited to lending platforms that have passed legal review and obtained regulatory approval, thereby ensuring that all recommended portfolios are generated and executed within a legal framework. This design not only protects the legitimate rights and interests of investors but also maintains the healthy and stable operation of the financial market.
[0026] This embodiment provides an online lending portfolio recommendation method, which can be applied to smart terminals, such as... Figure 1 As shown, the specific steps include the following: Step S100: Using the database of the online lending platform, construct a list of loan records, which includes settled loan records and investable loan records.
[0027] In this embodiment, firstly, a loan record list is constructed using the database of the online lending platform. This list includes settled loan records and investable loan records. Settled loan records are those whose repayment period has ended. For example, a self-employed individual borrows money through the platform for short-term working capital and successfully repays the entire amount within the agreed-upon year. This record is crucial for assessing the borrower's creditworthiness, as it directly reflects the borrower's repayment ability and credit level, reducing risk in future investment decisions. Investable loan records are those that can still be invested in. For example, a startup applies for a loan to expand production and plans to repay it gradually over the next three years. These types of loans attract the attention of investors seeking higher returns. This approach not only provides an effective financing channel for businesses and individuals but also offers investors diversified investment options. The advantages of this method are improved risk management capabilities, increased market transparency, and efficient resource allocation, thereby enhancing the health and stability of the entire financial system. By clearly classifying different types of loan status, both borrowers seeking suitable loan products and investors looking for potential investment projects can obtain more accurate information support and make more informed decisions.
[0028] Specifically, step S100 includes the following steps: Step S101: Using the database of the online lending platform, collect all loan records of the online lending platform since its establishment, and record the collection time; Step S102: Based on the loan records and the collection time, obtain the records of settled loans and the records of investable loans; Step S103: Based on the settled loan records and the investable loan records, construct a loan record list.
[0029] In one implementation, such as Figure 2 As shown, firstly, using the database of an online lending platform, all loan records since the platform's inception are collected, and the collection time is recorded. This provides a solid foundation for subsequent data analysis, ensuring the integrity and accuracy of the data, and enabling researchers to fully understand the lending activities on the platform. Next, based on these loan records, each loan is retrieved to accurately distinguish its status. If the collection time is greater than the sum of the loan's bidding end date and repayment date, the loan is considered settled. This method effectively filters out settled loan records, which is crucial for assessing borrowers' creditworthiness and the platform's risk management capabilities. Based on the settled loans, records of settled loans are obtained, where settled loans are those whose repayment period has ended. If the data collection time is during the active bidding phase of the loan, the loan is considered an investable loan, thus helping investors identify current investment opportunities. Based on the investable loans, records of investable loans are obtained, where the active phase refers to the period from when the loan begins accepting investment until when it stops accepting investment. Investable loans are those that can still be invested in. Defining the active phase as the period from when the loan begins accepting investment to when it stops helps to clarify investment windows, enabling potential investors to better plan their fund allocation. Then, based on the settled loan records and the investable loan records, a loan record list is constructed, facilitating data analysis, providing decision support for investors, and improving information transparency and market efficiency. For example, an online lending platform may contain a collection of all loans... Based on whether the loan data was collected after the loan repayment period ended or during the loan bidding stage, loans with completed repayment periods and investable loans are selected to construct a loan record list. Accordingly, all loans with completed repayment periods are defined as a subset of this list. Define all investable loans as its subset. Overall, this approach enhances our understanding of the lending market and facilitates more efficient resource allocation through precise classification and recording of loan status. It also provides a strong basis for risk management and investment decisions.
[0030] Step S200: Based on the settled loan records, obtain the first income data and the first risk data of the settled loan records, and establish a risk prediction model and an income prediction model based on the first risk data and the information of the settled loan records.
[0031] In this embodiment, as Figure 2 As shown, firstly, based on the settled loan records in the loan record list, the first revenue data and first risk data of the settled loan records are obtained. This process helps to clearly understand the actual revenue of past loan projects and their corresponding risk levels, providing a solid data foundation for subsequent analysis. Then, based on the first risk data and the information from the settled loan records, a risk prediction model and a revenue prediction model are established. This step, by training the model using historical data, improves the accuracy and reliability of predicting the potential risks and returns of future loan projects. Overall, this method not only makes full use of existing loan data resources but also achieves a more accurate assessment of new loan projects by constructing scientifically sound prediction models. This helps financial institutions or investors better manage risks and optimize returns, improving decision-making efficiency and the safety and effectiveness of fund utilization.
[0032] Specifically, step S200 includes the following steps: Step S201: Based on the settled loan records, obtain each loan in the settled loan records and the loan information corresponding to each loan; Step S202: Based on the loan information, calculate the revenue data for each loan, and based on the revenue data for each loan, obtain the first revenue data for the settled loan records; Step S203: Based on the first revenue data, the credit rating of each loan, and the number of loans with the same credit rating as each credit rating in the settled loan records, obtain the risk data of the loans belonging to each credit rating in the settled loan records, and obtain the first risk data of the settled loan records based on the risk data of the loans with each credit rating. Step S204: Based on the first risk data, establish a risk prediction model; Step S205: Based on the information of the settled loan records, obtain the relevant attributes of the loan and the borrower of the loan, and establish a revenue prediction model based on the relevant attributes.
[0033] In one implementation, firstly, based on the settled loan records, each loan in the settled loan records and the corresponding loan information for each loan are obtained. The loan information for each loan includes, but is not limited to, cash inflows, cash outflows, and the total time elapsed from loan issuance to repayment, ensuring that subsequent analysis is based on detailed and accurate data, thus improving the quality of the model input. Next, based on the loan information, the revenue data for each loan is calculated, and the revenue for each loan is denoted as... The calculation formula is: ,in, Indicates cash inflow. This represents the total cash outflow, from the time the loan is obtained to the time it is repaid. A period (usually one month). For example, if an investor invests 1,000 yuan in January 2019 and then receives 210 yuan in principal and interest each month for up to 6 months, the cash outflow can be represented as (-1000, 0, 0, 0, 0, 0), and the cash inflow can be represented as (0, 210, 210, 210, 210, 210, 210). The calculation result is 2%. By calculating the return data for each loan and quantifying the investment return using specific formulas, investors are provided with a clear cost-benefit perspective, facilitating an understanding of the investment value of different loans. After calculating the return data for each loan, the first return data for the settled loan records can be obtained based on the return data for each loan. That is, the first return data includes the return data for each loan in the settled loan records. Then, based on the first return data, the credit rating of each loan, and the number of loans in the settled loan records with the same credit rating as each of those credit ratings, the risk data for loans belonging to each of those credit ratings in the settled loan records is obtained. For example, the credit rating of a particular loan... Select all the settled loan records that belong to the credit rating. The loan records are used to construct a subset of loan records. Based on the first income data, the variance of the internal rate of return of these loans in the subset of loan records is calculated and denoted as the total internal rate of return of all loans belonging to the credit rating among the settled loan records. The risk value of a loan is calculated using the following formula: ,in, This indicates that all settled loan records belong to the credit rating category. The risks of loans This indicates that all settled loan records belong to the credit rating category. The amount of loans, This indicates that all settled loan records belong to the credit rating category. The average internal rate of return (IRR) of the loans. For example, if there are 10 loans with credit rating A, and their yields are (0.052, 0.045, 0.049, 0.065, 0.055, 0.042, 0.051, 0.072, 0.040, 0.050), then the risk of loans with credit rating A is 9.83E-05. Risk is measured by calculating the variance of the IRR. This method considers not only the average return level but also the volatility of the return, thus more comprehensively reflecting the risk characteristics of loans under a specific credit rating. After calculating the risk value of all loans belonging to a certain credit rating, the first risk data of the settled loan records can be obtained based on the risk values of all loans belonging to a certain credit rating. That is, the first risk data includes the risk data of each credit rating loan in the settled loan records. Next, based on the first risk data, a risk prediction model is established to help predict the risk status of future loans, enabling financial institutions to better manage risk. The risk prediction model is as follows: ,in, Indicates loan The risk, This indicates that all belong to the credit rating. The risks of loans, here, loans The credit rating is Then, based on the information of the settled loan records, relevant attributes of the loans and the borrowers are obtained. Based on these attributes, a revenue forecasting model is established. Through quantitative analysis, key factors affecting loan yields are identified, providing a basis for making more scientific and reasonable credit decisions. The revenue forecasting model is as follows: ,in, Indicates loan Internal rate of return, Indicates loan and the relevant attributes of the borrower, It is the intercept term. It is the coefficient of each attribute. and It needs to be estimated through a model. For example, if the relevant attributes of the loan and its borrower are income, loan amount, and borrower's credit score, the model would be: ,So, , , , These are the variables that need to be estimated. Furthermore, other machine learning models (such as neural networks and regression trees) can be considered to build risk and return prediction models for loans. Overall, this framework integrates multiple stages, including data collection, return calculation, risk assessment, and predictive model construction. It not only enhances the understanding of historical loan data but also strengthens the ability to predict future loan performance, which is of great significance for optimizing loan portfolio management, reducing default risk, and improving investment returns.
[0034] Step S300: Calculate the second risk data and second return data of the investable loan record according to the risk prediction model and the return prediction model, and construct an optimization model based on the second risk data, the second return data and modern portfolio theory.
[0035] In this embodiment, as Figure 2 As shown, firstly, based on the obtained risk prediction model and return prediction model, the second risk data and second return data of the investable loan records are calculated. This ensures accurate quantitative assessment of each potential investment opportunity, providing investors with a clear risk-return profile and helping to identify investment options that both match their risk appetite and achieve expected return goals. Then, based on the second risk data, the second return data, and modern portfolio theory, an optimization model is constructed. This model aims to maximize the utility of the portfolio through mathematical methods, i.e., seeking maximum return at a given risk level or minimizing risk at a given return level. This reflects a close integration of theory and practice, making investment decisions not only scientific but also effective. Overall, the advantage of the above process lies in its ability to systematically integrate risk management and return optimization into the investment decision-making framework. It utilizes rigorous data analysis and advanced financial theory to guide practical operations, ultimately helping investors build a more robust and efficient portfolio, achieve optimal asset allocation, and pursue continuous and stable investment returns in an uncertain market environment.
[0036] Specifically, step S300 includes the following steps: Step S301: Based on the risk prediction model and the credit rating of each loan in the investable loan record, calculate the risk data of each loan in the investable loan record, and based on the risk data of each loan in the investable loan record, obtain the second risk data of the investable loan record. Step S302: Calculate the revenue data for each loan in the investable loan record based on the revenue prediction model, each loan in the investable loan record, and the relevant attributes of the borrower of each loan. Based on the revenue data for each loan in the investable loan record, obtain the second revenue data for the investable loan record.
[0037] In one implementation, firstly, based on the risk prediction model and the credit rating of each loan in the investable loan record, risk data for each loan in the investable loan record is calculated. Based on this risk data, second risk data for the investable loan record can be obtained; that is, the second risk data includes the risk data of all loans in the investable loan record. This step helps identify potential investment risks, providing investors with a scientific basis to screen suitable investment targets and improve capital security. Next, based on the return prediction model, each loan in the investable loan record, and the relevant attributes of the borrowers of each loan, return data for each loan in the investable loan record is calculated. Based on this return data, second return data for the investable loan record can be obtained; that is, the second return data is the return data of all loans in the investable loan record. This step aims to accurately quantify expected returns, helping investors make more informed investment decisions. Then, based on the second risk data, the second return data, and modern portfolio theory, an optimization model is constructed. This optimization model includes a first model that maximizes the second return data given a fixed second risk data, and a second model that minimizes the second risk data given a fixed second return data. If the investor's goal is to maximize the expected return while maintaining a fixed expected risk, i.e., to maximize the second return data given a fixed second risk data, then the first model is: , If the investor's goal is to minimize the rate-of-realization risk given a fixed expected return, that is, to minimize the second risk given a fixed second return, then the second model is: , .in, This is the second return, also known as the portfolio return; This is the second risk, also known as portfolio risk; It is the first The investment ratio of a loan; It is the first The return on a loan, i.e., the expected return; It is the first The risk of a loan, namely, the expected risk; and The risk and return are predefined. The first model is particularly suitable for investors who are willing to accept a certain level of risk in pursuit of higher returns; while the second model is for investors who want to minimize risk while ensuring a certain return. Overall, this approach can not only effectively balance the risk and return in a portfolio, but also customize the optimal investment strategy according to the investor's specific goals, ultimately achieving scientific and rational asset allocation and improving investment efficiency and effectiveness.
[0038] Step S400: Based on the optimization model, make portfolio recommendations.
[0039] In this embodiment, as Figure 2 As shown, based on the first or second model of the aforementioned optimization model, the optimal investment loan portfolio calculation result is obtained. Through this step, investors can gain a deeper understanding of different loan products, thereby making more informed investment decisions. Then, based on the optimal investment loan portfolio calculation result, an investment portfolio recommendation is made, which presents the user with the basic information of each loan they should invest in, the specific proportion of each loan to be invested in, and the expected risk and expected return of the entire portfolio. This transparency allows users to clearly understand the maximum potential loss and expected profit level, helping them make investment decisions with full knowledge. Overall, the advantages of this method are reflected in improving the scientific nature and accuracy of investment decisions, reducing the risks caused by market uncertainty, and enhancing users' confidence in their investment strategies. Guiding investment behavior through a systematic methodology not only contributes to the growth of personal wealth but also promotes the healthy development of the capital market.
[0040] Preferably, after a period of time (e.g., 3 months), the optimization model is updated and re-optimized based on new data. This method enables the recommendation algorithm to evolve with updated loan records by continuously adding new data, resulting in more accurate results.
[0041] Figure 3 This invention demonstrates the optimal investment-loan portfolio derived using the method of this invention, including loan ID, investment ratio, expected return, expected risk, portfolio return, and portfolio risk. The proposed method addresses the problem of effective investment allocation for investors in online lending and investment scenarios, overcoming the limitations of traditional credit scoring and profit scoring models that only consider investment risk or return. Based on modern portfolio theory, it utilizes nonlinear programming to construct an optimization model and solve for the optimal portfolio. In summary, compared to traditional credit scoring models, this invention's method can improve the investor's return on investment while maintaining the same expected risk; compared to traditional profit scoring models, this invention's method can reduce the investor's expected risk while maintaining the same return on investment.
[0042] In summary, this embodiment first utilizes the database of an online lending platform to construct a loan record list. This list includes settled loan records and investable loan records. Settled loan records are those whose repayment period has ended, while investable loan records are those still eligible for investment. This step lays the data foundation for subsequent analysis, enabling model construction and optimization based on real historical data, thus enhancing the model's accuracy and practicality. Next, based on the settled loan records, first-level return data and first-level risk data are obtained. Based on the first-level risk data and the information from the settled loan records, a risk prediction model and a return prediction model are established. This step helps to deeply understand the impact of different factors on loan returns and risks. This invention provides a more scientific and reasonable portfolio recommendation method by comprehensively integrating historical loan data, constructing and applying risk and return prediction models, and introducing modern portfolio theory. This method not only overcomes the limitations of existing technologies that rely solely on a single indicator to evaluate loan projects, but also achieves the goals of maximizing returns while controlling risk and minimizing risk while maintaining a certain return, providing investors with more accurate and effective investment strategy suggestions. The method also incorporates risk and return prediction models to improve the accuracy of risk and return predictions for investable loans. Furthermore, based on the risk and return prediction models and modern portfolio theory, it calculates second risk data and second return data for the investable loan records. This optimization model not only considers maximizing returns but also minimizes risk while maintaining a certain return.
[0043] like Figure 4As shown in the illustration, this embodiment also provides an online lending portfolio recommendation system, which includes: a loan record list construction module 10, a prediction model building module 20, an optimization model construction module 30, and a portfolio recommendation module 40. Specifically, the loan record list construction module 10 is used to construct a loan record list using the database of an online lending platform. The loan record list includes settled loan records and investable loan records. The prediction model building module 20 is used to obtain first return data and first risk data for the settled loan records based on the settled loan records, and to build a risk prediction model and a return prediction model based on the first risk data and the information of the settled loan records. The optimization model construction module 30 is used to calculate second risk data and second return data for the investable loan records according to the risk prediction model and the return prediction model, and to construct an optimization model based on the second risk data, the second return data, and modern portfolio theory. The optimization model consists of a first model and a second model. The first model maximizes the second return data given a certain second risk data, and the second model minimizes the second risk data given a certain second return data. The portfolio recommendation module 40 is used to recommend portfolios based on the optimization model.
[0044] In one implementation, the loan record list construction module 10 includes: The loan record collection and collection time recording unit is used to collect all loan records of the online lending platform since its establishment using the platform's database, and to record the collection time. The unit for obtaining settled loan records and investable loan records is used to obtain settled loan records and investable loan records based on the loan records and the collection time. The loan record list construction unit is used to construct a loan record list based on the settled loan records and the investable loan records.
[0045] In one implementation, the unit for obtaining settled loan records and investable loan records includes: The loan acquisition subunit is used to acquire each loan in the loan record based on the loan record. The settled loan record acquisition subunit is used to determine if the loan is settled if the collection time is greater than the sum of the bidding end date and the repayment time of the loan, and to acquire settled loan records based on the settled loans, wherein the settled loans are loans whose repayment period has ended. The investable loan record acquisition subunit is used to acquire investable loan records based on the active bidding phase of the loan if the collection time is during that phase. The active phase refers to the period from when the loan starts accepting investment until when the loan stops accepting investment. The investable loan is the loan that is still eligible for investment.
[0046] In one implementation, the prediction model building module 20 includes: The loan and loan information acquisition unit is used to acquire each loan in the settled loan records and the corresponding loan information based on the settled loan records. The first revenue data acquisition unit is used to calculate the revenue data of each loan based on the loan information, and to acquire the first revenue data of the settled loan records based on the revenue data of each loan. The first risk data acquisition unit is used to obtain risk data of loans belonging to each credit rating in the settled loan records based on the first revenue data, the credit rating of each loan, and the number of loans with the same credit rating as each credit rating in the settled loan records, and to obtain the first risk data of the settled loan records based on the risk data of each credit rating. The risk prediction model building unit is used to build a risk prediction model based on the first risk data. The revenue prediction model building unit is used to obtain relevant attributes of the loan and the borrower based on the information of the settled loan records, and to build a revenue prediction model based on the relevant attributes.
[0047] In one implementation, the optimization model construction module 30 includes: The second risk data acquisition unit is used to calculate the risk data of each loan in the investable loan record based on the risk prediction model and the credit rating of each loan in the investable loan record, and to obtain the second risk data of the investable loan record based on the risk data of each loan in the investable loan record. The second revenue data acquisition unit is used to calculate the revenue data of each loan in the investable loan record based on the revenue prediction model, each loan in the investable loan record, and the relevant attributes of the borrower of each loan, and to obtain the second revenue data of the investable loan record based on the revenue data of each loan in the investable loan record.
[0048] The working principle of each module in the online lending portfolio recommendation system of this embodiment is the same as that of each step in the above method embodiment, and will not be repeated here.
[0049] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 5 As shown. The terminal may include one or more processors 100 ( Figure 5 (Only one is shown in the image), memory 101, and a computer program 102 stored in memory 101 and executable on one or more processors 100, such as an online lending portfolio recommendation program. When one or more processors 100 execute computer program 102, they can implement various steps in the online lending portfolio recommendation method embodiment. Alternatively, when one or more processors 100 execute computer program 102, they can implement the functions of various modules / units in the online lending portfolio recommendation method embodiment, which is not limited here.
[0050] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.
[0051] In one embodiment, memory 101 may be an internal storage unit of an electronic device, such as a hard drive or RAM. Memory 101 may also be an external storage device of the electronic device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, memory 101 may include both internal and external storage units. Memory 101 is used to store computer programs and other programs and data required by the terminal. Memory 101 can also be used to temporarily store data that has been output or will be output.
[0052] Those skilled in the art will understand that Figure 5The schematic diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0053] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, operational databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual operating data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. An online lending portfolio recommendation method, characterized by, The method comprises: constructing a loan record list by using a database of an online lending platform, the loan record list comprising closed loan records and investable loan records; obtaining first yield data and first risk data of the closed loan records based on the closed loan records, and establishing a risk prediction model and a yield prediction model based on the first risk data and information of the closed loan records; calculating second risk data and second yield data of the investable loan records according to the risk prediction model and the yield prediction model, and constructing an optimization model based on the second risk data, the second yield data and modern portfolio theory; performing portfolio recommendation based on the optimization model. 2.The online lending portfolio recommendation method of claim 1, wherein, The constructing a loan record list by using a database of an online lending platform comprises: collecting all loan records of the online lending platform since its establishment by using a database of the online lending platform, and recording the collection time; obtaining closed loan records and investable loan records based on the loan records and the collection time; constructing a loan record list based on the closed loan records and the investable loan records. 3.The online lending portfolio recommendation method of claim 2, wherein, The obtaining closed loan records and investable loan records based on the loan records and the collection time comprises: obtaining each loan in the loan records based on the loan records; if the collection time is greater than the sum of the end date of bidding and the repayment time of the loan, the loan is a closed loan, and closed loan records are obtained based on the closed loan; if the collection time is in the active stage of bidding of the loan, the loan is an investable loan, and investable loan records are obtained based on the investable loan, wherein the active stage refers to the time period from the start of accepting investment of the loan to the stop of accepting investment of the loan. 4.The online lending portfolio recommendation method of claim 1, wherein, The obtaining first yield data and first risk data of the closed loan records based on the closed loan records comprises: obtaining each loan in the closed loan records and the loan information corresponding to each loan based on the closed loan records; calculating yield data of each loan based on the loan information, and obtaining first yield data of the closed loan records based on the yield data of each loan; obtaining risk data of loans belonging to each credit rating in the closed loan records based on the first yield data, the credit rating of each loan and the number of loans with the same credit rating as each credit rating in the closed loan records, and obtaining first risk data of the closed loan records based on the risk data of loans of each credit rating. 5.The online lending portfolio recommendation method of claim 4, wherein, The establishing a risk prediction model and a yield prediction model based on the first risk data and information of the closed loan records comprises: establishing a risk prediction model based on the first risk data; obtaining related attributes of loans and borrowers of the loans based on the information of the closed loan records, and establishing a yield prediction model according to the related attributes. 6.The online lending portfolio recommendation method of claim 1, wherein, The second risk data and the second return data of the investable loan records are obtained according to the risk prediction model and the return prediction model, and the method comprises the following steps: According to the risk prediction model and the credit rating of each loan in the investable loan records, the risk data of each loan in the investable loan records is calculated, and the second risk data of the investable loan records is obtained based on the risk data of each loan in the investable loan records. According to the return prediction model, each loan in the investable loan records and the related attributes of the borrower of each loan, the return data of each loan in the investable loan records is calculated, and the second return data of the investable loan records is obtained based on the return data of each loan in the investable loan records. 7.The online lending portfolio recommendation method of claim 1, wherein, The optimization model comprises a first model and a second model, the first model is used to maximize the second return data under the condition of the second risk data, and the second model is used to minimize the second risk data under the condition of the second return data.
8. An online peer-to-peer investment portfolio recommendation system, characterized by, The system comprises: A loan record list construction module is configured to construct a loan record list by using a database of an online lending platform, wherein the loan record list comprises cleared loan records and investable loan records; A prediction model establishment module is configured to obtain first return data and first risk data of the cleared loan records based on the cleared loan records, and establish a risk prediction model and a return prediction model based on the first risk data and information of the cleared loan records; An optimization model construction module is configured to calculate second risk data and second return data of the investable loan records according to the risk prediction model and the return prediction model, and construct an optimization model based on the second risk data, the second return data and modern portfolio theory; A portfolio recommendation module is configured to perform portfolio recommendation based on the optimization model.
9. A terminal, characterized by comprising: The terminal comprises a memory, a processor and an online lending portfolio recommendation program stored in the memory and executable on the processor, and the processor implements the steps of the online lending portfolio recommendation method according to any one of claims 1-7 when executing the online lending portfolio recommendation program.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores an online lending portfolio recommendation program, and the online lending portfolio recommendation program implements the steps of the online lending portfolio recommendation method according to any one of claims 1-7 when executed by the processor.