TWM597468U - System for scoring investment item - Google Patents

Abstract

The disclosure is related to a system for scoring an investment item according to correlations between an investor and the similar investors. In the system, a plurality of routes are formed according to correlations between multiple investment items and the related attributes that are obtained from a user who submits an investment recommendation request and the similar investors. Every route is configured to have a weight. When performing the scoring process, a plurality of calculation routes can be randomly selected according to connectivity of the plurality of routes. A score can be calculated based on the weights assigned to the plurality of selected calculated routes. After repeatedly performing the scoring process, multiple scores can be obtained for providing a recommendation list of the investment items.

Description

Investment project scoring system

This creation involves a plan to provide clients with investment projects, especially an investment project scoring system based on personal relevance and using data analysis.

A financial commodity investor will determine the investment target based on various information. Some financial institutions will provide a self-assessment questionnaire. After the investors fill out the various questions in the questionnaire, they will make a comprehensive judgment based on the questionnaire results to evaluate the investor. The risk attributes, such as conservative, stable, active growth, etc., afterwards, the wealth management specialist or through software judgment, recommend investors some investment targets based on the assessed risk attributes.

A rational investor will choose the one with the lowest risk among several portfolios with the same expected return; or, in another case, if several portfolios have the same investment risk, the investor will choose the expected return The highest one. Such a portfolio is called an optimal portfolio (Efficient Portfolio).

The current portfolio method of recommended investors, such as a "Markowitz Efficient Frontier", which represents a collection of various optimal portfolios, and each point on the Markowitz efficiency frontier curve represents An optimal investment portfolio, which can be used to infer the optimal investment portfolio of investors.

The specification discloses a system that provides users with recommendations for investment projects based on their relevance to similar investors. According to an embodiment of the investment project scoring system, the system includes a recommendation server, an analysis server, and a model server. Method: First, the recommendation server receives the investment recommendation request from the user, and then the user's investment data can be obtained from the database, and the investment data of other investors similar to the user can be obtained to form the user and multiple investments The relationship between the project and various attributes related to each investment project.

In the analysis server, multiple paths can be formed according to the correlation between users, multiple investment projects, and various attributes related to each investment project, where each path is provided with a weight value.

In the model server, multiple calculation paths are randomly selected according to the connection relationship of multiple paths. According to the individual weight values of the selected multiple calculation paths, a score can be calculated, and after repeating the scoring algorithm, multiple calculation paths can be selected. Randomly select multiple different calculation paths on the path, and then obtain multiple scores. The multiple scores represent the scores of multiple investment projects.

Preferably, the model server performs a graph analysis to illustrate a plurality of scoring nodes obtained according to the relevance of users and their similar investors. Each scoring node is composed of users, investment items, and related attributes.

Preferably, in the scoring process, the user, multiple investment projects and each investment project can be formed based on the user's investment data recorded in the database of the computer system and the investment data of other investors similar to the user The correlation between various related attributes, and the similarity between the user and other investors includes a similar background, similar preferences, and similar investment behavior.

Preferably, the investment behavior includes one or any combination of various investment items, transaction amount and transaction time bought by the user or other investors; and the preference is personalized by the user or other investors The information obtained includes browsing data on financial web pages, current products, customer group labeling, various types of asset balances and their basic attributes.

Further, in the scoring algorithm, multiple scoring nodes are established based on multiple paths formed by the correlation between users, multiple investment projects, and various attributes related to each investment project, and the scoring algorithm is self-used The scoring node of the author starts, randomly selects one of the other scoring nodes connected to each scoring node, and finally connects to the scoring node formed by one of the multiple investment projects, in which multiple calculation paths are formed and multiple calculations are performed Multiple weight values of the path are added or multiplied to obtain a score.

Further, a recommendation list of investment projects can be proposed based on multiple scores obtained by the repeated scoring algorithm. The investment projects in the recommendation list include one of stocks, funds, futures, bonds, insurance, derivative commodities, and virtual currencies. One or any combination of them.

In order to further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and explanation only, and are not intended to limit the present invention.

The following are specific specific examples to illustrate the implementation of this creation. Those skilled in the art can understand the advantages and effects of this creation from the content disclosed in this specification. This creation can be implemented or applied through other different specific embodiments. The details in this specification can also be based on different views and applications, and various modifications and changes can be made without departing from the concept of this creation. In addition, the drawings in this creation are only a schematic illustration, not based on the actual size, and are declared in advance. The following embodiments will further describe the relevant technical content of the creation, but the disclosed content is not intended to limit the protection scope of the creation.

It should be understood that although terms such as “first”, “second”, and “third” may be used herein to describe various elements or signals, these elements or signals should not be limited by these terms. These terms are mainly used to distinguish one component from another component, or one signal from another signal. In addition, the term "or" as used herein should include any combination of any one or more of the associated listed items, depending on the actual situation.

The specification discloses an investment project scoring system that can be implemented using data analysis based on the relevance of investment behaviors of similar investors to perform investment scoring and recommendation. One of the purposes is to use the preferences of similar users Under the accepted risk attribute, users are recommended to be familiar with multiple sequential investment commodities that are acceptable and interested in purchasing, and can also provide suitable financial targets for new customers or customers with no investment experience.

Different from the current portfolio recommendation method, the proposed investment project scoring and recommendation method is mainly to find out other investors who are similar to the customer's investment behavior, and obtain the investment target of other investors with the same preference as this customer. Then analyze the investment attribute data and establish the investment attribute data related to this customer according to a certain degree of similarity, including multiple investment projects and various attributes related to each investment project, forming the nodes required by the algorithm, according to the association between each node The weight value is given by nature, and it can be randomly walked between multiple nodes obtained through mathematical algorithms (or in accordance with certain rules, such as the travel path cannot be repeated, limiting the number of traveling nodes), and the scores of multiple investment projects can be comprehensively calculated. The result is a portfolio that provides customer investment recommendations.

FIG. 1 shows a flowchart of an embodiment of an investment project scoring system for performing scoring and recommendation. The investment project scoring and recommendation method mainly runs on a computer system. A computer system processor executes a scoring algorithm to allow banks, Securities companies and other institutions provide recommendation lists based on investment recommendation requests made by users (such as bank customers).

In this method, as in step S101, the software program running on the computer system can query the database according to the characteristics of the customer, and after comparing the database data, other investors similar to the user can be obtained, including a similar background, Similar preferences and similar investment behaviors. This database records the investment information of various investors (including the requesting user), including the historical data of various investments, and the recorded investment behavior includes various investment projects, transaction amounts, and transaction time purchased by users or other investors One of them or any combination. The database also records the preferences of various investors. These preferences can be derived from the personalized information of users or other investors, including browsing data on financial web pages, current products, customer group labeling, and various types of assets. Balance and its basic attributes.

As such, in step S103, the method will obtain the related investment items and attributes based on the investment data of the user who made the investment recommendation request and the investment data of other investors similar to this user, and form this user , The correlation between multiple investment projects and various attributes related to each investment project.

In step S105, in the computer system, multiple paths are formed according to the correlation between users, multiple investment projects, and various attributes related to each investment project. These paths indicate that the users, investment projects, and attributes are based on the customer and The connection relationship established by the relevance of the investment project and its attributes, and the users, investment projects, and related attributes all form the scoring nodes required by the algorithm in the investment project scoring method. And in step S107, the weight value of each path will be set according to the relevance between the user and various investment items and attributes, so the personalized weight value of the connection relationship between the nodes is obtained.

In step S109, the executed scoring algorithm may randomly select multiple calculation paths according to the connection relationships of multiple paths between the scoring nodes, and then calculate a score based on the individual weight values of the selected multiple calculation paths The calculation is to show the index of each investment project for the user who made the investment recommendation request, and an algorithm can be proposed according to the rules set by the weight value, for example, the weights on these calculation paths can be added or multiplied Out. In this way, the scores of customers and different investment projects can be calculated.

Then, the scoring algorithm described above can be repeated. Each calculation randomly selects different calculation paths on multiple paths, and then obtains multiple scores. The multiple scores represent the scores of multiple investment projects. Sorting, such as step S111, can generate a list of recommended investments.

Figure 2 shows a schematic diagram of an embodiment of an implementation of a scoring system for investment projects.

The figure shows that the computer system is used to implement multiple functions on the system side, which can be implemented by different software or hardware servers, and the processor executes the relevant method flow. According to an embodiment, the system end in the figure is provided with a recommendation server 21, an analysis server 22, and a model server 23, and the servers (21, 22, 23) are connected through the network 20, and may include records of investors’ A database 24 of personal basic data, investment data and preference data, the investment project scoring system runs the investment project scoring method through these servers.

The user can operate the software program in the terminal device 201, and make an investment recommendation request via the network 20. This request may carry a client identification code, for example, the user should obtain a unique code through registration before obtaining the service, or may obtain an identity card One of the font size and web browsing records (such as cookies) is used as the basis for identifying the user's identity.

On the system side, the recommendation server 21 receives the investment recommendation request made by the user. The recommendation server 21 can obtain the investment information of the user from the database 24, and obtain the investment information of other investors similar to the user to form the relationship between the user, multiple investment projects, and various attributes related to each investment project. Relevance. For example, the recommendation server 21 obtains from the database 24 the investment projects that the user has invested in or has queried (according to the web browsing history) in the past period, and can query the investment information of other investors based on this investment project to obtain Information related to other investors also includes historical investment information.

After that, in the analysis server 22, the analysis can be performed based on the user and its surrounding information, including the investment behavior of the person with investment relevance, and the connection relationship between the user and various investment items and their attributes can be obtained, that is, according to the income The correlation between the outgoing users, multiple investment projects, and various attributes related to each investment project form multiple paths, each path represents a connection relationship, and each path is provided with an analyzed personalized weight value, so that The model server 23 randomly selects a plurality of calculation paths according to the connection relationships of the plurality of paths, and calculates a score based on the individual weight values of the selected plurality of calculation paths. Similarly, after repeatedly executing the scoring algorithm, multiple different calculation paths are randomly selected, and then multiple scores are obtained, and these scores respectively represent the scores of multiple investment projects.

It is worth mentioning that, in the analysis server 22, the current investment behavior can be analyzed against various investment project lists of other investors of similar users, and the user’s past transaction investment target (which can indicate user preferences) can be set Different weights, personalized and investment commodity data are analyzed by graphs and sent to the model server 23. In the model server 23, you can refer to the schematic diagram of the graph analysis shown in FIG. 3, and randomly select the calculation path through the scoring algorithm to generate each user’s preference score for various investment projects (such as financial commodities), and finally according to the preference The scores are sorted from high to low to provide a list of investment recommendations.

Among them, the user can be notified of the investment recommendation list generated by the calculation by means of newsletters, e-mails, push messages, etc. The investment recommendation list can record various algorithms to obtain investment project rankings suitable for the user's attributes.

In one embodiment, because there are many variables related to the user, the similar investors and related investment items obtained can be variable values, and the weight value on the calculation path between the scoring nodes will also change Therefore, the investment project scoring system that implements the investment project scoring and recommendation method can be set to regularly provide the user's investment recommendation list, and others can also immediately broadcast the risk level, investment and financial management items, and the system updates and recommends commodities every day.

Fig. 3 shows an example diagram of calculating the score of an investment project using a graph analysis method. This example shows a schematic diagram of graph analysis performed by a model server in a computer system on the system side. The various scoring nodes 32 are composed of users, investment projects and related attributes. In practical applications, the investment project includes one or any combination of stocks, funds, futures, bonds, insurance, derivative commodities, and virtual currencies.

In this example, the scoring nodes represented by images include: User 30, according to the analysis, a variety of commodities related to User 30 are obtained, such as Item 1 P1, Item 2 P2, and Item 3 P3, and the analysis and various Commodity-related attributes such as net worth (C1, C2, C3), risk (T1, T2, T3) and associated market (M1, M2). According to an implementation example, the meanings expressed by the scoring nodes 32 include: net value C1 may indicate growth of more than 5%, net value C2 may indicate growth of more than 10%, and net value C3 may indicate growth of less than 5%; various risk levels such as Risk T1 can be expressed as a positive type, risk T2 can be expressed as a growth type, and risk T3 can be expressed as a stable and conservative type; market M1 such as the United States, and market M2 such as China can also be expressed in currencies such as the US dollar and the RMB.

The figure also shows that, in the analysis server, multiple scoring nodes 32 are established based on multiple paths formed by the association between the user 30, multiple investment items and various attributes related to each investment item. Nodes 32 establish paths with algorithms that randomly walk through calculations according to the association, where each path is set with a weight value.

Furthermore, there is a path between each node in the scoring node 32, and each path is provided with a personalized weight value. In the investment project scoring method, the higher the weight value of each path, it means that the user and The higher the relevance of the investment project or related attributes, the higher the weight value of the investment project or its attribute directly related to the user, and the higher the weight value of the investment project or its attribute that is only related to other investors similar to the user low.

When the scoring algorithm is executed, the weight relationship between nodes can be learned through the algorithm. At the beginning of the calculation, you can start from the node of user 30, adopt a random walk method, according to rules (such as limiting the number of steps, the same attributes (such as net worth, risk and market) are not directly connected, etc.), at each scoring node Randomly select one of the other connected scoring nodes on 32, and finally connect to a node formed by one of multiple investment projects (P1, P2, P3), where multiple calculation paths are formed, and multiple calculation paths are formed. The weight values are added or multiplied to get a score. This example shows that the scoring algorithm links the net worth C1, the market M1, the net worth C2, the risk T2, the net worth C3, and the risk T1, and finally links to item two P2 to form the illustrated calculation path 305.

Similarly, the above scoring algorithm can be repeated, multiple calculation paths are randomly selected based on the connection relationship of multiple paths, and multiple scores corresponding to multiple items are calculated according to the individual weight values of the selected multiple calculation paths. A score indicates the score of multiple investment projects, which are ranked to form an investment recommendation list for users.

For the weight calculation function, refer to the embodiment diagram shown in FIG. 4. The diagram shows the relationship between the subject h, the relationship r and the object t expressed in vectors (including direction and value). The subject h indicates the user, the object r indicates the investment project, and the relationship r indicates that the relationship between the user and the investment project is the purchase , Sell, etc. When the system acquires users (subject h) and other investors with similar backgrounds, similar preferences and similar investment behaviors, they can derive a variety of investment projects (object t) based on their respective investment behaviors, including users or other investments One or any combination of various investment items, transaction amount and transaction time that the purchaser bought. The strength of the relationship (relation r) between the subject h and the object t can be obtained through learning (such as machine learning, big data analysis), thereby obtaining the weight value shown schematically in Figure 3 on the connection of each node, which can be calculated The scores of various investment projects recommended to users.

FIG. 5 continues to show a flowchart of another embodiment of an investment project scoring system performing scoring and recommendation.

When the system receives an investment recommendation request from a user (such as a bank customer) (step S501), and obtains customer characteristics (step S503), the characteristics such as data generated by the user’s past investment behavior, during actual operation, the system proposes The model server will obtain the customer's past investment behavior (the proportion of each financial product transaction amount) and refer to the customer's investment behavior and basic information of the wealth management product (currency, risk level, market) in the past. , Dynamic performance of wealth management products (net worth changes, hot ranking), Internet browsing clicks, products of the bank, customer group labeling, various types of asset balances, customer basic data and other data for graphical analysis, so available Out of other investors who have investment relevance to this client (step S505), and obtain the associated investment items and attributes (step S507) and establish a scoring node based on the relevance and determine the weight value between the nodes (step S509).

You can also refer to Figure 3 to show the schematic diagram of the scoring algorithm after the graph. The above investment items and attributes derived from the user's personal information and similar investors form multiple scoring nodes 32 as shown in Figure 3, each scoring node according to the association , There can be a connection relationship with each other, forming a variety of paths, each path is given a personalized weight value, in addition to the scoring node 32 formed by various investment projects and attributes in the middle, it is ultimately necessary to connect to multiple investment projects P1, P2, and P3 . When executing the scoring algorithm, starting from the user 30, the algorithm randomly walks to select the connected scoring node 32, forming a calculation path 305 shown schematically in this legend during the calculation process, and then traversing the multiple paths on the calculation path 305 The weighted value of is added up or a score is obtained by a specific algorithm to become the score of an investment project P1, P2 or P3.

In the process, through the model calculation, according to the set conditions, randomly connect the scoring nodes to calculate the investment project score, that is, calculate the preference score of each customer for each financial product (step S511), which is calculated by the repeated scoring algorithm The multiple scores are listed, and finally ranked according to the preference score from the highest to the lowest as the order of recommended commodities at this time, providing an investment recommendation list (step S513).

In summary, according to the description of the above embodiment, the proposed investment project scoring method is to calculate the investment project score for the relevance of investors and similar investors, and then can provide a recommended investment list according to the scoring results, and includes implementation The system of these methods, and the proposed method does not use the conventional Markowitz efficiency frontier curve to infer the best investment portfolio, but based on analysis of other investors similar to users (such as bank customers). Investment projects, through an algorithm, are recommended to recommend this user to be familiar with and acceptable to purchase multiple sequential investment projects.

The content disclosed above is only a preferred and feasible embodiment of the new model, and does not limit the scope of the patent application of the new model. Therefore, any equivalent technical changes made by using the description of the new model and the content of the drawings are included in the application of the new model. Within the scope of the patent.

201: terminal device 20: Internet 21: Recommended server 22: Analysis server 23: Model server 24: database h: main body r: relationship t: object 32: Scoring node C1, C2, C3: Net worth M1, M2: market T1, T2, T3: Risk P1: Project one P2: Project 2 P3: Project three 30: user 305: calculation path Steps S101～S111: Investment project scoring and recommendation process Steps S501～S513: Investment project scoring and recommendation process

FIG. 1 shows one of the flow charts of an embodiment of an investment project scoring system performing scoring and recommendation;

Figure 2 shows a schematic diagram of an embodiment of an implementation of the scoring system for investment projects;

Figure 3 shows an example diagram of calculating the score of an investment project using a graph analysis method;

4 is a schematic diagram showing an embodiment of a weight calculation function;

FIG. 5 shows the second flowchart of the implementation example of the scoring and recommendation of the investment project scoring system.

S101: draw similar investors based on customer characteristics

S103: Obtain related investment projects and attributes

S105: Establish a connection relationship based on the relevance of customers and investment projects and their attributes

S107: Obtain the personalized weight value of the connection relationship

S109: Calculate the scores of customers and different investment projects

S111: Generate a list of recommended investments

Claims (10)

An investment project scoring system, including: One recommended server; An analysis server; and A model server, connecting the recommendation server and the analysis server via a network; among them: The recommendation server receives an investment recommendation request from a user; Obtain the investment information of the user from a database, and obtain the investment information of other investors similar to the user to form the association between the user, multiple investment projects, and various attributes related to each investment project ; In the analysis server, multiple paths are formed according to the correlation between the user, the multiple investment projects, and the multiple attributes related to each investment project, where each path is provided with a weight value; In the model server, a plurality of calculation paths are randomly selected according to the connection relationship of the plurality of paths, and a score is calculated according to the individual weight values of the selected plurality of calculation paths; and Repeating the scoring algorithm, randomly selecting different multiple calculation paths on the multiple paths, and then obtaining multiple scores, the multiple scores respectively representing the scores of multiple investment projects. The investment project scoring system as described in claim 1, wherein other investors similar to the user include similar backgrounds, similar preferences and similar investment behaviors. The investment project scoring system according to claim 2, wherein the investment behavior includes one or any combination of various investment projects, transaction amount, and transaction time purchased by the user or other investors. The investment project scoring system as described in claim 2, wherein the preference is derived from the personalized data of the user or other investors, including browsing data on financial webpages, current products, customer group labeling, and various types of assets Balance and its basic attributes. The investment project scoring system according to claim 1, wherein the higher the weight value of each path, the higher the relevance of the user to the investment project or related attributes, wherein the investment directly related to the user The weight value of the project or its attribute is high, and the weight value of the investment project or its attribute only related to other investors similar to the user is low. The investment project scoring system according to claim 1, wherein, in the scoring algorithm, a recommendation list of investment projects is proposed based on the multiple scores obtained by repeating the scoring algorithm. The investment project scoring system according to claim 6, wherein the investment project includes one or any combination of stocks, funds, futures, bonds, insurance, derivative commodities and virtual currency. The investment project scoring system according to any one of claims 1 to 7, wherein in the scoring algorithm, based on the association between the user, the plurality of investment projects, and the plurality of attributes related to each investment project The multiple paths formed by the property establish multiple scoring nodes. The scoring algorithm begins with the user’s scoring node, randomly selects one of the other scoring nodes connected to each scoring node, and finally connects to the multiple investments A scoring node formed by one of the items, wherein the plurality of calculation paths are formed, and the weight values of the plurality of calculation paths are added or multiplied to obtain the score. The investment project scoring system according to claim 8, wherein the model server performs a graph analysis to illustrate a plurality of scoring nodes obtained according to the relevance of the user and its similar investors. Each scoring node is determined by the user , Investment projects and related attributes. The investment project scoring system according to claim 9, wherein the analysis server analyzes the current investment behavior of various investment project lists of other investors similar to the user, and sets different personalized weights for the user’s past investment target, Provide the model server for graph analysis.

Images