CN114742657A

CN114742657A - Investment target planning method and system

Info

Publication number: CN114742657A
Application number: CN202210508605.6A
Authority: CN
Inventors: 刘恒江
Original assignee: Hangzhou Hexin Software Technology Co ltd
Current assignee: Hangzhou Hexin Software Technology Co ltd
Priority date: 2022-05-11
Filing date: 2022-05-11
Publication date: 2022-07-12

Abstract

The embodiment of the specification provides an investment objective planning method, which comprises the steps of obtaining an investment objective and a risk objective of a user; determining target profit information of the user through a profit prediction model based on the user portrait, the risk target and the market information, wherein the profit prediction model is a machine learning model; and determining a target realization probability and a target execution decision through an optimization model based on a plurality of target profit information.

Description

Investment target planning method and system

Technical Field

The specification relates to the field of big data, in particular to an investment target planning method and an investment target planning system.

Background

In recent years, more and more users have financial needs. Matching a financial investment scheme more suitable for the user based on the actual situation of the user is a problem which needs to be solved urgently at present.

Therefore, it is desirable to provide an investment objective planning method for providing a user with an investment plan with a higher matching degree only under the condition of acquiring limited user data, so as to assist the user in achieving the financial objective.

Disclosure of Invention

One or more embodiments of the present specification provide a method of investment goal planning. The investment goal planning method comprises the following steps: acquiring an investment target and a risk target of a user; determining target profit information of the user through a profit prediction model based on the user portrait, the risk target and market information, wherein the profit prediction model is a machine learning model; and determining a target realization probability and a target execution decision through an optimization model based on a plurality of target income information.

One or more embodiments of the present specification provide an investment goal planning system. The investment goal planning system comprises: the acquisition module is used for acquiring an investment target and a risk target of a user; the first determination module is used for determining target income information of the user through an income prediction model based on the user portrait, the risk target and the market information, and the income prediction model is a machine learning model; and the second determination module is used for determining target realization probability and target execution decision through an optimization model based on a plurality of target income information.

One or more embodiments of the present specification provide an investment goal planning apparatus comprising at least one processor and at least one memory; the at least one memory is for storing computer instructions; the at least one processor is configured to execute at least some of the computer instructions to implement the investment goal planning method according to any one of the above embodiments.

One or more embodiments of the present specification provide a computer-readable storage medium storing computer instructions which, when read by a computer, cause the computer to perform a method of investment target planning as described in any one of the above embodiments.

Drawings

The present description will be further explained by way of exemplary embodiments, which will be described in detail by way of the accompanying drawings. These embodiments are not intended to be limiting, and in these embodiments like numerals are used to indicate like structures, wherein:

FIG. 1 is a schematic diagram of an application scenario of an investment goal planning system according to some embodiments of the present description;

FIG. 2 is a block diagram of an investment goal planning system according to some embodiments of the present description;

FIG. 3 is an exemplary flow diagram for determining a target achievement probability and a target execution decision, according to some embodiments of the present description;

FIG. 4 is an exemplary flow diagram illustrating the determination of structured target information according to some embodiments of the present description;

FIG. 5 is an exemplary flow diagram illustrating the determination of target revenue information for a user in accordance with some embodiments of the present description;

FIG. 6 is an exemplary flow diagram of a process flow of an optimization model according to some embodiments shown herein.

Detailed Description

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only examples or embodiments of the present description, and that for a person skilled in the art, without inventive effort, the present description can also be applied to other similar contexts on the basis of these drawings. Unless otherwise apparent from the context, or otherwise indicated, like reference numbers in the figures refer to the same structure or operation.

It should be understood that "system", "apparatus", "unit" and/or "module" as used herein is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, other words may be substituted by other expressions if they accomplish the same purpose.

As used in this specification and the appended claims, the terms "a," "an," "the," and/or "the" are not to be taken in a singular sense, but rather are to be construed to include a plural sense unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that steps and elements are included which are explicitly identified, that the steps and elements do not form an exclusive list, and that a method or apparatus may include other steps or elements.

Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of the present specification. It should be understood that the preceding or following operations are not necessarily performed in the exact order in which they are performed. Rather, the steps may be processed in reverse order or simultaneously. Meanwhile, other operations may be added to the processes, or a certain step or several steps of operations may be removed from the processes.

FIG. 1 is a schematic illustration of an application scenario of an investment objective planning system according to some embodiments of the present description.

In some embodiments, the application scenario 100 of the investment goal planning system may include a processor 110, a storage device 120, a user terminal 130, and a network 140. The application scenario 100 may determine target profit information by analyzing and processing the acquired investment target and risk target of the user by implementing the method and/or process disclosed in this specification, thereby determining a target implementation probability and a target execution decision, and implementing an optimal investment plan.

The processor 110 may be used to process data and/or information from at least one component of the application scenario 100 (e.g., the storage device 120, the user terminal 130, and/or the network 140) or an external data source (e.g., a cloud data center). Processor 110 may execute program instructions based on such data, information, and/or processing results to perform one or more of the functions described herein. For example, the processor 110 may obtain the investment target and the risk target set by the user from the user terminal 130. As another example, the processor 110 may determine target revenue information for the user through a revenue prediction model based on the user representation, risk targets, and market information. As another example, the processor 110 may also determine a goal achievement probability and a goal performance decision through an optimization model based on a plurality of goal revenue information.

In some embodiments, the processor 110 may be a single processor or a group of processors. The set of processors may be centralized or distributed (e.g., the processors 110 may be a distributed system), may be dedicated, or may be serviced by other devices or systems at the same time. In some embodiments, the processor 110 may include one or more sub-processing devices (e.g., single core processing devices or multi-core processing devices). Merely by way of example, the processor 110 may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an Application Specific Instruction Processor (ASIP), a Graphics Processing Unit (GPU), a Physical Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a programmable logic circuit (PLD), a controller, a micro-controller unit, a Reduced Instruction Set Computer (RISC), a microprocessor, or the like or any combination thereof.

In some embodiments, processor 110 may be connected locally to network 140 or remotely from network 140. In some embodiments, the processor 110 may be implemented on a cloud platform. By way of example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-tiered cloud, and the like, or any combination thereof.

Storage device 120 may be used to store data and/or instructions. For example, the storage device 120 may store historical user representations, corresponding investment and risk goals, market information, and the like. As another example, the storage device 120 may store parameters for a revenue prediction model, an optimization model, and a target prediction model. Storage device 120 may include one or more storage components, each of which may be a separate device or part of another device.

In some embodiments, storage device 120 may include Random Access Memory (RAM), Read Only Memory (ROM), mass storage, removable storage, volatile read and write memory, and the like, or any combination thereof. Illustratively, mass storage may include magnetic disks, optical disks, solid state disks, and the like. In some embodiments, the storage device 120 may be implemented on a cloud platform.

User terminal 130 refers to one or more user terminals or software used by a user. In some embodiments, the user (e.g., investor, etc.) may be the owner of the user terminal 130. In some embodiments, user terminal 130 may be used for interaction and display with a user. For example, the user terminal 130 may obtain the investment goal and the risk goal of the user through user input. As another example, the user terminal 130 may display the goal achievement probability and the goal performance decision to the user. In some embodiments, the user terminal 130 may be used by one or more users, may include users who directly use the service, and may also include other related users. In some embodiments, the user terminal 130 may be one or any combination of a mobile device 130-1, a tablet computer 130-2, a laptop computer 130-3, a desktop computer 130-4, or other device having input and/or output capabilities.

Network 140 may connect the various components of the system and/or connect the system with external resource components. The network 140 enables communication between the various components, as well as with other components outside the system, facilitating the exchange of data and/or information. In some embodiments, the network 140 may be any one or more of a wired network or a wireless network. For example, network 140 may include a cable network, a fiber optic network, a telecommunications network, the internet, a Local Area Network (LAN), a Wide Area Network (WAN), a Wireless Local Area Network (WLAN), a Metropolitan Area Network (MAN), a Public Switched Telephone Network (PSTN), a bluetooth network, a ZigBee network (ZigBee), Near Field Communication (NFC), an in-device bus, an in-device line, a cable connection, and the like, or any combination thereof. The network connection between the parts can be in one way or in multiple ways.

In some embodiments, the network may be a point-to-point, shared, centralized, etc. variety of topologies or a combination of topologies. In some embodiments, network 140 may include one or more network access points. For example, the network 140 may include wired or wireless network access points, such as base stations and/or network switching points 140-1, 140-2, …, through which one or more components of the access point system may connect to the network 140 to exchange data and/or information.

In some embodiments, storage 120 may be included in processor 110, user terminal 130, and possibly other system components. In some embodiments, the processor 110 may be included in the user terminal 130, as well as other possible system components.

Fig. 2 is a block diagram of an investment goal planning system according to some embodiments of the present description.

In some embodiments, the investment goal planning system 200 may include an acquisition module 210, a first determination module 220, and a second determination module 230.

The acquisition module 210 may be used to acquire investment goals and risk goals for a user. In some embodiments, the acquisition module 210 may be configured to determine an investment goal and a risk goal of the user based on the user profile via a goal prediction model. In some embodiments, the acquisition module 210 may also be used to translate the investment goals and the risk goals into structured investment goals. See FIG. 4 and its associated description for specific details on how to transform the structured investment goals.

The first determination module 220 may be configured to determine target revenue information for the user based on the user representation, the risk target, and the market information via a revenue prediction model, the revenue prediction model being a machine learning model. In some embodiments, the revenue prediction model includes a first model, the first model being a machine learning model. In some embodiments, the first determination module 220 may be configured to determine deviation information based on the user representation, market information; and determining the third-order distance skewness of the user through the first model based on the user portrait, the risk target and the deviation information, wherein the third-order distance skewness represents a target yield probability density curve of the user. In some embodiments, the revenue prediction model includes a second model, the second model being a regression model. In some embodiments, the first determination module 220 may be configured to determine a user forecasted revenue and a market forecasted revenue via the second model based on the user representation, the market information; deviation information is determined based on the user forecasted revenue and the market forecasted revenue. For specific details regarding the revenue prediction model, reference is made to FIG. 5 and its associated description.

The second determination module 230 may be configured to determine a target achievement probability and a target execution decision through the optimization model based on the plurality of target revenue information. In some embodiments, the process of optimizing the model includes constructing expected utilities for multiple phases, and solving the expected utilities for the goal phases via bellman equations. In some embodiments, the processing of the optimization model further includes back-solving a simulation path of the forecast and investment strategies for the future market based on each of the expected utilities of the plurality of phases, e.g., back-solving the market forecast revenue and strategy functions based on each of the expected utilities of the plurality of phases, to obtain the goal achievement probability and the goal execution decision. See FIG. 6 and its associated description for specific details regarding the optimization model.

It should be noted that the above description of the investment objective planning system and its modules is merely for convenience of description and should not be construed as limiting the present disclosure to the illustrated embodiments. It will be appreciated by those skilled in the art that, given the teachings of the present system, any combination of modules or sub-system configurations may be used to connect to other modules without departing from such teachings. In some embodiments, the obtaining module 210, the first determining module 220, and the second determining module 230 disclosed in fig. 2 may be different modules in a system, or may be a module that implements the functions of two or more modules described above. For example, each module may share one memory module, and each module may have its own memory module. Such variations are within the scope of the present disclosure.

FIG. 3 is an exemplary flow diagram illustrating the determination of a target achievement probability and a target execution decision according to some embodiments of the present description. As shown in fig. 3, the process 300 includes the following steps:

step 310, the investment objective and the risk objective of the user are obtained. In some embodiments, step 310 may be performed by an acquisition module.

An investment objective refers to the result or purpose a user desires to achieve during an investment process. In some embodiments, the investment goals may be numerical or text-only information, for example, the investment goals of the user may be 10% annual return, 20000 dollars in profit, copybook, house buying, etc. information.

The risk objective refers to the loss of investment that the user can tolerate during the investment process. In some embodiments, the risk objective may be numerical or plain text information, e.g., the risk objective may be that the loss does not exceed 10% of the principal, the loss does not exceed 10000 dollars, bear a certain loss proportion, etc.

In some embodiments, the investment goal and the risk goal may be set by the user, and the obtaining module may obtain the investment goal and the risk goal set by the user. For example, the user may set information such as "annual profit 20%", "profit 1000 yuan", "copybook", "house buying", "car buying" and the like as an investment target. For example, the user can set information such as "loss does not exceed 10%", "bear a certain loss ratio," and "probability of investment failure is small" as a risk target.

In some embodiments, the investment goals and the risk goals set by the user may be stored in a database (e.g., the storage device 120), and the acquisition module may collect information such as the investment goals and the risk goals of the user directly in the database.

In some embodiments, for a user who does not set an investment goal and a risk goal by himself, the acquisition module may determine the investment goal and the risk goal of the user through a goal prediction model based on the user profile.

The user profile is information reflecting various features in the user investment process. The user profile may include characteristics of the user's gender, age, application function usage preferences, information reading preferences, investment philosophy category, etc. For example, a user representation of user A may be: the male, 30 years old, prefers to invest in stocks, index funds, likes the 'news' function of application programs, has higher attention to financial information, and adopts the investment concept of short-line trading and the like.

The target prediction model can analyze and process the input user portrait and determine the investment target and the risk target of the user.

In some embodiments, the target prediction model may be a semi-supervised learning model, such as FlexMatch.

In some embodiments, a sequence of multiple user images may be used as an input of the target prediction model, and investment targets and risk targets of users corresponding to different user images may be used as an output of the target prediction model.

The parameters of the target prediction model can be obtained through training. In some embodiments, multiple sets of training samples may be obtained based on a large amount of user information, and each set of training samples may include multiple training data and labels corresponding to the training data. The training data may include historical user profiles and the tags may be investment goals and risk goals for the user corresponding to the historical user profiles. For example, a plurality of user representations over a historical period of time (e.g., a week, a month, a year, etc.) may be obtained as training data, and investment goals and risk goals of the user corresponding to the historical user representations (e.g., investment goals and risk goals set by the user themselves) may be obtained as training labels.

In some embodiments, the parameters of the initial target prediction model may be iteratively updated based on a plurality of training samples such that the loss function of the model satisfies a preset condition. For example, the loss function converges, or the loss function value is less than a preset value. And finishing model training when the loss function meets the preset condition to obtain a trained target prediction model.

The method according to some embodiments of the present description can quickly and accurately predict the investment target and the risk target of the user without inputting the investment-related information through the model, so as to facilitate the subsequent determination of the target income information of the user.

In some embodiments, the obtaining module may splice the output result of the target prediction model with user data in which the investment target and the risk target have been set by itself, and finally obtain the investment target and the risk target of all users.

In some embodiments, the stitching may refer to summarizing the investment target and the risk target of the user predicted based on the target prediction model and the investment target and the risk target set by the user, for example, arranging the investment target and the risk target corresponding to each user portrait according to an arrangement order of the user portraits, and the like.

In some embodiments, the acquisition module may process the previously acquired investment goals and risk goals for the totality of users to convert them into structured goals.

Structured targets refer to quantifiable investment targets and risk targets having a standard format. For example, the structured target can be expressed as { desired profitability (μ), standard deviation profitability (σ), investment period (t), affordable loss fraction (l), target achievement/exceeding risk probability (p) }.

In some embodiments, the structured target may be obtained by performing extraction and conversion processing on the obtained effective information of the investment target and risk target of the user based on a pre-adjusted computing device (such as the processor 110). See fig. 4 for more description of the conversion of the investment goals and the risk goals into structured goals.

And step 320, determining target income information of the user through an income prediction model based on the user image, the risk target and the market information. In some embodiments, step 320 may be performed by the first determining module.

Market information refers to information of financial markets, wherein the financial markets may include money markets, bond markets, stock markets, and the like. In some embodiments, market information may include market factors, scale factors, value factors, momentum factors, and the like.

The target profit information refers to information related to profits that the user can obtain under the influence of the risk target and the market information. For example, the target profit information may be annual profit 3.4%, profit 2500 yuan, or the like.

In some embodiments, the first determination module may determine the target revenue information for the user through a revenue prediction model, wherein the revenue prediction model is a machine learning model.

The profit prediction model can analyze and process the input user portrait, risk target and market information to determine the target profit information of the user.

In some embodiments, a sequence of a plurality of user images, risk targets and market information may be used as an input of the target prediction model, and target profit information corresponding to different user images under the influence of the risk targets and the market information may be used as an output of the profit prediction model.

The parameters of the revenue prediction model may be obtained through training. In some embodiments, multiple sets of training samples may be obtained based on a large amount of user information, and each set of training samples may include multiple training data and labels corresponding to the training data.

The training data may include historical user profiles, historical risk goals, and historical market information, and the tags may be target revenue information corresponding to the historical user profiles under the influence of the historical risk goals and the historical market information. For example, the first determining module may obtain a plurality of user portraits, risk targets and market information as training data over a historical period of time (e.g., a week, a month, a year, etc.), and obtain a determination result (e.g., a determination result manually made by relying on professional experience) of target revenue information corresponding to the risk targets and the market information of the plurality of user portraits over the historical period of time as a training label or obtain actual target revenue information of the user as the training label.

In some embodiments, parameters of the initial revenue prediction model may be iteratively updated based on a plurality of training samples such that a loss function of the model satisfies a preset condition. For example, the loss function converges, or the loss function value is smaller than a preset value. And when the loss function meets the preset condition, completing model training to obtain a trained revenue prediction model.

For more explanation regarding the determination of the user's target revenue information by the revenue prediction model, see FIG. 5.

Step 330, determining a target realization probability and a target execution decision through an optimization model based on a plurality of target profit information. In some embodiments, step 330 may be performed by the second determining module.

The target achievement probability refers to a likelihood size of achieving the target benefit. E.g., 70%, 67%, etc. In some embodiments, if a single user has targeted revenue information for multiple investment events:

the target achievement probability may include a probability that all target revenue information for a single user is achieved. For example, the target profit information I of the user a is: 10000 yuan is earned, and the corresponding target realization probability is 70 percent; the target profit information II is: the three-month annual yield is not lower than 4.0%, and the corresponding target realization probability is 80%; the probability that all the target revenue information for that user is fully realized is 70%. 80%. 56%.

In some embodiments, the target achievement probability may include a probability of achieving as much target revenue information as possible for the user. For example, it may be preset that if the implementation probability of the single target profit information exceeds a certain threshold (e.g., 50%), the single target profit information is regarded as the target profit to be implemented, and the probability of implementing all the target profits to be implemented is the target implementation probability. In some embodiments, a weight value may be set for each target profit information, and the target profit information whose weight value exceeds a preset threshold value may be used as the target profit to be achieved. In some embodiments, the weight value of each target revenue information is positively correlated with its degree of contribution to the desired utility of the user. The higher the contribution degree, the higher the weight value. See fig. 6 and its corresponding text for a description of the desired utility.

In some embodiments, the target achievement probability may include a probability at which the overall utility is optimal. The consideration criterion of the overall optimal utility may be determined based on the user, for example, the user may set the highest profit value within a predetermined time (e.g., 1 month, half year, etc.) as the overall optimal utility, or, for example, the user may set the lowest risk value as the overall optimal utility.

The objective execution decision refers to an investment strategy and/or scheme to be executed for realizing objective income, and can be divided into a single objective execution decision and a multi-objective execution decision. The single-target execution decision is for a user with only single target profit information, and the single-target execution decision may be represented based on the optimized structured target, for example, the target execution decision corresponding to the user D is { marriage, the profit rate is expected to be 15%, the profit rate standard deviation is 15.5%, the investment cycle is 1, the bearable loss is 5%, and the success/super-risk probability is 80%/20% }.

The multi-objective implementation decision includes the priority of the user to invest in the objectives, and the expected profitability and standard deviation of the investment to be achieved under each objective, for example, the objective implementation decision may include: each investment objective, corresponding priority and corresponding optimized structured objective data. For example, the investment goals for user E are: planning to buy the car in the next year, wherein the target profit is 10 percent; buying a house three years later is planned, and the target income is 20%. The corresponding execution decision may be expressed as: { buying car, in priority, yield is expected to be 10%, yield standard deviation is 5%, investment cycle 1, bearable loss is 5%, success/excess risk probability is 80%/20% }, { buying room, high priority, yield is expected to be 20%, yield standard deviation is 5%, investment cycle 3, bearable loss is 10%, success/excess risk probability is 90%/20% }.

In some embodiments, the second determination module may determine a target achievement probability and a target execution decision through an optimization model based on a plurality of the target revenue information. In some embodiments, the optimization model may be a multi-objective optimization model implemented based on Mean-Variance Theory (MVT). For example, the obtained multiple target revenue information of the user may be input into a multi-objective optimization model implemented based on Mean-Variance Theory (MVT), a utility function of the user asset is constructed based on the multi-objective optimization model, the overall utility is optimized and calculated under the constraint of user risk, the implementation cost and the implementation period for each target are balanced, the implemented and failed targets are dynamically removed, and the overall implementation probability of the comprehensive optimization target and the target implementation decision for each target in the whole process are finally output.

See FIG. 6 for further explanation regarding determining goal achievement probabilities and goal execution decisions through the optimization model.

The method according to some embodiments of the present specification determines the corresponding target profit information by analyzing the actual investment goals and risk goals of different users, thereby enabling risk control and making an execution decision more suitable for the user.

FIG. 4 is an exemplary flow diagram illustrating the determination of structured target information according to some embodiments of the present description. In some embodiments, the flow 400 may be performed by an acquisition module. As shown in fig. 4, the process 400 includes the following steps:

and step 410, classifying the text content of the acquired investment target and risk target of the user.

In some embodiments, the investment goals and risk goals of the user may be textual information. In some embodiments, the acquisition module may divide the contents of the investment objective and the risk objective into a content including valid numerical information, a content not including valid numerical information, a risk awareness content, and the like.

The content containing effective numerical information refers to the related content of which the investment goals of the user have specific numerical values. For example, the content containing the effective numerical value information may be "the expected annual profit margin is more than 7%" "10000 yuan are earned", or the like.

The content not containing the effective numerical information refers to the related content having no specific numerical value in the investment goals of the user. For example, the content that does not contain valid numerical information may be: "buy room", "educate savings", etc.

Risk-aware content refers to the relevant content of the loss that a user can afford during the investment process. For example, the risk awareness content may be "5% loss acceptable", "must be maintained", and the like.

In some embodiments, the obtaining module may classify the investment goal and the risk goal of the user according to the text content of the above criteria based on the obtained content of the information in the investment goal and the risk goal of the user.

In step 420, different processing is performed on the content of different classification results.

In some embodiments, the acquisition module can translate content containing significant numerical information into a profitability expectation in the structured targets. For example, the acquisition module may extract values from content containing significant numerical information (e.g., "annual rate of interest is 20%") and convert the values to a desired rate of interest (e.g., 20%) by a regularization method.

In some embodiments, for the investment target part in the content not containing the valid numerical information, the acquisition module may determine the corresponding investment period (t), principal and target amount according to the category of the investment target in combination with the user image. For example, the acquisition module determines a corresponding investment period (e.g., 1 year, 3 years, etc.), principal (e.g., 20 ten thousand, etc.) and target amount (e.g., 25 ten thousand, etc.), by determining a category (e.g., wedding, car, etc.) of the investment target in the content (e.g., "wedding and car"), which does not contain valid numerical information, in combination with a user profile (e.g., male, 29 year old, preferred investment product is bond-type fund, investment philosophy is long-term holding, etc.), wherein the investment period corresponds to the investment period in the structured target.

In some embodiments, for risk objective portions of content that do not contain valid numerical information, the acquisition module can translate the risk-aware content into an affordable deficit proportion (l) in the structured objective. For example, a risk tolerance category (e.g., conservative) may be determined based on risk awareness content (e.g., "not losing"), thereby determining an affordable deficit proportion (e.g., 0).

At step 430, a yield expectation and a yield standard deviation are determined.

In some embodiments, for a user who does not obtain a rate of return expectation based on valid numerical information, the obtaining module may calculate the rate of return expectation (μ) and the rate of return standard deviation (σ) in the structured target by using a Capital Asset Pricing Model (CAMP) in combination with the obtained information of the user image, the investment period, the principal and the target amount. Wherein the capital asset pricing model may be E (r)_i)＝r_f+β_im(E(r_m)-r_f) Wherein, E (r)_i) Is the expected rate of return, r, for asset i_fIs risk-free, beta_imI.e. systematic risk of asset i, E (r)_m) Is the expected market return for market m, E (r)_m)-r_fIs the market risk premium, i.e., the difference between the expected market return and the no risk return. And inputting information such as user portrayal, investment period, principal and target amount and the like into a capital asset pricing model to obtain the corresponding expected income ratio (mu) and standard deviation of the income ratio (sigma).

At step 440, structured target information is determined.

In some embodiments, the obtaining module may calculate the target achievement/exceeding Risk probability (p) in the structured target by importing the profitability expectation and the standard deviation into a Risk Value model (VaR), and then combining the obtained bearable loss proportion of the user, to finally obtain the structured target.

Wherein the format of the structured target is: { expected rate of return (μ), standard deviation of return (σ), investment cycle (t), affordable deficit fraction (l), target achievement/excess risk probability (p) }. For example, the structuring target may be (μ ═ 9.00%, σ ═ 0.015, t ═ 1 year, l ═ 10%, p ═ 35%/15%).

The traditional investment target risk questionnaire can only show limited and fixed options to users, so that the expression appeal of the users is greatly limited, and the threshold of the users for understanding the professional terms is increased. The method according to some embodiments of the present description can effectively convert the content about the investment target and the risk target, which is set or input by the user, into the quantifiable and structured investment and risk targets, thereby effectively reducing the understanding cost of the user and improving the interaction experience of the user.

FIG. 5 is an exemplary flow diagram illustrating the determination of target revenue information for a user according to some embodiments of the present description. As shown in fig. 5, the process 500 includes the following steps. In some embodiments, the flow 500 may be performed by a first determination module.

The first determination module may determine deviation information based on the user representation, the market information, step 510.

The deviation information is information indicating a deviation of the predicted user profitability from the predicted market profitability, and for example, the deviation information may include a deviation and a right deviation, where a left deviation indicates that the predicted user profitability is smaller than the predicted market profitability, and may indicate that the user is more likely to make a profit. The right bias means that the user predicted profitability is greater than the market predicted profitability, and the user is less likely to be profitable on behalf. The user predicted profitability may represent a profitability that a user's investment goal, as determined in combination with the current user representation and market information, may achieve at a future pre-set time; the market predicted profit represents predicted profit margin information of the financial market invested by the user at a preset time in the future. The reference market may include different indices such as Shanghai depth 300, Shanghai, and Shenzhou, which may be determined based on the user profile.

In some embodiments, the first determination module may determine the deviation information in a variety of ways, for example, the first determination module may manually obtain the deviation information based on experience, automatically obtain the deviation information based on a preset prediction table, obtain the deviation information based on a multiple linear regression fit, and the like.

For example only, the first determining module may respectively realize the predicted user predicted profitability and the predicted market profitability by fitting based on a plurality of user portraits in the historically stored data and the user historical predicted profitability and the market historical predicted profitability corresponding to the market information.

For example, the first determining module may fit a plurality of user images and market information in the historically stored data by a least square method, establish a fit curve, fit a relationship between the user images and the market information and the predicted profitability of the user and the predicted profitability of the market, and then calculate the predicted profitability of the user and the predicted profitability of the market according to the obtained user images and the market information. In some embodiments, the first determination module may further record a history fitting process, and modify the calculated predicted profitability of the user and the calculated predicted profitability of the market through the history fitting process to obtain the predicted profitability of the user and the predicted profitability of the market with higher accuracy.

After the predicted profitability of the user and the predicted market profitability are determined, the first determining module may determine deviation information based on a deviation condition of the predicted profitability of the user compared with the predicted market profitability, and if the predicted profitability of the user is smaller than the predicted market profitability, the deviation information is left deviation. And when the predicted profitability of the user is greater than the market predicted profitability, the deviation information is right deviation.

In some embodiments, the first determination module may also determine deviation information based on the user representation, the market information, by way of model processing. In determining deviation information based on the model implementation, the first determination module may perform the following operational steps:

and 511, the first determining module determines the user predicted profitability and the market predicted profitability through the second model based on the user portrait and the market information.

In some embodiments, the revenue prediction model may include a second model, which may be a logistic regression model. The first determination module may derive a user predicted rate of return and a market predicted rate of return based on processing of the user representation, the market information, and the second model.

In some embodiments, the input to the second model may include a user representation, market information, and the output of the second model may include a user predicted rate of return and a market predicted rate of return.

In some embodiments, the second model may be trained using a plurality of labeled training samples. For example, a plurality of labeled training samples may be input into the initial second model, a loss function may be constructed from the labels and the results of the initial second model, and parameters of the initial second model may be iteratively updated based on the loss function. And finishing model training when the loss function of the initial second model meets the preset condition to obtain the trained second model. The preset condition may be that the loss function converges, the number of iterations reaches a threshold, and the like.

In some embodiments, the training samples may include at least multiple sets of historical sample data, and each set of historical sample data may include one sample user icon and its corresponding multiple sample market information. The label can be a user historical yield corresponding to the corresponding sample market information and a corresponding market historical yield when a specific sample user draws the image. The labels may be obtained based on manual labeling, for example, manually labeling the samples based on the actual user profitability and the actual market profitability corresponding to the historical actual users corresponding to the sample user portraits under various types of sample market information.

The first determination module determines deviation information based on the user predicted profitability and the market predicted profitability, step 512.

In some embodiments, the first determining module may obtain deviation information of the user predicted profitability compared with the market predicted profitability based on a comparison process of the user predicted profitability and the market predicted profitability obtained in step 511, for example, the deviation information is left deviation when the user predicted profitability is smaller than the market predicted profitability. And when the predicted profitability of the user is greater than the market predicted profitability, the deviation information is right deviation.

In some embodiments, for users for whom historical data does not store a user's historical profitability of the corresponding user representation for each market information, the first determination module may utilize the obtained information user representation to predict deviation information for such users via a neural network model, such as an information prediction model, e.g., the first determination module may predict deviation information for the user's predicted profitability as compared to the market predicted profitability based on processing of the user representation by the information prediction model.

In some embodiments, when training the information prediction model, a plurality of labeled training samples may be used for training by a plurality of methods (e.g., gradient descent method), so that the parameters of the model may be learned. And when the trained model meets the preset conditions, finishing the training and acquiring the trained information prediction model.

The training samples may include a user profile of a historically stored user with historical rate of return information. The label of the training sample may be the total deviation information of the historical profitability of the user with the historical profitability information. For example, all the categories of the deviation information obtained by the user may be counted, and the category of the deviation information with the largest number may be used as the label of the training sample corresponding to the user. The labels of the training samples may be obtained through statistics by manual labeling, for example, 100 pieces of historical profit data of a user corresponding to a certain user portrait sample are stored, where deviation information corresponding to 70% of the historical profit rate data is left deviation, and when the deviation information exceeds a threshold score threshold (e.g., 60%), the deviation information corresponding to the user portrait sample is considered left deviation. In some embodiments, the information prediction model may be trained in another device or module.

Some embodiments of the present description may implement, through processing of the model, determining deviation information based on the user portrait and the market information, which may effectively improve accuracy of predicted deviation information, reduce difficulty of manual calculation, and improve efficiency of data processing.

In step 520, the first determination module may determine a third order moment skewness of the user through the first model based on the user representation, the risk objective, and the deviation information.

The third-order distance skewness may be used as one of the factors representing reasonable target rate-of-return information for the user, e.g., based on the third-order distance skewness in combination with the expected rate-of-return and the standard deviationA probability density curve representing a reasonable target profitability of the user can be drawn. In some embodiments, the third order skewness of the user may be expressed based on the following formula,

where X represents the variable in the distribution, i.e., the different historical profitability of the user, μ is the profitability expectation, and σ is the profitability variance. E2]Indicating the desire.

In some embodiments, the revenue prediction model includes a first model, the first model may be a machine learning model, and the first determination module may determine the third order distance skewness of the user based on processing of the user representation, the risk objective, and the deviation information by the first model.

The probability density curve representing a reasonable target profitability of the user based on the third-order distance skewness of the user means that, for example, it can be assumed that the profitability of the user follows a lognormal distribution, wherein the decision parameters are the desired profitability and the standard deviation of the profitability. And adjusting the randomly initialized distribution shape (adjusting the fixed expectation and the standard deviation) according to the third-order distance skewness output by the first model, so as to obtain a reasonable target yield probability density curve of the user.

In some embodiments, the first determining module may determine the target profit information of the user based on the obtained target profit-rate probability density curve that is reasonable for the user, for example, may determine a profit-loss interval with a higher probability (e.g., exceeding a preset probability threshold) based on the target profit-rate probability density curve that is reasonable for the user, define a super-parameter (e.g., a fixed expectation and a standard deviation) based on the profit-loss interval with the higher probability, and set a suitable probability (e.g., the probability is greater than a preset minimum probability threshold) corresponding to the profit information as the target profit information.

The first model is a machine learning model that processes the user representation, risk objectives, and deviation information to derive a third order distance skewness of the user, and in some embodiments, the first model may include a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), or the like, or any combination thereof.

In some embodiments, the first model may be trained using a plurality of labeled training samples. For example, a plurality of labeled training samples may be input into an initial first model, a loss function may be constructed from the labels and the results of the initial first model, and parameters of the initial first model may be iteratively updated based on the loss function. And finishing model training when the loss function of the initial first model meets the preset condition to obtain the trained first model. The preset condition may be that the loss function converges, the number of iterations reaches a threshold, and the like.

In some embodiments, the training sample may include at least a plurality of sets of historical user data, each set of historical user data may include a sample user representation and its corresponding risk objective and deviation information. In some embodiments, the lead investor may be used as a user sample in a training sample, and the label may be a third-order offset of each lead investor calculated based on the distribution of the history of the investor and future income information. The tags may be obtained based on manual calculations, e.g., based on calculation formulas manually based on various types of historical revenue information obtained for the head investor

And calculating the third-order offset of the head investor.

In some embodiments, training the loss function of the first model includes classifying a cross-entropy loss function, a skewness-ordering loss function, a skewness regularizing constraint loss function. The classification cross entropy loss function (namely, left-right-bias cross entropy loss) can be constructed based on the skewness information of the obtained user prediction income relative to the market prediction income or constructed based on the label of an information prediction model, and the classification cross entropy loss function is mainly used for supervising the strength of the prediction income of the learning investor (namely, the predicted user) relative to the average income capacity of the market. The skewness ranking loss function (namely, left-right-skewness ranking loss) can also be constructed based on skewness information of the obtained user predicted income relative to the market predicted income or label construction based on an information prediction model, and the skewness ranking loss function can mainly supervise the strength of the predicted income of the learning investor relative to the income capacity of the rest investors. The partial regularization constraint loss function refers to partial regularization constraint loss based on an expected target, is a regularization constraint and is mainly based on an economic theory, and monitors psychological expected deviation of investors of different expected targets.

In some embodiments, the first model is supervised and trained based on three loss functions, so that an output result obtained based on the trained first model may include at least supervised optimization information corresponding to each loss function.

Some embodiments of the present description can determine target profit information in combination with the user's own situation based on the user portrait, the risk target and the market information through processing of the profit prediction model, and further determine a reasonable investment target and risk in combination with the user's own situation, thereby effectively improving accuracy of predicted target profit information, reducing difficulty of manual calculation, and improving efficiency of data processing.

FIG. 6 is an exemplary flow diagram of a process flow of an optimization model in accordance with certain embodiments shown herein. In some embodiments, the flow 600 may be performed by the second determination module 230. As shown in fig. 6, the process 600 includes the following steps:

and step 610, constructing expected utility of multiple periods, and solving the expected utility of the target period through a Bellman equation.

The expected utility may represent aggregated utility information for the utility of various investment results that may be available to the user under uncertain conditions. In some embodiments, the desired utilities may include asset utility u (w) and consumer utility u (c).

In some embodiments, a period of time may be preset as a period to construct a corresponding expected utility based on the preset period of time, e.g., a period may correspond to a year, and the expected utility for period t may include asset utility u (W) for period t_t) And the consumption utility u (C) of the t-th stage_t). The expected utility for stage t +1 may include asset utility u (W) for stage t +1_t+1) And the consumption utility u (C) of the t +1 th stage_t+1)。

In some embodiments, the optimization model may solve for the expected utility of the goal period based on bellman's equations, e.g., if the goal period is the tth period, the optimization model may achieve the expected utility of the tth period based on:

wherein, the optimization conditions of the optimization model are the target k to be achieved at each stage and the simulated investment strategy l, and see step 330 of fig. 3 for further description of the optimization model.

In some embodiments, the optimization model may be based on a bellman equation, and in combination with the investment objective k to be realized at each stage and the simulated investment strategy l, the solution of the expected utility at the objective stage is realized, which may facilitate the subsequent inverse solution of the market prediction profit and the strategy function based on the expected utility at each stage, so as to obtain the objective realization probability and the objective execution decision.

And step 620, reversely solving the simulation path of the market prediction yield and the investment strategy at each stage based on each stage expected utility in the expected utilities at the plurality of stages.

The simulation path of the investment strategy can represent the cash flow change situation of the user constructed according to the generated balance of a plurality of investment targets, namely the cash flow situation after the investment targets are added; such as information that may include the amount of cash, security assets, liabilities, etc. of the year t.

In some embodiments, the processing of user representations and market information based on the wealth lifecycle model may result in a user's revenue expenditure and asset forecasts throughout the life cycle.

In some embodiments, the second determining module may obtain the cash flow condition after adding the investment targets by dividing the fund change condition of each investment target of the user in each period of the investment cycle into items (such as cash, securities and liabilities) and adding the items to obtain the net asset data, and then merging the net asset data into the asset data of each period output by the wealth life cycle model.

By way of example only, if the wealth lifecycle model, the cash flow for user a in the next three years is: [300000,400000,500000] Yuan, the corresponding structured targets of the existing one-investment-target buying vehicle are: { 30%, 20%, 3 years, 10%, 80%/40% }, with a principal of 100000 dollars, and a net investment budget of [ -50000, -50000,130000] for the next three years according to a certain investment strategy, then the cash flow after the implementation of the investment strategy becomes [250000,350000,630000 ].

In some embodiments, the parameters of the wealth lifecycle model may be derived through training. In some embodiments, multiple sets of training samples may be obtained based on a large amount of user information, and each set of training samples may include multiple training data and labels corresponding to the training data. The training data may include historical user profiles and historical market information, and the tags may be revenue expenditure and asset values of the user corresponding to the historical user profiles.

In some embodiments, the parameters of the initial wealth lifecycle model may be iteratively updated based on a plurality of training samples such that the loss function of the model satisfies a preset condition. For example, the loss function converges, or the loss function value is smaller than a preset value. And finishing model training when the loss function meets the preset condition to obtain a trained wealth life cycle model.

The reverse solution represents that calculation is deduced from back to front in the solution process, the second determination module is based on each expected utility in the expected utilities of the multiple stages, when the reverse solution is carried out on the simulation path of the market prediction yield and the investment strategy in each stage, the last stage (if the last stage is the t stage) is the maximum life span of the calculation object, and the utility only needs to consider the asset utility u (W stage)_t) I.e. assets in penultimate period need to reach W_tFrom this, the expected utility u (W) of the penultimate stage can be written_t-1)+u(C_t-1)+τ[u(W_t)]Wherein [ tau ], []Representing a discount on a cross-term growth. By analogy, the bellman equation described in step 610 can be obtained.

In some embodiments, the first order partial derivatives of the Bellman equation are calculated to obtain the maximum expected utility value at each stageConditions, constructing a strategy function k for each period of the investment objective k and the simulated investment strategy l according to the result of the Monte Carlo simulation (i.e., the expected value for obtaining the return on the security asset)_t～W_tAnd l_t～W_tAnd completing reverse solution based on the strategy function.

It should be noted that as the number of investment targets of a single user increases, the complexity of the model increases. When the complexity of the model reaches a certain degree, the solution efficiency can be improved by utilizing deep reinforcement learning, for example, the solution can be completed by replacing the strategy function in the solution process through a deep neural network based on a strategy gradient deep reinforcement learning method.

Step 630, obtaining a target realization probability and a target execution decision.

Based on the solution in step 620, the corresponding optimized structured target of the investment target k under the corresponding expected utility can be obtained. The second determination module can calculate the corresponding target realization probability based on the structured targets of the investment targets. For a description of the target achievement probability and the calculation thereof, refer to step 330 of fig. 3, and are not described herein again. Based on the structured targets of the investment targets, which are used to calculate the target realization probability, corresponding target execution decisions can be obtained. For the description of the target execution decision, refer to step 330 in fig. 3, which is not described herein again.

In some embodiments, after determining the target achievement probability and the target execution decision, the second determination module may further determine an investment strategy based on the target achievement probability and the target execution decision, and recommend the investment strategy to the user.

In some embodiments, in conjunction with the goal performance decision and market information obtained in step 630, the second determination module may match the corresponding investment strategy with the standard deviation of profitability according to the desired profitability expected to be achieved by the investment at each goal from the investment strategy scenario base and recommend the matched investment strategy to the user. When the matching is performed, the second determining module may preferentially match a policy with the same expected profitability and standard deviation of the profitability from the scenario base according to the target profitability and standard deviation of the profitability of the target (e.g., the investment target corresponding to the target execution decision with the highest priority) that needs to be executed currently, and recommend the policy to the user based on a Click-Through-Rate (CTR) recommendation manner.

In some embodiments, the optimization model is implemented by constructing an optimization model such as a multi-objective optimization model based on Mean-Variance Theory (MVT); multiple investment goals of a user can be comprehensively considered, and reasonable trade-offs can be made among the investment goals so as to obtain the overall profit optimization effect.

It should be noted that the above description related to the flow 600 is only for illustration and description, and does not limit the application scope of the present specification. Various modifications and changes to flow 600 will be apparent to those skilled in the art in light of this description. However, such modifications and variations are intended to be within the scope of the present description.

Having thus described the basic concept, it will be apparent to those skilled in the art that the foregoing detailed disclosure is to be regarded as illustrative only and not as limiting the present specification. Various modifications, improvements and adaptations to the present description may occur to those skilled in the art, although not explicitly described herein. Such modifications, improvements and adaptations are proposed in the present specification and thus fall within the spirit and scope of the exemplary embodiments of the present specification.

Also, the description uses specific words to describe embodiments of the specification. Reference throughout this specification to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic described in connection with at least one embodiment of the specification is included. Therefore, it is emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, some features, structures, or characteristics of one or more embodiments of the specification may be combined as appropriate.

Additionally, the order in which the elements and sequences of the process are recited in the specification, the use of alphanumeric characters, or other designations, is not intended to limit the order in which the processes and methods of the specification occur, unless otherwise specified in the claims. While various presently contemplated embodiments of the invention have been discussed in the foregoing disclosure by way of example, it is to be understood that such detail is solely for that purpose and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements that are within the spirit and scope of the embodiments herein. For example, although the system components described above may be implemented by hardware devices, they may also be implemented by software-only solutions, such as installing the described system on an existing server or mobile device.

Similarly, it should be noted that in the preceding description of embodiments of the present specification, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the embodiments. This method of disclosure, however, is not intended to imply that more features than are expressly recited in a claim. Indeed, the embodiments may be characterized as having less than all of the features of a single embodiment disclosed above.

Numerals describing the number of components, attributes, etc. are used in some embodiments, it being understood that such numerals used in the description of the embodiments are modified in some instances by the use of the modifier "about", "approximately" or "substantially". Unless otherwise indicated, "about", "approximately" or "substantially" indicates that the number allows a variation of ± 20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximations that may vary depending upon the desired properties of the individual embodiments. In some embodiments, the numerical parameter should take into account the specified significant digits and employ a general digit preserving approach. Notwithstanding that the numerical ranges and parameters setting forth the broad scope of the range are approximations, in the specific examples, such numerical values are set forth as precisely as possible within the scope of the application.

For each patent, patent application publication, and other material, such as articles, books, specifications, publications, documents, etc., cited in this specification, the entire contents of each are hereby incorporated by reference into this specification. Except where the application history document does not conform to or conflict with the contents of the present specification, it is to be understood that the application history document, as used herein in the present specification or appended claims, is intended to define the broadest scope of the present specification (whether presently or later in the specification) rather than the broadest scope of the present specification. It is to be understood that the descriptions, definitions and/or uses of terms in the accompanying materials of this specification shall control if they are inconsistent or contrary to the descriptions and/or uses of terms in this specification.

Finally, it should be understood that the embodiments described herein are merely illustrative of the principles of the embodiments of the present disclosure. Other variations are also possible within the scope of the present description. Thus, by way of example, and not limitation, alternative configurations of the embodiments of the present specification can be seen as consistent with the teachings of the present specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. An investment objective planning method comprising:

acquiring an investment target and a risk target of a user;

determining target revenue information of the user through a revenue prediction model based on the user portrait, the risk target and the market information, wherein the revenue prediction model is a machine learning model;

and determining a target realization probability and a target execution decision through an optimization model based on a plurality of target income information.

2. The method of claim 1, the obtaining investment goals and risk goals for a user comprising:

and determining an investment target and a risk target of the user through a target prediction model based on the user portrait.

3. The method of claim 1, the obtaining investment goals and risk goals for a user further comprising:

converting the investment objective and the risk objective into a structured objective.

4. The method of claim 1, the revenue prediction model comprising a first model, the first model being a machine learning model;

the determining, by a revenue prediction model, target revenue information for the user based on the user representation, the risk target, and market information includes:

determining deviation information based on the user representation and the market information; the deviation information is used for representing the deviation condition of the user predicted yield and the market predicted yield;

and determining a third-order distance skewness of the user through the first model based on the user portrait, the risk target and the deviation information, wherein the third-order distance skewness represents a target rate of return probability density curve of the user.

5. The method of claim 4, the revenue prediction model comprising a second model, the second model being a regression model;

the determining deviation information based on the user representation, the market information, comprises:

determining, by the second model, the user forecasted profitability and the market forecasted profitability based on the user representation, the market information,

determining the deviation information based on the user forecasted profitability and the market forecasted profitability.

6. The method of claim 1, the processing of the optimization model comprising: and constructing expected utility of multiple phases, and solving the expected utility of the target phase through a Bellman equation.

7. The method of claim 6, the processing of the optimization model comprising: and reversely solving on a simulation path of market prediction yield and investment strategy at each stage based on each expected utility in the expected utilities at multiple stages to obtain the target realization probability and the target execution decision.

8. An investment objective planning system comprising:

the acquisition module is used for acquiring an investment target and a risk target of a user;

the first determination module is used for determining target income information of the user through an income prediction model based on the user portrait, the risk target and the market information, and the income prediction model is a machine learning model;

and the second determination module is used for determining target realization probability and target execution decision through an optimization model based on a plurality of target income information.

9. An investment goal planning apparatus, comprising at least one processor and at least one memory;

the at least one memory is for storing computer instructions;

the at least one processor is configured to execute at least some of the computer instructions to implement the method of any of claims 1-7.

10. A computer-readable storage medium, wherein the storage medium stores computer instructions that, when executed by a processor, implement the method of any one of claims 1-7.