WO2019192134A1

WO2019192134A1 - Portfolio optimization method, device, and storage medium

Info

Publication number: WO2019192134A1
Application number: PCT/CN2018/102222
Authority: WO
Inventors: 李正洋; 周璟
Original assignee: 平安科技（深圳）有限公司
Priority date: 2018-04-03
Filing date: 2018-08-24
Publication date: 2019-10-10
Also published as: CN108717664A

Abstract

The present application provides a portfolio optimization method. The method comprises: receiving stocks in an input initial portfolio to be optimized and an initial weight corresponding to each stock in the initial portfolio, and obtaining a market corresponding to each stock and history data of each stock in the initial portfolio in a first preset time period; calculating a weight corresponding to each stock in the initial portfolio, and determining a first portfolio subjected to preliminary optimization; and predicting a prospective logarithmic rate of return of each stock in the first portfolio, and determining a second portfolio. The present application further provides an electronic device and a storage medium. By using the present application, the correlation between each two individual stocks in an initial portfolio is minimized as much as possible, and a weight corresponding to each stock in a portfolio is adjusted by predicting a prospective rate of return of each stock, so as to improve the portfolio return.

Description

Portfolio optimization method, device and storage medium

The present application is based on the priority of the Chinese Patent Application entitled "Investment Portfolio Optimization Method, Apparatus and Storage Medium" filed on April 3, 2018, the disclosure of which is hereby incorporated by reference. The way is combined in this application.

Technical field

The present application relates to the field of financial big data mining, and in particular, to a portfolio optimization method, an electronic device, and a computer readable storage medium.

Background technique

A portfolio refers to a collection of stocks, bonds, and derivative financial products held by a financial institution or investor. Taking stocks as an example, how to allocate the proportion of each stock in the portfolio to achieve certain goals, such as income pair To minimize language or risk, the purpose of building a portfolio is to spread risk.

However, in the portfolio, the weight of each stock is not static. Currently, the industry's weighting of stocks in portfolios is usually in the form of manual adjustments. Such methods rely more on the individual trading experience of the researchers, lack of objectivity, low efficiency, and are not conducive to controlling risks.

Summary of the invention

The present application provides a portfolio optimization method, an electronic device, and a computer readable storage medium, the main purpose of which is to minimize the correlation between individual stocks in the initial investment portfolio, and to predict the future yield of each stock by comparing The weights corresponding to each stock in the portfolio are adjusted to increase the portfolio income.

To achieve the above objective, the present application provides a portfolio optimization method, the method comprising:

Receiving the initial weights of the stocks in the initial portfolio to be optimized and the stocks in the initial portfolio, and obtaining the historical data of the market corresponding to each stock and the stocks in the initial investment portfolio in the first preset time;

Calculating, according to the historical data and the preset first analysis rule, a weight corresponding to each stock in the initial investment portfolio when the correlation between the stocks in the initial investment portfolio is minimum, and initial investment Adjusting the weight corresponding to each stock in the portfolio to determine the first optimized investment portfolio; and

Obtaining historical data of each stock in the first investment portfolio in a second preset time, constructing a feature vector of each stock, and inputting a pre-trained future profit rate prediction model corresponding to each stock, predicting the first The future logarithmic yield of each stock in a portfolio is adjusted according to a preset second analysis rule for the weight corresponding to each stock in the first portfolio, and the second portfolio is determined.

In addition, in order to achieve the above object, the present application further provides an electronic device, including: a memory, a processor, and a portfolio optimization program stored on the memory, where the portfolio optimization program is executed by the processor The steps of the portfolio optimization method as described above are implemented.

In addition, in order to achieve the above object, the present application further provides a computer readable storage medium, where the portfolio optimization program is stored, and the portfolio optimization program is executed by a processor to implement the above The steps of the portfolio optimization method.

Compared with the prior art, the portfolio optimization method, the electronic device and the computer readable storage medium proposed by the present application minimize the correlation between individual stocks in the initial investment portfolio while keeping the income unchanged. The initial investment portfolio is initially optimized to obtain the first portfolio with the least risk; in addition, the future logarithmic yield of each stock in the first portfolio is predicted by the LSTM model, and the future yield is relatively high or low. The weight of the stock is adjusted to obtain the second investment portfolio, which improves the investment income to a certain extent.

DRAWINGS

1 is a schematic diagram of a preferred embodiment of an electronic device of the present application;

2 is a program module diagram of the portfolio optimization program of FIG. 1;

3 is a flow chart of a preferred embodiment of a portfolio optimization method of the present application.

The implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings.

detailed description

It is understood that the specific embodiments described herein are merely illustrative of the application and are not intended to be limiting.

The application provides an electronic device 1 . Referring to FIG. 1 , it is a schematic diagram of a preferred embodiment of the electronic device 1 of the present application.

In the embodiment, the electronic device 1 includes a memory 11, a processor 12, a network interface 13, and a communication bus 14.

The memory 11 includes at least one type of readable storage medium. The at least one type of readable storage medium may be a non-volatile storage medium such as a flash memory, a hard disk, a multimedia card, a card type memory, or the like. In some embodiments, the readable storage medium may be an internal storage unit of the electronic device 1, such as a hard disk of the electronic device 1. In other embodiments, the readable storage medium may also be an external storage device of the electronic device 1, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (SMC). , Secure Digital (SD) card, Flash Card, etc.

In the present embodiment, the readable storage medium of the memory 11 is generally used to store a portfolio optimization program 10 or the like installed in the electronic device 1. The memory 11 can also be used to temporarily store data that has been output or is about to be output.

The processor 12, in some embodiments, may be a Central Processing Unit (CPU), microprocessor or other data processing chip for running program code or processing data stored in the memory 11, such as performing portfolio optimization. Program 10 and so on.

The network interface 13 may include a standard wired interface, a wireless interface (such as a WI-FI interface). Usually used for data transmission with terminals (not identified in the figure).

Communication bus 14 is used to implement connection communication between these components.

Figure 1 shows only the electronic device 1 with components 11-14 and portfolio optimization program 10, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.

Optionally, the electronic device 1 may further include a user interface, and the user interface may include a display, an input unit such as a keyboard, and the optional user interface may further include a standard wired interface and a wireless interface.

Optionally, the electronic device 1 may further include a display, and in some embodiments, an LED display, a liquid crystal display, a touch liquid crystal display, and an Organic Light-Emitting Diode (OLED) touch device. The display is used to display information processed in the electronic device and a user interface for displaying visualizations.

In the apparatus embodiment shown in FIG. 1, a memory optimization program 10 is included in the memory 11 as a computer storage medium. When the processor 12 executes the portfolio optimization program 10 stored in the memory 11, the following steps are implemented:

In the present embodiment, the scheme is described with stocks in the CSI 300, but is not limited to the CSI 300. Understandably, when adjusting the weights of each stock in the portfolio, on the one hand, it is to control the risk of the portfolio, and on the other hand, to improve the return of the portfolio, the scheme considers both risk minimization and consideration. Increased revenue. It should be noted that the weight mentioned in this scheme refers to the proportion of the investment amount corresponding to each stock in the portfolio to the total amount of investment.

When receiving the initial investment portfolio to be optimized input by the investor, respectively obtaining the weights corresponding to each stock and each stock in the initial investment portfolio, for example, the initial investment portfolio includes N (for example, 20) stocks, and the N The weights corresponding to only stocks are W ₀ =[w ₀₁ ,w ₀₂ ,w ₀₃ ,...,w _0n ]. Obtain historical data of N stocks and markets (for example, CSI 300) in the initial portfolio for the first preset time (for example, 2 months before the optimization of the initial portfolio), for example, N stocks The closing price of each period within 2 months before the optimization, the closing price of each period of the market within 2 months before the optimization.

After determining the historical data corresponding to each stock in the initial investment portfolio and the historical data of the market, it is necessary to adjust the weight corresponding to each stock in the initial investment portfolio according to the existing data, and under the condition that the profit rate remains unchanged, Minimizing the correlation of stocks in a portfolio can minimize the risk of a portfolio.

Preferably, the steps to minimize the correlation of each stock in the portfolio include:

First, a logarithmic rate of return corresponding to each stock in the market and the initial investment portfolio in a first preset time period is separately calculated.

Specifically, taking the weight corresponding to each stock in the initial investment portfolio every other week as an example, the number of stocks in the initial portfolio is N=20, and the first preset time T=60.

On the day of optimizing the initial investment portfolio, calculate the logarithmic rate of return of the 20 stocks in the initial portfolio for the first 60 trading days, synthesize the daily logarithmic rate of return, and generate the corresponding logarithmic rate of return, logarithmic returns. The formula for calculating the rate is:

RS _it =lnP _it -lnP _i(t-1)

Where RS _it represents the logarithmic rate of return of the stock at the t-th time in the logarithmic rate of return corresponding to the i-th stock in the initial portfolio, and P _it represents the closing price of the i-th stock at time t, P _{i (t-1)} indicates the closing price of the i-th stock at time t-1, and t indicates a certain time before the optimization of the initial investment portfolio, that is, a trading day.

Calculate the daily logarithmic rate of return of the market corresponding to 20 stocks in the initial portfolio for the first 60 trading days, and synthesize the daily logarithmic rate of return to generate a corresponding logarithmic rate of return. The formula for calculating the logarithmic rate of return is :

RM _t =lnPM _t -lnPM _t-1

Where RM _t represents the logarithmic rate of return of the market at the t-th time in the logarithmic rate of return corresponding to the market, PM _t represents the closing price of the market at time t, PM _t-1 represents the closing price of the market at time t-1, and t represents At some point before the optimization of the initial portfolio, that is, a trading day.

The log yield series corresponding to each stock in the initial investment portfolio in the first preset time is [RS _i1 , RS _i2 , RS _i3 ,..., RS _it ], and the market corresponding log yield series is [RM ₁ , RM ₂ , RM ₃ , . . . , RM _t ] wherein t=T=60, 0<i≤N=20.

Then, the logarithmic income sequence corresponding to each stock in the initial investment portfolio is respectively regressed with the logarithmic income sequence of the market, and the residual sequence corresponding to each stock in the initial investment portfolio is calculated respectively, and the obtained residue is obtained. The covariance matrix corresponding to the initial portfolio.

Specifically, for each stock in the initial portfolio, the logarithmic rate of return series and the market's logarithmic rate of return series are respectively returned to obtain:

RM _t =α _i +β _i *RS _it +e _it

By regression, we obtain: α _i =[α ₁ ,α ₂ ,α ₃ ,...,α _n ], β _i =[β ₁ ,β ₂ ,β ₃ ,...,β _n ], and according to: e _it = RM _t -(α _i +β _i *RS _it ), and obtain the residual sequence of the i-th stock in the initial investment portfolio within the first preset time (60 days). Among them, for the i-th stock, α _i represents the excess return of the stock relative to the market, β _i represents the degree of linkage between the trend of the logarithmic yield of the stock and the trend of the market logarithmic yield, and e _it is the residual item. Represents a random interference term other than α _i , β _i , that is, a residual sequence corresponding to the stock in the initial portfolio, and e _it =[e _1i , e _2i , e _3i , . . . , e _Ti ].

Finally, the solution of the covariance weighted sum between the stocks in the covariance matrix corresponding to the initial portfolio is calculated, and the weight corresponding to each stock in the initial portfolio is determined according to the calculation result.

Specifically, using the residual sequence of each stock in the first preset time, the covariance between each stock is calculated as:

X=e _it =[e _1i ,e _2i ,e _3i ,...,e _Ti ]

Cov(X,Y)=E[(X-E[X])(Y-E[Y])]=E[XY]-E[X]E[Y]

Covariance is used in probability theory and statistics to measure the overall error of two variables. A covariance matrix of N*N (ie, 20*20) can be written by the covariance between two and two stocks, denoted as Σ.

It can be understood that the covariance matrix represents the degree of correlation between individual stocks in the portfolio. When the covariance is smaller, it often represents the more dispersed risk. Therefore, after confirming the covariance matrix corresponding to the initial portfolio, the demand is to minimize the covariance weighted sum between the stocks. The formula is:

Min W ^T ∑W

St∑ _i w _0i *RS _i =R

e ^T W=1

Where W is to be solved, W=[w ₁ ,w ₂ ,...,w _n ], indicating the weight corresponding to each stock, ∑ represents the covariance matrix corresponding to the initial portfolio, and RS _i represents the optimized portfolio The logarithmic rate of return corresponding to each stock in the previous moment, w _0i represents the initial weight corresponding to each stock in the initial portfolio, and RS is a constant, indicating the total logarithmic yield of the initial portfolio at the previous moment of the optimized investment portfolio, meaning In the case of controlling the profit of the optimized portfolio to be equal to the gain of the previous trading day, the correlation between the two stocks is reduced as much as possible, and e ^T W=1 indicates that the sum of the weights of each stock in W is 1.

After obtaining W=[w ₁ , w ₂ , . . . , w _n ] by the above steps, comparing W=[w ₁ , w ₂ , . . . , w _n ] and W ₀ =[w ₀₁ ,w ₀₂ ,w ₀₃ , ...,w _0n ], when the weight corresponding to each stock in the initial portfolio is inconsistent with W=[w ₁ , w ₂ ,...,w _n ], the weights of the stocks whose initial portfolio needs to be adjusted are adjusted accordingly. , get the first portfolio. It should be noted that the main purpose of the above steps is to control the income unchanged and to spread the risk as much as possible, that is, the first investment portfolio controls the portfolio risk to the minimum while maintaining the previous yield.

The frequency of initial optimization for the initial portfolio awakening is not only optimized by week, but in other embodiments, it can also be optimized monthly. For example, the objective function W = [w ₁ , w ₂ , ..., w _n ] is calculated monthly to determine the best investment portfolio for the next period.

Understandably, there are two investors' decision-making goals: the lowest possible uncertainty risk and the highest possible rate of return, but not the pursuit of the lowest risk or the pursuit of high returns. The best goal should be The best balance between these two mutually constrained goals.

After determining the minimum risk of the first investment portfolio through the above steps, it is necessary to further consider the income of the first investment portfolio, that is, to convert the risk into profit, and pursue excess returns. For example, the predicted future return rate is higher in the first investment portfolio. The individual stocks adopt an increase in holding operation and take a reduction operation for individual stocks with lower forecasting future yields. Then, it is necessary to predict the future rate of return corresponding to each stock in the first portfolio.

In this embodiment, the historical yield data of each stock in the second preset time (for example, the first 30 trading days) is used to predict the future profit rate corresponding to each stock, in view of the long-short-term memory network (Long Short-Term) Memory, LSTM) is a time recurrent neural network that is more suitable for processing and predicting important events with relatively long intervals and delays in time series. LSTM has been proven to be an effective technique for solving long-order dependent problems. Therefore, the LSTM is selected to predict the future rate of return corresponding to each stock in the initially optimized first portfolio.

Before predicting the future rate of return of each stock in the first portfolio, the LSTM needs to be trained, including the following steps:

Construct sample data: obtain the characteristic data of each stock in the first investment portfolio for each trading day in the third preset time (the past two years), expand the sample data in the future, and randomly collect the arbitrarily 30 days of characteristic data and The corresponding logarithmic rate of return for the next 5 trading days is used as a sample set, and the first proportion (for example, 80%) of the sample data is extracted from the sample set as the training set, and the second ratio (for example, 20%) of the sample data is used as the verification. Set; and

Model training: The training set is used to supervise the LSTM model to obtain the future rate prediction model. The verification set is used to verify the future rate prediction model until the condition is met. For example, the model prediction accuracy reaches 90%.

Repeat the above steps to fit a future yield prediction model for each stock. After the training, each stock in the first portfolio corresponds to one model, namely Model=[Model1, Model2,...,ModelN].

When it is required to predict the future profit rate corresponding to the i-th stock in the first investment portfolio, firstly, the feature data of the stock in the second preset time (for example, the first 30 trading days) is obtained, and the feature data has 5 ,['close', 'open', 'high', 'low', 'volume'], construct the feature vector of 30*5 corresponding to the i-th stock; call the future yield prediction corresponding to the i-th stock The model uses the feature vector as the input to input the model corresponding to the i-th stock, and predicts the logarithmic rate of return for the next five trading days. The model Output is the daily logarithmic yield of the next five trading days.

After determining the logarithmic rate of return for each of the stocks in the first optimized investment portfolio for the next 5 trading days, in order to maximize the benefits, the total logarithm of each stock in the next 5 trading days is required. The rate of return is adjusted.

Specifically, the step of adjusting the weight corresponding to each stock in the first investment portfolio according to the preset second analysis rule includes:

Calculating the future comprehensive logarithmic rate of return for each stock separately, sorting the stocks in the first portfolio according to the order of the future comprehensive logarithm of each stock; in view of the output of the future yield prediction model For the logarithmic rate of return of each stock in the next 5 trading days, therefore, the logarithmic rate of return for each stock in the first portfolio for the next 5 trading days will be added as the future comprehensive logarithm of each stock. Yield, and sort the stocks in the first portfolio according to the order of future comprehensive logarithmic returns.

For the first type of stocks whose future comprehensive logarithmic rate of return is greater than or equal to the first preset threshold, the weight corresponding to the stocks is raised; for example, for stocks with a future comprehensive logarithm of return greater than or equal to 10%, indicating future gains It is larger, so it can be overweighted, assuming that its weight in the first portfolio is a, and the weight corresponding to the stock is adjusted to (1+50%)a.

For the second type of stocks whose future comprehensive logarithmic rate of return is less than or equal to the second predetermined threshold, the weight corresponding to the stocks is adjusted to a third preset threshold; for example, for a future comprehensive logarithmic yield of less than -10% Stocks, indicating a large decline in the future, it can be reduced or even short-selling operations, assuming that its weight in the first portfolio is a, adjust the corresponding weight of the stock to (1-80%) a or directly The corresponding weight is adjusted to 0.

For the third type of stocks with a predetermined proportion, the corresponding weights of the stocks are lowered; for example, 20 stocks in the first portfolio are sorted into the next 20% of the 4 stocks for the future comprehensive rate of return. It shows that its future income in the first investment portfolio is relatively low, so the stock can be reduced. Assuming the original proportion is a, adjust the proportion of the stock to (1-50%) a.

After the above steps, after determining the stocks in the first portfolio that should be adjusted in weight, the weights corresponding to the stocks in the first portfolio are adjusted to determine the weights corresponding to the stocks in the second portfolio W ^* = [w ₁ ^* , w ₂ ^* , w ₃ ^* , ..., w _n ^* ], that is, determine the target portfolio.

The electronic device 1 proposed in the above embodiment optimizes the initial investment portfolio by minimizing the correlation between the individual stocks in the initial investment portfolio while maintaining the same income, and obtains the first investment with the least risk. In addition, the LSTM model predicts the future logarithmic yield of each stock in the first portfolio, and adjusts the weight of stocks with relatively higher or lower yields to obtain a second portfolio, to a certain extent. Increased investment income.

Alternatively, in other embodiments, portfolio optimization program 10 may also be partitioned into one or more modules, one or more modules being stored in memory 11 and executed by one or more processors 12 To complete this application. A module as referred to in this application refers to a series of computer program instructions that are capable of performing a particular function. Referring to FIG. 2, it is a program module diagram of the portfolio optimization program in FIG. In this embodiment, the portfolio optimization program 10 can be divided into: a receiving module 110, a first optimization module 120, and a second optimization module 130. The functions or operational steps implemented by the modules 110-130 are similar to the above, and are not described in detail herein, by way of example, for example:

The receiving module 110 is configured to receive the input stocks in the initial portfolio to be optimized, the initial weights corresponding to the stocks in the initial investment portfolio, obtain the market corresponding to each stock, and obtain the stocks in the initial investment portfolio at the first preset time. Historical data within;

a first optimization module 120, configured to calculate, according to the historical data and a preset first analysis rule, when the correlation between each stock in the initial investment portfolio is minimum, each stock in the initial investment portfolio should respond The weights should be adjusted, and the weights corresponding to the stocks in the initial portfolio should be adjusted to determine the first optimized investment portfolio; and

The second optimization module 130 is configured to acquire historical data of each stock in the first investment portfolio in a second preset time, construct a feature vector of each stock, and input a pre-trained future profit rate corresponding to each stock. In the prediction model, predicting a future logarithmic yield of each stock in the first portfolio, and adjusting a weight corresponding to each stock in the first portfolio according to a preset second analysis rule to determine a second portfolio.

In addition, the present application also provides a portfolio optimization method. Referring to FIG. 3, it is a flowchart of a preferred embodiment of the portfolio optimization method of the present application. The method can be performed by a device that can be implemented by software and/or hardware.

In this embodiment, the portfolio optimization method includes: steps S1-S4.

Step S1, receiving the initial weights of the stocks in the initial portfolio to be optimized and the stocks in the initial portfolio, and obtaining the market corresponding to each stock and the history of each stock in the initial investment portfolio in the first preset time. data;

Step S2: calculating, according to the historical data and the preset first analysis rule, a weight corresponding to each stock in the initial investment portfolio when the correlation between the stocks in the initial investment portfolio is minimum, and Adjusting the weights corresponding to the stocks in the initial portfolio to determine the first optimized investment portfolio; and

Step S3: acquiring historical data of each stock in the first investment portfolio in a second preset time, constructing a feature vector of each stock, and inputting a pre-trained future profit rate prediction model corresponding to each stock, and predicting The future logarithmic yield of each stock in the first investment portfolio is adjusted according to a preset second analysis rule, and the weight corresponding to each stock in the first investment portfolio is adjusted to determine a second investment portfolio.

RS _it =lnP _it -lnP _i(t-1)

RM _t =lnPM _t -lnPM _t-1

RM _t =α _i +β _i *RS _it +e _it

X=e _it =[e _1i ,e _2i ,e _3i ,...,e _Ti ]

Cov(X,Y)=E[(X-E[X])(Y-E[Y])]=E[XY]-E[X]E[Y]

Min W ^T ∑W

St∑ _i w _0i *RS _i =R

e ^T W=1

The above embodiment proposes a portfolio optimization method, which minimizes the correlation between individual stocks in the initial investment portfolio while keeping the returns unchanged, and initially optimizes the initial investment portfolio to obtain the first investment with the least risk. In addition, the LSTM model predicts the future logarithmic yield of each stock in the first portfolio, and adjusts the weight of stocks with relatively higher or lower yields to obtain a second portfolio, to a certain extent. Increased investment income.

In addition, the embodiment of the present application further provides a computer readable storage medium, where the portfolio optimization program is stored, and when the portfolio optimization program is executed by the processor, the following operations are implemented:

The specific implementation manner of the computer readable storage medium of the present application is substantially the same as the specific implementation manner of the above-mentioned portfolio optimization method, and details are not described herein again.

It is to be understood that the term "comprises", "comprising", or any other variants thereof, is intended to encompass a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a series of elements includes those elements. It also includes other elements not explicitly listed, or elements that are inherent to such a process, device, item, or method. An element that is defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, the device, the item, or the method that comprises the element.

The serial numbers of the embodiments of the present application are merely for the description, and do not represent the advantages and disadvantages of the embodiments. Through the description of the above embodiments, those skilled in the art can clearly understand that the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better. Implementation. Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM as described above). , a disk, an optical disk, including a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the methods described in the various embodiments of the present application.

The above is only a preferred embodiment of the present application, and is not intended to limit the scope of the patent application, and the equivalent structure or equivalent process transformations made by the specification and the drawings of the present application, or directly or indirectly applied to other related technical fields. The same is included in the scope of patent protection of this application.

Claims

A portfolio optimization method for an electronic device, the method comprising:

Receiving the initial weights of the stocks in the initial portfolio to be optimized and the stocks in the initial portfolio, and obtaining the historical data of the market corresponding to each stock and the stocks in the initial investment portfolio in the first preset time;

Calculating, according to the historical data and the preset first analysis rule, a weight corresponding to each stock when the correlation between the stocks in the initial investment portfolio is minimum, and corresponding to each stock in the initial investment portfolio The weights are adjusted to determine the first optimized investment portfolio; and

Obtaining historical data of each stock in the first investment portfolio in a second preset time, constructing a feature vector of each stock, and inputting a pre-trained future profit rate prediction model corresponding to each stock, predicting the first The future logarithmic yield of each stock in a portfolio is adjusted according to a preset second analysis rule for the weight corresponding to each stock in the first portfolio, and the second portfolio is determined.
The portfolio optimization method according to claim 1, wherein said "based on said historical data and a preset first analysis rule, calculating a correlation between each stock in said initial portfolio In the hour, the steps in which the stocks in the initial portfolio should correspond to the weights include:

Calculating a logarithmic rate of return sequence corresponding to each stock in the market and the initial investment portfolio in a first preset time period;

Regressing the logarithmic income sequence corresponding to each stock in the initial portfolio with the logarithmic yield sequence of the market, respectively calculating a residual sequence corresponding to each stock in the initial portfolio, and obtaining the initial The covariance matrix corresponding to the portfolio; and

Calculating a minimum solution of the covariance weighted sum between the stocks in the covariance matrix corresponding to the initial portfolio, and determining a weight corresponding to each stock in the initial portfolio according to the calculation result.
The portfolio optimization method according to claim 2, wherein the step of "adjusting the weight corresponding to each stock in the first portfolio according to a preset second analysis rule" comprises:

Calculating the future comprehensive logarithmic rate of return corresponding to each stock, and sorting the stocks in the first portfolio according to the order of the future comprehensive logarithmic returns of each stock;

For the first type of stocks whose future comprehensive logarithm rate of return is greater than or equal to the first preset threshold, the weight corresponding to the stocks is raised;

For the second type of stocks whose future comprehensive logarithmic rate of return is less than or equal to the second predetermined threshold, the weight corresponding to the stocks is adjusted to a third preset threshold;

For the third type of stocks that are sorted by the preset ratio, the weights corresponding to the stocks are lowered.
The portfolio optimization method according to any one of claims 1 to 3, wherein the pre-trained future rate of return prediction model is a Long Short-Term Memory (LSTM).
The portfolio optimization method according to claim 4, wherein the calculation formula of the logarithmic rate of return corresponding to the market is:

RM t =lnPM t -lnPM t-1

Where RM t represents the logarithmic rate of return of the market at time t in the logarithmic rate of return corresponding to the market, PM t represents the closing price of the market at time t, and t represents a certain time before the optimization of the initial investment portfolio.
The portfolio optimization method according to claim 5, wherein the calculation formula of the logarithmic rate of return sequence corresponding to each stock in the initial investment portfolio is:

RS it =lnP it -lnP i(t-1)

Where RS it represents the logarithmic rate of return of the stock at the t-th time in the logarithmic rate of return corresponding to the i-th stock in the initial portfolio, Pit represents the closing price of the i-th stock at time t, and t represents Optimize a moment before the initial portfolio.
The portfolio optimization method according to claim 6, wherein the calculation formula of the residual sequence corresponding to each stock is:

e it =RM t -(α i +β i *RS it )

Where RM t represents the logarithmic rate of return of the market at the t-th time in the logarithmic rate of return sequence corresponding to the market, and RS it represents the stock at the t-th time in the logarithmic rate of return sequence corresponding to the i-th stock in the initial portfolio Logarithmic rate of return, α = [α 1 , α 2 , α 3 , ..., α n ], β = [β 1 , β 2 , β 3 , ..., β n ], e it represents the initial portfolio The i-th stock corresponds to the residual sequence, e it =[e 1i , e 2i , e 3i ,..., e Ti ].
The portfolio optimization method according to claim 7, wherein the calculation formula of the covariance weighted sum minimum between the stocks in the covariance matrix corresponding to the initial portfolio is:

Min W T ∑W

St∑ i w 0i *RS i =R

e T W=1

Where W is to be solved, W=[w 1 ,w 2 ,...,w n ], indicating the weight corresponding to each stock, ∑ represents the covariance matrix corresponding to the initial portfolio, and RS i represents the optimized portfolio The logarithmic rate of return corresponding to each stock in the previous moment, w 0i represents the initial weight corresponding to each stock in the initial portfolio, and RS represents the total logarithmic yield corresponding to the initial portfolio of the optimized investment portfolio, e T W=1 Indicates that the sum of the weights corresponding to each stock in W is 1.
An electronic device, comprising: a memory, a processor, and a portfolio optimization program stored on the memory, where the portfolio optimization program is executed by the processor to implement the following steps:

Receiving the initial weights of the stocks in the initial portfolio to be optimized and the stocks in the initial portfolio, and obtaining the historical data of the market corresponding to each stock and the stocks in the initial investment portfolio in the first preset time;

Calculating, according to the historical data and the preset first analysis rule, a weight corresponding to each stock when the correlation between the stocks in the initial investment portfolio is minimum, and corresponding to each stock in the initial investment portfolio The weights are adjusted to determine the first optimized investment portfolio; and

Obtaining historical data of each stock in the first investment portfolio in a second preset time, constructing a feature vector of each stock, and inputting a pre-trained future profit rate prediction model corresponding to each stock, predicting the first The future logarithmic yield of each stock in a portfolio is adjusted according to a preset second analysis rule for the weight corresponding to each stock in the first portfolio, and the second portfolio is determined.
The electronic device according to claim 9, wherein said "based on said historical data and a preset first analysis rule, when the correlation between the stocks in said initial portfolio is calculated to be minimal, The steps of the weights of the stocks in the initial portfolio should correspond to:

Calculating a logarithmic rate of return sequence corresponding to each stock in the market and the initial investment portfolio in a first preset time period;

Regressing the logarithmic income sequence corresponding to each stock in the initial portfolio with the logarithmic yield sequence of the market, respectively calculating a residual sequence corresponding to each stock in the initial portfolio, and obtaining the initial The covariance matrix corresponding to the portfolio; and

Calculating a minimum solution of the covariance weighted sum between the stocks in the covariance matrix corresponding to the initial portfolio, and determining a weight corresponding to each stock in the initial portfolio according to the calculation result.
The electronic device according to claim 10, wherein the step of "adjusting the weight corresponding to each stock in the first portfolio according to a preset second analysis rule" comprises:

Calculating the future comprehensive logarithmic rate of return corresponding to each stock, and sorting the stocks in the first portfolio according to the order of the future comprehensive logarithmic returns of each stock;

For the first type of stocks whose future comprehensive logarithmic rate of return is greater than or equal to the first preset threshold, the weight corresponding to the stocks is raised;

For the second type of stocks whose future comprehensive logarithmic rate of return is less than or equal to the second predetermined threshold, the weight corresponding to the stocks is adjusted to a third preset threshold;

For the third type of stocks that are sorted by the preset ratio, the weights corresponding to the stocks are lowered.
The electronic device according to any one of claims 9 to 11, wherein the pre-trained future rate of return prediction model is a Long Short-Term Memory (LSTM).
The electronic device according to claim 12, wherein the calculation formula of the logarithmic rate of return corresponding to the market is:

RM t =lnPM t -lnPM t-1

Where RM t represents the logarithmic rate of return of the market at time t in the logarithmic rate of return corresponding to the market, PM t represents the closing price of the market at time t, and t represents a certain time before the optimization of the initial investment portfolio.
The electronic device according to claim 13, wherein the calculation formula of the logarithmic rate of return corresponding to each stock in the initial portfolio is:

RS it =lnP it -lnP i(t-1)

Where RS it represents the logarithmic rate of return of the stock at the t-th time in the logarithmic rate of return corresponding to the i-th stock in the initial portfolio, Pit represents the closing price of the i-th stock at time t, and t represents Optimize a moment before the initial portfolio.
The electronic device according to claim 14, wherein the calculation formula of the residual sequence corresponding to each stock is:

e it =RM t -(α i +β i *RS it )

Where RM t represents the logarithmic rate of return of the market at the t-th time in the logarithmic rate of return sequence corresponding to the market, and RS it represents the stock at the t-th time in the logarithmic rate of return sequence corresponding to the i-th stock in the initial portfolio Logarithmic rate of return, α = [α 1 , α 2 , α 3 , ..., α n ], β = [β 1 , β 2 , β 3 , ..., β n ], e it represents the initial portfolio The i-th stock corresponds to the residual sequence, e it =[e 1i , e 2i , e 3i ,..., e Ti ].
The electronic device according to claim 15, wherein the formula for calculating the solution of the covariance weighted sum between the stocks in the covariance matrix corresponding to the initial portfolio is:

Min W T ∑W

St∑ i w 0i *RS i =R

e T W=1

Where W is to be solved, W=[w 1 ,w 2 ,...,w n ], indicating the weight corresponding to each stock, ∑ represents the covariance matrix corresponding to the initial portfolio, and RS i represents the optimized portfolio The logarithmic rate of return corresponding to each stock in the previous moment, w 0i represents the initial weight corresponding to each stock in the initial portfolio, and RS represents the total logarithmic yield corresponding to the initial portfolio of the optimized investment portfolio, e T W=1 Indicates that the sum of the weights corresponding to each stock in W is 1.
A computer readable storage medium, wherein the computer readable storage medium stores a portfolio optimization program, and when the portfolio optimization program is executed by the processor, the following steps are implemented:

Receiving the initial weights of the stocks in the initial portfolio to be optimized and the stocks in the initial portfolio, and obtaining the historical data of the market corresponding to each stock and the stocks in the initial investment portfolio in the first preset time;

Calculating, according to the historical data and the preset first analysis rule, a weight corresponding to each stock when the correlation between the stocks in the initial investment portfolio is minimum, and corresponding to each stock in the initial investment portfolio The weights are adjusted to determine the first optimized investment portfolio; and

Obtaining historical data of each stock in the first investment portfolio in a second preset time, constructing a feature vector of each stock, and inputting a pre-trained future profit rate prediction model corresponding to each stock, predicting the first The future logarithmic yield of each stock in a portfolio is adjusted according to a preset second analysis rule for the weight corresponding to each stock in the first portfolio, and the second portfolio is determined.
The computer readable storage medium according to claim 17, wherein said "according to said historical data and a preset first analysis rule, calculating a correlation between each stock in said initial portfolio At the very least, the steps in which the stocks in the initial portfolio should correspond to the weights include:

Calculating a logarithmic rate of return sequence corresponding to each stock in the market and the initial investment portfolio in a first preset time period;

Regressing the logarithmic income sequence corresponding to each stock in the initial portfolio with the logarithmic yield sequence of the market, respectively calculating a residual sequence corresponding to each stock in the initial portfolio, and obtaining the initial The covariance matrix corresponding to the portfolio; and

Calculating a minimum solution of the covariance weighted sum between the stocks in the covariance matrix corresponding to the initial portfolio, and determining a weight corresponding to each stock in the initial portfolio according to the calculation result.
The computer readable storage medium according to claim 18, wherein the step of "adjusting the weight corresponding to each stock in the first portfolio according to a preset second analysis rule" comprises:

Calculating the future comprehensive logarithmic rate of return corresponding to each stock, and sorting the stocks in the first portfolio according to the order of the future comprehensive logarithmic returns of each stock;

For the first type of stocks whose future comprehensive logarithmic rate of return is greater than or equal to the first preset threshold, the weight corresponding to the stocks is raised;

For the second type of stocks whose future comprehensive logarithmic rate of return is less than or equal to the second predetermined threshold, the weight corresponding to the stocks is adjusted to a third preset threshold;

For the third type of stocks that are sorted by the preset ratio, the weights corresponding to the stocks are lowered.
The computer readable storage medium according to any one of claims 17 to 19, wherein the pre-trained future rate of return prediction model is a Long Short-Term Memory (LSTM).