Investment portfolio configuration method and device, equipment and storage medium

Abstract

The invention relates to the technical field of financial science and technology, in particular to an investment portfolio configuration method and device, equipment and a storage medium, and the method comprises the steps: carrying out the normal transformation of obtained index data, obtaining normal adjusted data, and calculating a category factor based on the normal adjusted data; calculating a regression coefficient estimation value based on the category factor, the relative value index market value data corresponding to the index data and the total asset data; calculating a high-decline return index based on the regression coefficient estimation value, and determining a return value coefficient based on the high-decline return index; and determining an asset weight based on the return value coefficient, and configuring an investment portfolio based on the asset weight and a preset investment strategy. The configuration requirement of investors for investment portfolios can be met conveniently.

Description

Technical Field

[0001] This application relates to the technical field of fintech, and particularly to a method, device, equipment and storage medium for portfolio allocation. Background Art

[0002] There are various types of assets available for investors to invest in the market, such as stocks, bonds, funds, real estate, etc. Allocating multiple assets can optimize the risk-return characteristics of the portfolio according to the investor's risk tolerance and return target, so as to achieve the expected return target. Conservative investors may allocate a relatively large proportion of funds to fixed-income assets such as bonds to obtain stable returns; while aggressive investors may increase the allocation ratio of equity assets such as stocks to pursue higher returns. Through reasonable portfolio allocation, investors can pursue a suitable return level on the premise of controllable risks.

[0003] In addition, the performances of different assets vary in different market environments. Through portfolio allocation, dispersing funds to multiple assets can reduce the impact of the volatility of a single asset on the overall assets. Even if a certain type of asset performs poorly, the performances of other assets may make up for part of the losses, thereby reducing the overall risk of the portfolio. Therefore, currently, investors have a strong demand for portfolio allocation. Summary of the Invention

[0004] To facilitate meeting the investors' allocation requirements for portfolios, this application provides a method, device, equipment and storage medium for portfolio allocation.

[0005] In a first aspect, this application provides a method for portfolio allocation, including:

[0006] Performing a normal transformation on the obtained indicator data to obtain normally adjusted data, and calculating a category factor based on the normally adjusted data;

[0007] Calculating an estimated regression coefficient based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data;

[0008] Calculating a high-low reduction return indicator based on the estimated regression coefficient, and determining a return value coefficient based on the high-low reduction return indicator;

[0009] Determining an asset weight based on the return value coefficient, and allocating a portfolio based on the asset weight and a preset investment strategy.

[0010] In a second aspect, this application provides a device for portfolio allocation, including:

[0011] A factor calculation module, configured to perform a normal transformation on the obtained indicator data to obtain normally adjusted data, and calculate a category factor based on the normally adjusted data;

[0012] A coefficient estimation module, configured to calculate an estimated regression coefficient based on the category factor, the market value data of the relative value index corresponding to the index data, and the total asset data;

[0013] A coefficient determination module, configured to calculate a high-low reduction return index based on the estimated regression coefficient, and determine a return value coefficient based on the high-low reduction return index;

[0014] A portfolio configuration module, configured to determine asset weights based on the return value coefficient, and configure an investment portfolio based on the asset weights and a preset investment strategy.

[0015] In a third aspect, the present application provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in the above method are implemented.

[0016] In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method are implemented.

[0017] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0018] In the above investment portfolio configuration method, device, equipment and storage medium, the obtained index data is subjected to a normal transformation to obtain normally adjusted data, and a category factor is calculated based on the normally adjusted data; an estimated regression coefficient is calculated based on the category factor, the market value data of the relative value index corresponding to the index data, and the total asset data; a high-low reduction return index is calculated based on the estimated regression coefficient, and a return value coefficient is determined based on the high-low reduction return index; asset weights are determined based on the return value coefficient, and an investment portfolio is configured based on the asset weights and a preset investment strategy. Through the above implementation, the skewness of all variables is improved by using the normal transformation, and at the same time, the two-way clustering adjustment makes the regression result of the model robust and significant. Then, the regression coefficient estimate is calculated and the return value coefficient is calculated to regress the excess return of each enterprise, so as to facilitate the calculation of the asset weights corresponding to the assets and realize the construction of the investment portfolio, thereby facilitating the satisfaction of the investor's configuration requirements for the investment portfolio.

[0019] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. Description of the Drawings

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the embodiments will be briefly introduced below. It should be understood that the following accompanying drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related accompanying drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a flowchart of a portfolio allocation method provided in an embodiment of the present application;

[0022] Figure 2 It is a schematic structural diagram of a portfolio allocation device provided in an embodiment of the present application;

[0023] Figure 3 It is a schematic structural diagram of a computer device provided in an embodiment of the present application;

[0024] Figure 4 It is an internal structural diagram of a computer-readable storage medium provided in an embodiment of the present application. Specific Embodiments

[0025] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present disclosure, and are not used to limit the present disclosure.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of this article and the above accompanying drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present article described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or equipment.

[0027] In this article, the term "and / or" is only a relationship describing associated objects, indicating that three relationships can exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0028] Embodiment 1

[0029] Figure 1The flowchart of a portfolio allocation method provided in the first embodiment of this application is referred to Figure 1 , and this method can be executed by a device that executes this method. The device can be implemented in software and / or hardware. The method includes:

[0030] S110. Perform a normal transformation on the obtained index data to obtain the normally adjusted data, and calculate the category factor based on the normally adjusted data.

[0031] Among them, the index data is a variety of enterprise index data of different years of each enterprise that an investor may invest in. The variety of enterprise index data includes: financial index data, market performance data, and other index data; among them, the financial index data includes: earnings per share (EPS), price-earnings ratio (PE), price-to-book ratio (PB), return on equity (ROE), etc., the market performance data includes: stock price, trading volume, turnover rate, etc., and other index data includes: dividend rate and beta coefficient, etc.

[0032] Among them, earnings per share (EPS) is the ratio of a company's net profit to the total number of outstanding common shares, reflecting the net profit or net loss that each common stockholder can enjoy or bear per share, and is an important indicator to measure a company's profitability. The price-earnings ratio (PE) is the ratio of the stock market price to earnings per share, which reflects the price that an investor is willing to pay to obtain each unit of profit of the enterprise. The price-to-book ratio (PB) is the ratio of the stock market price to the net asset per share, which is used to measure the premium degree of the enterprise's stock price relative to its net assets. The calculation formula is: price-to-book ratio = stock market price / net asset per share. Return on equity (ROE) is the percentage of net profit to average net assets, reflecting the efficiency of the enterprise in using its own capital. The stock price is the trading price of the enterprise's stock in the securities market and is one of the most concerned indicators for investors. The trading volume refers to the number of stocks traded within a certain period of time, reflecting the activity of the stock trading. The turnover rate is the ratio of the trading volume of a stock to the outstanding shares within a certain period of time, which is used to measure the liquidity of the stock. The dividend rate is the ratio of the dividend distributed by the enterprise according to the shareholding ratio of the shareholders to the stock price within a specific period of time, reflecting the level of cash return obtained by the shareholders from the enterprise. The calculation formula is: dividend rate = (dividend per share / stock price) × 100%. The beta coefficient is an indicator that measures the volatility of a security or portfolio relative to the overall market. A beta coefficient greater than 1 indicates that the price fluctuation range of the security or portfolio is greater than the average market fluctuation range; a beta coefficient less than 1 indicates that its price fluctuation range is less than the average market fluctuation range.

[0033] It should be noted that all the various indicator data obtained above may have skewness. Among them, skewness means that the data distribution of the data presents an asymmetric form. If the tail of the data distribution is longer on the right side (i.e., the side of larger values), it is called right skewness (also called positive skewness), and at this time the mean is greater than the median; if the tail of the data distribution is longer on the left side (i.e., the side of smaller values), it is called left skewness (also called negative skewness); if the indicator data has skewness, it will reduce the accuracy of the subsequent calculated category factors and regression coefficient estimates. Therefore, it is necessary to perform a normal transformation on the indicator data. The normal transformation is used to process the indicator data to improve the skewness of the indicator data, and the data obtained after the indicator data is normally transformed is recorded as the normally adjusted data.

[0034] The weights of each type of indicator data can be determined through the normally adjusted data, and classification can be carried out according to the types of indicator data corresponding to each weight. Then, by performing corresponding calculations on the weights of the same category, the corresponding category factors can be obtained.

[0035] S120. Calculate the regression coefficient estimate based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data.

[0036] Among them, each enterprise has corresponding indicator data, and the enterprise corresponding to the indicator data has corresponding relative value indicator market value data MV and total asset data TA. Among them, the relative value indicator market value data MV is the ratio data obtained by comparing the market value of the enterprise with other relevant financial indicators or market data, and is used to evaluate the relative value of the enterprise in the market. By taking the category factor as the independent variable and MV / TA as the dependent variable, a factor model for calculating the subsequent regression coefficient estimate can be constructed. This factor model can process the category factor to calculate the regression coefficient estimate, and this regression coefficient estimate is used for cross-sectional regression of the excess return of the enterprise.

[0037] S130. Calculate the high-low return indicator based on the regression coefficient estimate, and determine the return value coefficient based on the high-low return indicator.

[0038] Among them, the high-low return indicator can be obtained by processing the regression coefficient estimate. The high-low return indicator represents the return difference between the portfolios of companies with high and low exposure to a given factor, captures the pricing premium related to specific risks, and facilitates ensuring that the analysis focuses on the factors that have the most significant impact on returns. The return value coefficient can be obtained by performing corresponding processing on the high-low return indicator, and the return value coefficient is used to determine the weights of the various assets that the investor wants to invest in.

[0039] S140. Determine the asset weights based on the return value coefficient, and allocate the investment portfolio based on the asset weights and the preset investment strategy.

[0040] Among them, the asset weight ω i In this embodiment, three calculation methods are provided:

[0041] First, Risk Neutral, that is i refers to the i-th enterprise; n represents the number of enterprises; D Buy represents the set of enterprises that can be bought (i.e., undervalued); D Sell represents the set of enterprises that can be sold (i.e., overvalued). This method gives equal weight to each asset, regardless of their risk or expected return differences.

[0042] Second, Risk Parity: that is σ i = std(θ i,1 , θ i,2 ,..., θ i,k ), θ i,k is the return value coefficient. To handle the case where σi = 0, the zero value can be replaced by a small constant ∈. Further, This method ensures an even distribution of risk across all assets in the portfolio.

[0043] Third, Factor-Weighted, that is, first calculate the scores μ k and σ k are the mean and standard deviation of the k-th high-low return index across all assets in D buy / sell respectively. Further, For the buying set D Buy , the sum of its asset weights is 1; for the selling set D Sell , the sum of its asset weights is -1; the core idea of this method is to calculate a comprehensive score based on the standardized exposure of assets to multiple factors, exclude assets with negative scores, and allocate portfolio weights according to the proportion of the scores of the remaining assets, ensuring that the weights reflect their advantages in multiple factors.

[0044] Among them, the asset weight is the weight of each asset that the investor intends to invest in; in this embodiment, the preset investment strategies include two types: long strategy and long-short strategy. Among them, the long strategy is that the investor expects the market or the price of a certain asset to rise, so the investor buys the asset and sells it after the price rises to earn the price difference. This is an investment strategy based on an optimistic expectation of the market or asset prospects; the long-short strategy is a more complex investment strategy that simultaneously involves long (buy) and short (sell) operations on different assets in the market. Through the long-short combination of different assets, it attempts to obtain returns in various market environments, reduce the risk of a single asset, and achieve relatively stable investment returns.

[0045] It should be noted that multiple sets of asset weights can be calculated through the return value coefficient. Each set of multiple asset weights corresponds to the weights of different assets in a group of assets. By combining multiple sets of asset weights with two preset investment strategies, the combined allocation of asset weights and investment strategies can be achieved, thereby obtaining the corresponding investment portfolio.

[0046] It should be noted that for the above investment portfolio allocation method, device, equipment and storage medium in this embodiment, the obtained index data is subjected to normal transformation to obtain normally adjusted data, and the category factor is calculated based on the normally adjusted data; based on the category factor, the relative value index market value data corresponding to the index data and the total asset data, the regression coefficient estimate value is calculated; based on the regression coefficient estimate value, the high-low reduction return index is calculated, and the return value coefficient is determined based on the high-low reduction return index; based on the return value coefficient, the asset weights are determined, and the investment portfolio is configured based on the asset weights and the preset investment strategy. Through the above implementation, the skewness of all variables is improved by using normal transformation, and at the same time, the two-way clustering adjustment makes the regression results of the model robust and significant. Then, the regression coefficient estimate is calculated and the return value coefficient is calculated to regress the excess return of each enterprise, so as to facilitate the calculation of the asset weights corresponding to the assets and realize the construction of the investment portfolio, thereby facilitating the satisfaction of the investor's allocation requirements for the investment portfolio.

[0047] Embodiment 2

[0048] An investment portfolio allocation method provided in Embodiment 2 of the present application optimizes the "performing normal transformation on the obtained index data to obtain normally adjusted data" in Embodiment 1; it should be noted that for parts not described in detail in this embodiment, reference can be made to the descriptions of other embodiments. The method includes:

[0049] S211. Perform linear interpolation on the obtained index data to obtain initial frequency modulation data.

[0050] Among them, the initially obtained index data has its corresponding acquisition frequency, but the index data of this acquisition frequency may not be suitable for application in this method. Exemplarily, the acquisition frequency of the initially obtained index data is times / year, but the acquisition frequency of the index data required by this method is times / month. Therefore, it is necessary to perform data frequency modulation on the initially obtained index data. The data frequency modulation method adopted in this embodiment is linear interpolation. Linear interpolation is to estimate the intermediate value between two data in the index data and insert this intermediate value between the above two data. Exemplarily, given two data points (x1, y1) and (x2, y2), it is required to estimate the y value corresponding to an x between x1 and x2. Linear interpolation is based on the assumption that between these two data points, y has a linear relationship with x, that is, a straight line can be used to connect these two points, and then the value of the intermediate point is estimated according to this straight line.

[0051] S212. Perform seasonal decomposition on the initial frequency modulation data to obtain target frequency modulation data.

[0052] Among them, in order to facilitate better understanding and prediction of the seasonal patterns in the initial frequency modulation data, in this embodiment, seasonal decomposition is also performed on the initial frequency modulation data. Seasonal decomposition is an analysis method that separates the seasonal components in time series data (initial frequency modulation data). The initial frequency modulation data often exhibits seasonal fluctuations. Through seasonal decomposition, the seasonal factors in the initial frequency modulation data can be separated from other trends, cycles, and random factors, so as to more clearly observe and analyze the different components of the initial frequency modulation data, which helps to more accurately predict the future data trend.

[0053] Specifically, the methods for performing seasonal decomposition on the initial frequency modulation data in this embodiment include the moving average method, the exponential smoothing method, etc. Taking the moving average method as an example, first, select a suitable moving average period, which is usually related to the seasonal period of the initial frequency modulation data. For example, for monthly data, if there is an annual seasonality, 12 months can be selected as the moving average period. Then, perform moving average calculation on the initial frequency modulation data to obtain a sequence of moving average values. This sequence eliminates part of the random fluctuations and can better reflect the trend and seasonality of the data. Next, divide the initial frequency modulation data by the moving average value to obtain a sequence of seasonal factors. This sequence reflects the fluctuations of the data relative to the average level in each season. Finally, through the analysis and adjustment of the sequence of seasonal factors, a more accurate seasonal pattern can be obtained, and it can be separated from the initial frequency modulation data to obtain the trend data after removing seasonality; and the data obtained after performing seasonal decomposition on the initial frequency modulation data is denoted as target frequency modulation data.

[0054] S213. Process the target frequency modulation data based on a preset normal transformation algorithm to obtain post-normal transformation data.

[0055] Among them, the normal transformation algorithm is used to process the target frequency modulation data to improve the skewness of the target frequency modulation data. The normal transformation algorithm adopted in this embodiment is the Yeo-Johnson transformation algorithm. In other embodiments, the Box-Cox transformation algorithm, the logarithmic transformation algorithm, etc. can also be used, and it is not specifically limited; the post-normal transformation data is the data obtained after processing the target frequency modulation data by the normal transformation algorithm.

[0056] It should be noted that the Yeo-Johnson transformation algorithm adopted in this embodiment can take care of zero values and negative values compared with the logarithmic transformation algorithm.

[0057] S214. Calculate the category factor based on the post-normal transformation data.

[0058] S220. Calculate the estimated regression coefficient based on the category factor, the market value data and the total asset data of the relative value index corresponding to the index data.

[0059] S230. Calculate the high-low reduction return index based on the estimated regression coefficient, and determine the return value coefficient based on the high-low reduction return index.

[0060] S240. Determine the asset weight based on the return value coefficient, and allocate the investment portfolio based on the asset weight and the preset investment strategy.

[0061] Embodiment III

[0062] An investment portfolio allocation method provided in Embodiment III of the present application optimizes "calculating the category factor based on the normally adjusted data" in Embodiment I. It should be noted that for parts not described in detail in this embodiment, the descriptions of other embodiments can be referred to. The method includes:

[0063] S311. Perform a normal transformation on the obtained index data to obtain normally adjusted data.

[0064] S312. Preprocess the normally adjusted data to obtain target index data, and calculate the prior weight and the prior covariance matrix based on the target index data.

[0065] Among them, the preprocessing includes data standardization and extreme value processing; the normally adjusted data are enterprise index data of different types, and the enterprise index data of different types may have different dimensions. In the subsequent process of calculating the category factor, it is necessary to eliminate the adverse effects brought by data of different types of dimensions. In addition, in order to facilitate the comparison and analysis of enterprise index data of different types, it is necessary to first perform data standardization on the normally adjusted data to obtain standard index data.

[0066] It should be noted that although the standard index data is data after standardization processing, there may be some data extreme values in it. Data extreme values refer to extremely large or extremely small values in the data set that are significantly deviated from other data, and are also called outliers. Numerical extreme values will reduce the accuracy of the calculated category factor. Therefore, it is necessary to perform extreme value processing on the standard index data to obtain the target index data.

[0067] Specifically, in this embodiment, at least one of the deletion method, the winsorization method, and the logarithmic transformation method can be adopted to perform extreme value processing on the standard index data. Among them, the deletion method is to directly delete the extreme values from the data set. The winsorization method is to adjust the extreme values to the values of specific percentiles. For example, the values in the data set that are less than the 5th percentile are adjusted to the value of the 5th percentile, and the values that are greater than the 95th percentile are adjusted to the value of the 95th percentile. In this way, not only the information contained in the extreme values can be retained, but also their impact on the overall data can be reduced. The logarithmic transformation method is to compress the scale of the data. This method can relatively reduce the gap between the extreme values and other data, and alleviate the impact of the extreme values to a certain extent. For example, for the data 1, 10, 100, 1000, after logarithmic transformation, it becomes 0, 1, 2, 3, and the gap between the extreme values and other data is significantly reduced.

[0068] It should be noted that to calculate the category factor corresponding to the target index data, it is necessary to first calculate the prior weight w0 and the prior covariance matrix P of the target index data. t ;

[0069] Among them, the target index data includes multiple indicators. To calculate the category factor, it is necessary to first calculate the prior weight w0 of each indicator data in the target index, where w0 = 1 / k, and k is the number of indicator data in the target index data; the prior covariance matrix P t = ζ × I × Var(X), where ζ is the regularization coefficient. In this embodiment, ζ = 0.1, I is the k×k identity matrix, Var(X) is the variance of the target index data, and X is the indicator data in the target index data.

[0070] S313. Calculate the target index weight based on the prior weight and the prior covariance matrix.

[0071] Among them, when the prior weight w0 and the prior covariance matrix P t are obtained, the target index weight w of each indicator data in the target index data can be further calculated posterior , where w posterior =(X T X + P t ) -1 (X T y + P t w0), where y = MV / TA, MV is the market value data of the enterprise's value indicator, and TA is the total asset data of the enterprise.

[0072] S314. Calculate the category factor based on the target index weight.

[0073] Among them, the corresponding target index weight w can be calculated for each indicator data in the target index data posterior, and each index data in the target index data has its corresponding data type; Exemplarily, the data types corresponding to index data such as earnings per share (EPS), price-earnings ratio (PE), price-to-book ratio (PB), and return on equity (ROE) are financial index data, and the data types corresponding to index data such as stock price, trading volume, and turnover rate are market performance data. The data types of dividend rate and beta coefficient are other index data.

[0074] Specifically, after calculating the target index weight w corresponding to each index data posterior , further, classify the target index weight w corresponding to each index data according to the data type corresponding to each index data posterior to obtain multiple weight classification sets, and each weight classification set corresponds to a data type.

[0075] Taking one of the weight classification sets as an example, perform weighted calculation on the target index weights w therein posterior , and the result of the weighted calculation is also the category factor corresponding to this weight classification set.

[0076] S320. Calculate the regression coefficient estimate based on the category factor, the relative value index market value data corresponding to the index data, and the total asset data.

[0077] S330. Calculate the high-low reduction return index based on the regression coefficient estimate, and determine the return value coefficient based on the high-low reduction return index.

[0078] S340. Determine the asset weight based on the return value coefficient, and allocate the investment portfolio based on the asset weight and the preset investment strategy.

[0079] Embodiment Four

[0080] An investment portfolio allocation method provided in Embodiment Four of the present application optimizes the "calculating the regression coefficient estimate based on the category factor, the relative value index market value data corresponding to the index data, and the total asset data" in Embodiment One; It should be noted that for parts not detailed in this embodiment, reference can be made to the descriptions of other embodiments. The method includes:

[0081] S410. Perform normal transformation on the obtained index data to obtain the normally adjusted data, and calculate the category factor based on the normally adjusted data.

[0082] S421. Construct a relative valuation factor model based on the category factor, the relative value index market value data corresponding to the index data, and the total asset data.

[0083] Among them, taking the category factor F as the independent variable and the ratio MV / TA of the market value data MV to the total asset data TA of the relative value index as the dependent variable, a corresponding relative valuation factor model can be constructed. The expression of the relative valuation factor model is as follows:

[0084]

[0085] Among them, i represents the i-th enterprise, t represents time, β0 is the intercept, k represents the k-th category factor, β k is the factor coefficient, u i,t is the total error term;

[0086] Among them, u i,t = μ i + λ t + ε i,t ;

[0087] μ i is to capture the individual random effect, which is the characteristic of a specific enterprise and does not change with time. Assume that these effects are independently and identically distributed normal distributions, with a mean of zero and a variance of σ 2 (μ), that is, μi ~ N(0, σ 2 (μ)). They represent the unique company-specific characteristics that the explanatory variables fail to control. μi is a random variable because it captures the random variation in individual-specific characteristics (such as companies, regions, etc.). To ensure that the model estimation results are unbiased, assume that it is uncorrelated with all explanatory variables and the time random effect λ t .

[0088] λ t is the time random effect. Considering a set of time random effects λ t , it captures the unobservable, time-specific shocks that affect the dependent variable. These effects are modeled as independently and identically distributed normal random variables, with a mean of zero and a variance of σ 2 (λ), that is, λ t ~ N(0, σ 2 (λ)). λ t is a random variable because it captures the common shocks across time, such as random events like economic cycles and market environment changes. Assume that it is uncorrelated with all explanatory variables and the individual random effects to ensure that the time random effect does not interfere with the effect of the explanatory variables.

[0089] ε i,t is the random error term, which simulates all other unmodeled random noises. Under the assumption of independently and identically distributed normal residuals, with a mean of zero and a variance of σ 2 (ε), these residuals, denoted as ε i,t ~ N(0, σ 2(ε)). This is because the residuals are a random variable that contains any random variation that the model fails to capture. They are assumed to be independent of the individual effect μ i and the time effect λ t as well as all explanatory variables.

[0090] S422. Determine the model standard error data of the relative valuation factor model.

[0091] Among them, the model standard error data is the standard deviation of the sampling distribution of the factor coefficients in the relative valuation factor model, which is used to reflect the average error degree between the estimated value and the overall true value.

[0092] S423. Cluster-adjust the model standard error data based on the individual dimension and the time dimension to obtain the regression coefficient estimate.

[0093] Among them, the individual dimension is the dimension of the i-th enterprise, and the time dimension is the dimension of time t; clustering and adjusting the model standard error data from the individual dimension i and the time dimension t can achieve double clustering adjustment of the model standard error data, and the result obtained after double clustering adjustment of the model standard error data is the regression coefficient estimate.

[0094] S430. Calculate the high-low reduction return index based on the regression coefficient estimate, and determine the return value coefficient based on the high-low reduction return index.

[0095] S440. Determine the asset weights based on the return value coefficient, and allocate the investment portfolio based on the asset weights and the preset investment strategy.

[0096] Example Five

[0097] An investment portfolio allocation method provided in Example Five of the present application optimizes the "cluster-adjusting the model standard error data based on the individual dimension and the time dimension to obtain the regression coefficient estimate" in Example Four; it should be noted that for the parts not detailed in this example, the descriptions of other examples can be referred to. The method includes:

[0098] S510. Perform a normal transformation on the obtained index data to obtain the normally adjusted data, and calculate the category factor based on the normally adjusted data.

[0099] S521. Construct a relative valuation factor model based on the category factor, the market value data of the relative value index corresponding to the index data, and the total asset data.

[0100] S522. Determine the model standard error data of the relative valuation factor model.

[0101] S523A. Cluster the model standard error data based on the individual dimension to obtain an individual clustering result, and calculate the individual clustering residuals based on the individual clustering result.

[0102] Among them, the data in the model standard error data corresponds to different individuals (enterprises). By clustering the data in the model standard error data according to different individuals, an individual clustering result can be obtained. The individual clustering result includes multiple individual clusters, and the number of individual clusters is the same as the number of individuals. Taking one of the individual clusters as an example, determine the clustering center of this individual cluster, and then calculate the distance from the data points in the individual cluster to this clustering center, so as to obtain the clustering residual corresponding to this data point, and record this clustering residual as the individual clustering residual.

[0103] S523B. Cluster the model standard error data based on the time dimension to obtain a time clustering result, and calculate the time clustering residuals based on the time clustering result.

[0104] Among them, the data in the model standard error data corresponds to different times (such as years). By clustering the data in the model standard error data according to different times, a time clustering result can be obtained. The time clustering result includes multiple time clusters, and the number of time clusters is the same as the number of times. Taking one of the time clusters as an example, determine the clustering center of this time cluster, and then calculate the distance from the data points in the time cluster to this clustering center, so as to obtain the clustering residual corresponding to this data point, and record this clustering residual as the time clustering residual.

[0105] S523C. Determine the regression coefficient estimate value based on the individual clustering residuals and the time clustering residuals.

[0106] Among them, the formula for double clustering adjustment of the model standard error data is as follows:

[0107]

[0108] Among them, is the factor coefficient variance, G1 is the number of individual clusters, G2 is the number of time clusters, and N is the product of G1 and G2; g1 represents the g1-th individual cluster, and g2 represents the g2-th time cluster.

[0109] It should be noted that two-way clustering adjustment focuses on adjusting the dependence between two grouping dimensions, such as individuals and time, and can solve several key statistical problems:

[0110] First of all, it reduces clustering correlation. The observations within the same company may be correlated because of shared characteristics (such as management style or capital structure), and the observations between different companies may be affected by common factors such as macroeconomic conditions or market fluctuations.

[0111] Secondly, the heteroscedasticity problem is solved. In traditional regression models, the standard error assumes homoscedasticity, that is, the variance of the error is constant among the observations. However, in panel data, the variances among different groups (such as companies) often vary due to different characteristics. Two-way clustering solves this problem by allowing the variance to vary in two dimensions (such as companies and time), ensuring that the standard error can adapt to the group-specific variability.

[0112] Furthermore, autocorrelation is corrected. Autocorrelation occurs when the errors in different periods of a time series are correlated. In panel data, this problem may occur both when companies change over time and among different companies in the same period, driven by lag dependence or shared macroeconomic effects. Two-way clustering adjusts the standard error to account for the time-series correlation within companies and the contemporaneous correlation among different companies in the same period.

[0113] It should be noted that by performing double-clustering adjustment on the model standard error data, the factor coefficient β k can be optimized, and the optimized factor coefficient is denoted as the regression coefficient estimate; the accuracy of the factor coefficient is improved, thus facilitating the guarantee of the robustness and significance of the relative valuation factor model.

[0114] S530. Calculate the high-low reduction return index based on the regression coefficient estimate, and determine the return value coefficient based on the high-low reduction return index.

[0115] S540. Determine the asset weights based on the return value coefficient, and allocate the investment portfolio based on the asset weights and the preset investment strategy.

[0116] Embodiment Six

[0117] An investment portfolio allocation method provided in Embodiment Six of the present application optimizes the "calculating the high-low reduction return index based on the regression coefficient estimate" in Embodiment One; it should be noted that for parts not detailed in this embodiment, the descriptions in other embodiments can be referred to. The method includes:

[0118] S610. Perform a normal transformation on the obtained index data to obtain the normally adjusted data, and calculate the category factor based on the normally adjusted data.

[0119] S620. Calculate the regression coefficient estimate based on the category factor, the market value data of the relative value index corresponding to the index data, and the total asset data.

[0120] S631. Perform a cross-sectional regression on the regression coefficient estimate to obtain a significant pricing factor.

[0121] Among them, the calculation formula for cross-sectional regression is as follows:

[0122]

[0123] Among them, represents the average return rate of the \(i\)th enterprise within the preset time range, while represents the mean of the risk-free return rate, which is defined here as the loan prime rate (LPR); \(\gamma\) i,0 is the intercept term, \(\gamma\) i,k is the significant pricing factor, \(\beta\) i,k is the regression coefficient estimate, \(\varepsilon\) i is the residual term, \(\varepsilon\) i is assumed to follow a normal distribution with a mean of zero and a variance of \(\sigma\) 2 ; a significant \(\gamma\) i,k means that the corresponding risk factor represented by \(\beta\) i,k is significantly priced by the capital market.

[0124] S632. Calculate the high-group return and the low-group return based on the significant pricing factor.

[0125] Among them, \(k\) significant pricing factors can be calculated through the above cross-sectional regression calculation formula. The significant pricing factors are sorted in descending order, and the first 1 / 3 of the significant pricing factors are used as the high-significant pricing factor group, and the last 1 / 3 of the significant pricing factors are used as the low-significant pricing factor group;

[0126] Furthermore, calculate the return \(R\) group,t corresponding to the high-significant pricing factor group. The calculation formula for the return is as follows:

[0127]

[0128] Among them, \(MV\) i,t is the market value of the relative value index corresponding to the \(i\)th enterprise at time \(t\) for the high-significant pricing factor group or the low-significant pricing factor group, \(R\) i,t is the preset weight corresponding to the \(i\)th enterprise at time \(t\); and the return corresponding to the high-significant pricing factor group is recorded as the high-group return \(R\) High,t , and the return corresponding to the low-significant pricing factor group is also recorded as the low-group return \(R\) Low,t .

[0129] S633. Calculate the high-minus-low return index based on the high-group return and the low-group return.

[0130] Among them, the high-minus-low return index \(HML\) k,t is the difference between the high-group return \(R\) High,t and the low-group return \(R\) Low,t .

[0131] S634. Determine the return value coefficient based on the high minus low return indicator.

[0132] S634. By performing a time series regression on the high minus low return indicator HML k,t the return value coefficient θ can be obtained. i,k .

[0133] S640. Determine the asset weights based on the return value coefficient, and configure the investment portfolio based on the asset weights and the preset investment strategy.

[0134] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are displayed in sequence according to the indications of the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps does not have a strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same moment, but can be executed at different moments. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0135] Embodiment Seven

[0136] Based on the same inventive concept, this embodiment also provides an investment portfolio allocation device for implementing the above-mentioned investment portfolio allocation method. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the investment portfolio allocation device provided below can refer to the limitations on the investment portfolio allocation method in the above text, and will not be repeated here.

[0137] In this embodiment, as Figure 2 shown, an investment portfolio allocation device is provided, including:

[0138] A factor calculation module, configured to perform a normal transformation on the obtained index data to obtain normally adjusted data, and calculate category factors based on the normally adjusted data;

[0139] A coefficient estimation module, configured to calculate the regression coefficient estimation value based on the category factors, the market value data of the relative value index corresponding to the index data, and the total asset data;

[0140] A coefficient determination module, configured to calculate the high minus low return indicator based on the regression coefficient estimation value, and determine the return value coefficient based on the high minus low return indicator;

[0141] A combined configuration module for determining asset weights based on the return value coefficients and configuring an investment portfolio based on the asset weights and a preset investment strategy.

[0142] Each module in the above investment portfolio configuration device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0143] It should be noted that in the above investment portfolio configuration method, device, equipment, and storage medium of this embodiment, the obtained index data is subjected to a normal transformation to obtain normally adjusted data, and category factors are calculated based on the normally adjusted data; regression coefficient estimates are calculated based on the category factors, the market value data of the relative value index corresponding to the index data, and the total asset data; high-low reduced return indicators are calculated based on the regression coefficient estimates, and return value coefficients are determined based on the high-low reduced return indicators; asset weights are determined based on the return value coefficients, and an investment portfolio is configured based on the asset weights and a preset investment strategy. Through the above implementation, the normal transformation is used to ensure that the skewness of all variables is improved, and at the same time, the two-way clustering adjustment makes the regression results of the model robust and significant. Then, the regression coefficient estimates are calculated and the return value coefficients are calculated to regress the excess returns of each enterprise, so as to facilitate the calculation of the asset weights corresponding to the assets and realize the construction of the investment portfolio, thus facilitating the satisfaction of the investor's configuration requirements for the investment portfolio. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description.

[0144] In one embodiment, in terms of performing a normal transformation on the obtained index data to obtain normally adjusted data, the factor calculation module is specifically used for:

[0145] Performing linear interpolation on the obtained index data to obtain initial frequency modulation data;

[0146] Performing seasonal decomposition on the initial frequency modulation data to obtain target frequency modulation data;

[0147] Processing the target frequency modulation data based on a preset normal transformation algorithm to obtain normally adjusted data.

[0148] In one embodiment, in terms of calculating category factors based on the normally adjusted data, the factor calculation module is specifically used for:

[0149] Preprocessing the normally adjusted data to obtain target index data, and calculating a priori weights and an a priori covariance matrix based on the target index data;

[0150] Calculate the target index weight based on the prior weight and the prior covariance matrix;

[0151] Calculate the category factor based on the target index weight.

[0152] In one embodiment, in terms of calculating the regression coefficient estimate based on the category factor, the relative value index market value data and the total asset data corresponding to the index data, the coefficient estimation module is specifically configured to:

[0153] Construct a relative valuation factor model based on the category factor, the relative value index market value data and the total asset data corresponding to the index data;

[0154] Determine the model standard error data of the relative valuation factor model;

[0155] Perform clustering adjustment on the model standard error data based on the individual dimension and the time dimension to obtain the regression coefficient estimate.

[0156] In one embodiment, in terms of performing clustering adjustment on the model standard error data based on the individual dimension and the time dimension to obtain the regression coefficient estimate, the coefficient estimation module is specifically configured to:

[0157] Perform clustering on the model standard error data based on the individual dimension to obtain an individual clustering result, and calculate an individual clustering residual based on the individual clustering result;

[0158] Perform clustering on the model standard error data based on the time dimension to obtain a time clustering result, and calculate a time clustering residual based on the time clustering result;

[0159] Determine the regression coefficient estimate based on the individual clustering residual and the time clustering residual.

[0160] In one embodiment, in terms of calculating the high-low return index based on the regression coefficient estimate, the coefficient determination module is specifically configured to:

[0161] Perform cross-sectional regression on the regression coefficient estimate to obtain a significant pricing factor;

[0162] Calculate the high-group return and the low-group return based on the significant pricing factor;

[0163] Calculate the high-low return index based on the high-group return and the low-group return.

[0164] Embodiment VIII

[0165] In this embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows Figure 3As shown in the figure. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a portfolio configuration method.

[0166] Those skilled in the art can understand that Figure 3 the structure shown in the figure is only a block diagram of some structures related to the solution of the present disclosure, and does not constitute a limitation on the computer device to which the solution of the present disclosure is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0167] Embodiment Nine

[0168] In this embodiment, a computer-readable storage medium is provided. As Figure 4 shown, a computer program is stored thereon, and when the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0169] Embodiment Ten

[0170] In this embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by the processor, it implements the steps in the above method embodiments.

[0171] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present disclosure are all information and data authorized by the user or fully authorized by all parties.

[0172] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided by the present disclosure can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided by the present disclosure can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided by the present disclosure can be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0173] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0174] The above-described embodiments only represent several implementation manners of the present disclosure. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present disclosure. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present disclosure, several modifications and improvements can be made, and these all belong to the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the appended claims.

Claims

1. A portfolio allocation method, characterized in that, Including: Performing a normal transformation on the obtained indicator data to obtain normally adjusted data, and calculating a category factor based on the normally adjusted data; Calculating an estimated regression coefficient based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data; Calculating a high-low reduction return indicator based on the estimated regression coefficient, and determining a return value coefficient based on the high-low reduction return indicator; Determining an asset weight based on the return value coefficient, and allocating an investment portfolio based on the asset weight and a preset investment strategy.

2. The method according to claim 1, wherein The performing a normal transformation on the obtained indicator data to obtain normally adjusted data includes: Performing linear interpolation on the obtained indicator data to obtain initial frequency modulation data; Performing seasonal decomposition on the initial frequency modulation data to obtain target frequency modulation data; Processing the target frequency modulation data based on a preset normal transformation algorithm to obtain normally adjusted data.

3. The method according to claim 1, wherein The calculating a category factor based on the normally adjusted data includes: Performing preprocessing on the normally adjusted data to obtain target indicator data, and calculating a prior weight and a prior covariance matrix based on the target indicator data; Calculating a target indicator weight based on the prior weight and the prior covariance matrix; Calculating a category factor based on the target indicator weight.

4. The method according to claim 1, characterized in that, The calculating an estimated regression coefficient based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data includes: Constructing a relative valuation factor model based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data; Determining the model standard error data of the relative valuation factor model; Performing clustering adjustment on the model standard error data based on the individual dimension and the time dimension to obtain an estimated regression coefficient.

5. The method according to claim 4, characterized in that The performing clustering adjustment on the model standard error data based on the individual dimension and the time dimension to obtain an estimated regression coefficient includes: Performing clustering on the model standard error data based on the individual dimension to obtain an individual clustering result, and calculating an individual clustering residual based on the individual clustering result; Performing clustering on the model standard error data based on the time dimension to obtain a time clustering result, and calculating a time clustering residual based on the time clustering result; Determining an estimated regression coefficient based on the individual clustering residual and the time clustering residual.

6. The method according to claim 1, wherein The calculating a high-low reduction return indicator based on the estimated regression coefficient includes: Performing cross-sectional regression on the estimated regression coefficient to obtain a significant pricing factor; Calculating a high-group return and a low-group return based on the significant pricing factor; Calculating a high-low reduction return indicator based on the high-group return and the low-group return.

7. An investment portfolio allocation device, characterized in that, The device includes: A factor calculation module, configured to perform a normal transformation on the obtained indicator data to obtain normally adjusted data, and calculate a category factor based on the normally adjusted data; A coefficient estimation module, configured to calculate an estimated regression coefficient based on the category factor, the market value data of the relative value indicator corresponding to the indicator data, and the total asset data; A coefficient determination module, configured to calculate a high-low reduction return indicator based on the estimated regression coefficient, and determine a return value coefficient based on the high-low reduction return indicator; A combined configuration module, configured to determine asset weights based on the return value coefficients, and configure an investment portfolio based on the asset weights and a preset investment strategy.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images