Asset allocation method and related device
By generating high-frequency factor data and adjusting the asset portfolio using sparse jump models and asset price-volume data, the problem of capturing market dynamics and short-term fluctuations in existing technologies is solved, achieving more optimized asset allocation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-04
- Publication Date
- 2026-04-14
AI Technical Summary
Existing static allocation models based on historical data, mean-variance analysis, and investor risk preferences are insufficient to capture the dynamic characteristics and short-term fluctuations of the market when dealing with nonlinear changes and sudden jumps, leading to limitations in asset allocation strategies.
By generating multiple high-frequency factor data, a sparse jump model is used to determine market conditions and factor returns. The initial portfolio is then adjusted based on asset price and volume data to generate an adjusted portfolio.
It can capture changes in market conditions, balance risks and returns, and provide users with more optimized asset allocation strategies.
Smart Images
Figure CN121860780A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of asset allocation technology, and in particular to an asset allocation method and related apparatus. Background Technology
[0002] Asset allocation is one of the core issues in the field of financial investment, aiming to achieve a balance between risk and return through the rational allocation of various asset classes such as stocks, bonds, commodities, and cash.
[0003] Currently, asset allocation can be achieved using mean-variance analysis based on historical data or static allocation models based on investor risk preferences. However, these methods have significant limitations in dealing with non-linear market changes and sudden jumps, and struggle to capture the dynamic characteristics and short-term fluctuations of the market.
[0004] Therefore, there is an urgent need for a solution to address the aforementioned technical problems. Summary of the Invention
[0005] In view of the above problems, this application provides an asset allocation method and related apparatus.
[0006] The embodiments of this application disclose the following technical solutions:
[0007] Firstly, this application provides an asset allocation method, including:
[0008] Multiple high-frequency factor data are generated based on asset data from multiple asset classes and raw factor data;
[0009] The multiple high-frequency factor data are input into a sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state.
[0010] An initial asset portfolio is generated based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state.
[0011] Based on the asset price and volume data corresponding to each of the high-frequency factor data, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0012] In one possible implementation, the generation of multiple high-frequency factor data based on asset data from multiple asset classes and raw factor data includes:
[0013] Based on the correlation between asset data from multiple asset classes and the original factor data, highly correlated asset data are identified from the asset data of the multiple asset classes.
[0014] Based on the highly correlated asset data and the original factor data, multiple high-frequency factor data are generated.
[0015] In one possible implementation, the generation of multiple high-frequency factor data based on the highly correlated asset data and the original factor data includes:
[0016] Using the asset return rate of the highly correlated asset data as the independent variable and the original factor data as the dependent variable, the regression coefficient of each highly correlated asset data is determined.
[0017] Multiple high-frequency factor data are generated based on the regression coefficients of each highly correlated asset data.
[0018] In one possible implementation, generating an initial portfolio based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state includes:
[0019] Based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state, the high-frequency factor data is divided into a training set and a validation set for the sparse jump model, and the test period is determined.
[0020] Based on the training set, validation set, and test period, an initial asset portfolio is generated.
[0021] In one possible implementation, adjusting the initial asset portfolio based on the asset price and volume data corresponding to each of the high-frequency factor data to generate an adjusted asset portfolio includes:
[0022] Based on the asset price and volume data corresponding to each of the high-frequency factor data, the correlation between the assets in the initial asset portfolio is determined;
[0023] Based on the relationships between the assets, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0024] In one possible implementation, adjusting the initial asset portfolio based on the relationships between the assets to generate an adjusted asset portfolio includes:
[0025] Based on the relationships between the assets, the weight of each asset is determined.
[0026] Based on the weights corresponding to each asset, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0027] Second aspect: This application provides an asset allocation device, including:
[0028] The system comprises a first generation unit, an acquisition unit, a second generation unit, and an adjustment unit.
[0029] The first generation unit is used to generate multiple high-frequency factor data based on asset data of multiple asset classes and original factor data;
[0030] The obtaining unit is used to input the multiple high-frequency factor data into a sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state.
[0031] The second generation unit is used to generate an initial asset portfolio based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state;
[0032] The adjustment unit is used to adjust the initial asset portfolio based on the asset price and volume data corresponding to each of the high-frequency factor data, and generate an adjusted asset portfolio.
[0033] In one possible implementation, the first generating unit is specifically used for:
[0034] Based on the correlation between asset data from multiple asset classes and the original factor data, highly correlated asset data are identified from the asset data of the multiple asset classes.
[0035] Based on the highly correlated asset data and the original factor data, multiple high-frequency factor data are generated.
[0036] In one possible implementation, the first generating unit is specifically used for:
[0037] Using the asset return rate of the highly correlated asset data as the independent variable and the original factor data as the dependent variable, the regression coefficient of each highly correlated asset data is determined.
[0038] Multiple high-frequency factor data are generated based on the regression coefficients of each highly correlated asset data.
[0039] In one possible implementation, the second generating unit is specifically used for:
[0040] Based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state, the high-frequency factor data is divided into a training set and a validation set for the sparse jump model, and the test period is determined.
[0041] Based on the training set, validation set, and test period, an initial asset portfolio is generated.
[0042] In one possible implementation, the adjustment unit is specifically used for:
[0043] Based on the asset price and volume data corresponding to each of the high-frequency factor data, the correlation between the assets in the initial asset portfolio is determined;
[0044] Based on the relationships between the assets, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0045] In one possible implementation, the adjustment unit is specifically used for:
[0046] Based on the relationships between the assets, the weight of each asset is determined.
[0047] Based on the weights corresponding to each asset, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0048] Third aspect: This application provides a computer device, which includes a processor and a memory;
[0049] The memory is used to store program code and transmit the program code to the processor;
[0050] The processor is used to execute the steps of an asset allocation method as described above, according to the instructions in the program code.
[0051] Fourth aspect: Embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of an asset allocation method as described above.
[0052] Fifth aspect: This application provides a computer program product, which, when run on a computer, enables the computer to execute the steps of an asset configuration method as described above.
[0053] Sixth aspect: This application provides a chip including a processor coupled to a memory for executing a computer program or instructions stored in the memory, such that the chip implements the steps of an asset allocation method as described above.
[0054] Compared with the prior art, this application has the following beneficial effects:
[0055] This application provides an asset allocation method and related apparatus. It generates multiple high-frequency factor data based on asset data from multiple asset classes and original factor data. The high-frequency factor data is then input into a sparse jump model to obtain the market state and factor return rate corresponding to each high-frequency factor data. An initial asset portfolio is generated based on the market state and factor return rate of each high-frequency factor data. Finally, the initial asset portfolio is adjusted based on the asset price-volume data corresponding to each high-frequency factor data to generate an adjusted asset portfolio. This application, by constructing high-frequency factor data and using a sparse jump model to determine the market state and factor return rate of the high-frequency factor data, captures changes in market state. Based on this, asset price-volume data is introduced to adjust the initial asset portfolio, generating an adjusted asset portfolio that balances risk and return, providing users with a more optimized asset allocation strategy. Attached Figure Description
[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0057] Figure 1 This is a schematic diagram illustrating an application scenario of an asset allocation method provided in an embodiment of this application.
[0058] Figure 2 A flowchart illustrating an asset allocation method provided in this application embodiment;
[0059] Figure 3 A schematic diagram illustrating the generation of an initial asset portfolio, provided as an embodiment of this application;
[0060] Figure 4 A schematic diagram of DQN model training provided in an embodiment of this application;
[0061] Figure 5 A schematic diagram illustrating the generation of asset status as provided in an embodiment of this application;
[0062] Figure 6 This is a schematic diagram of an asset allocation device provided in an embodiment of this application. Detailed Implementation
[0063] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The terminology used in the following embodiments is for the purpose of describing specific embodiments only and is not intended to be a limitation of this application. As used in the specification and appended claims of this application, the singular expressions "a," "an," "the," "the," "the," and "this" are intended to also include expressions such as "one or more," unless the context clearly indicates otherwise. It should also be understood that in the embodiments of this application, "one or more" refers to one, two, or more; "and / or" describes the relationship between related objects, indicating that three relationships may exist; for example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.
[0064] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.
[0065] The "multiple" mentioned in the embodiments of this application refers to two or more. It should be noted that in the description of the embodiments of this application, terms such as "first" and "second" are used only for the purpose of distinguishing descriptions and should not be construed as indicating or implying relative importance, nor should they be construed as indicating or implying order.
[0066] Currently, asset allocation can be achieved using mean-variance analysis based on historical data or static allocation models based on investor risk preferences. However, these methods have significant limitations in dealing with non-linear market changes and sudden jumps, and struggle to capture the dynamic characteristics and short-term fluctuations of the market.
[0067] Based on this, embodiments of this application provide an asset allocation method and related apparatus, which generates multiple high-frequency factor data based on asset data of multiple asset classes and original factor data; inputs the multiple high-frequency factor data into a sparse jump model to obtain the market state and factor return rate corresponding to each high-frequency factor data; generates an initial asset portfolio based on the market state and factor return rate corresponding to each high-frequency factor data; and adjusts the initial asset portfolio based on the asset price and volume data corresponding to each high-frequency factor data to generate an adjusted asset portfolio.
[0068] In this embodiment, high-frequency factor data is constructed, and a sparse jump model is used to determine the market state of the high-frequency factor data and the factor return rate under the market state. This process can capture changes in the market state. Based on this, asset price and volume data are introduced to adjust the initial asset portfolio and generate an adjusted asset portfolio that can balance risk and return, providing users with a more optimized asset allocation strategy.
[0069] like Figure 1 As shown in the figure, this figure is a schematic diagram of an application scenario of an asset allocation method provided in an embodiment of this application. The method can be implemented based on the interaction between client 101 and server 102.
[0070] Client 101 sends asset data for multiple asset categories to server 102.
[0071] Server 102 generates multiple high-frequency factor data based on asset data from multiple asset classes and original factor data; inputs the multiple high-frequency factor data into a sparse jump model to obtain the market state and factor return rate corresponding to each high-frequency factor data; generates an initial asset portfolio based on the market state and factor return rate corresponding to each high-frequency factor data; and adjusts the initial asset portfolio based on the asset price and volume data corresponding to each high-frequency factor data to generate an adjusted asset portfolio.
[0072] After generating the adjusted asset portfolio, server 102 can send the adjusted asset portfolio to client 101.
[0073] To facilitate understanding, the following will be combined with... Figure 2 This application introduces an asset allocation method provided by an embodiment. Figure 2 A flowchart of an asset allocation method provided in this application embodiment includes S201-S204.
[0074] S201. Generate multiple high-frequency factor data based on asset data from multiple asset classes and raw factor data.
[0075] In this embodiment of the application, asset data with high correlation can be determined from the asset data of the multiple asset categories based on the correlation between asset data of multiple asset categories and the original factor data; and multiple high-frequency factor data can be generated based on the asset data with high correlation and the original factor data.
[0076] In the process of generating multiple high-frequency factor data based on the highly correlated asset data and the original factor data, the regression coefficient of each highly correlated asset data can be determined by using the asset return rate of the highly correlated asset data as the independent variable and the original factor data as the dependent variable; and multiple high-frequency factor data are generated based on the regression coefficient of each highly correlated asset data.
[0077] For example, taking raw factor data as an example of growth factors, growth factors are usually represented by indicators such as the year-on-year difference of the Purchasing Managers' Index (PMI), the year-on-year growth of fixed asset investment, the year-on-year growth of total retail sales of consumer goods, and the year-on-year growth of import and export amounts.
[0078] The original factor data conforms to economic logic and is widely tracked, observed and predicted by the market. However, it inevitably suffers from problems such as low frequency, delayed updates and inconsistent year-on-year and month-on-month comparisons. It is suitable for subjective tracking, observation and prediction, but difficult to use for quantitative modeling.
[0079] Based on this, the embodiments of this application generate multiple high-frequency factor data based on asset data from multiple asset classes and the original factor data. Taking the growth factor as an example, the method provided in the embodiments of this application determines asset data with a high correlation to the growth factor from the asset data of multiple asset classes. For example, asset data with a high correlation to the growth factor can be asset data with a high correlation to economic growth.
[0080] Based on this, a univariate regression model is established using the asset return rate of highly correlated asset data as the independent variable and the original factor data as the dependent variable to screen the highly correlated asset data and obtain the screened asset data.
[0081] For example, if the regression coefficient between the return of a stock index and the growth rate of GDP is significantly positive, it indicates that the asset data is sensitive to economic growth. That is, when GDP growth increases, the return of the asset data tends to rise in tandem, reflecting its strong correlation with the macroeconomic cycle.
[0082] A multiple regression model is established based on the screened asset data, regression coefficients are determined, and multiple high-frequency factor data are generated based on the regression coefficients.
[0083] S202. Input the multiple high-frequency factor data into the sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state.
[0084] Among them, the sparse skip model can allocate data to different groups (or states) and optimize the allocation by minimizing the objective function.
[0085] The sparse skip model in this application adds an additional constraint term to the objective function to capture the skip characteristics in time series data, as shown in equation (1):
[0086] (1)
[0087] in, This represents the t-th data point; yes Status The center; , indicating the state assignment of data points; It is an indicator function that indicates whether a state jump has occurred; This is a penalty parameter used to control the sparsity of state jumps.
[0088] In the embodiments of this application, It is a vector, where each element represents the return of each high-frequency factor data point at the current time. For example, The form can be [1%, 2%, -1%, ...], where 1% represents the return of the first high-frequency factor data; 2% represents the return of the second high-frequency factor data; and -1% represents the return of the second high-frequency factor data.
[0089] Based on equation (1), the market state corresponding to each high-frequency factor data and the factor return under the market state can be determined. The market state corresponding to the high-frequency factor data can be represented by the latent state of the high-frequency factor data at each time point; the factor return under the market state can be represented by the central vector of each latent state.
[0090] For example, when the number of hidden states is set to 2, The value can be 0 or 1, representing two market states (e.g., "bull market" or "bear market"), and each market state corresponds to a central vector, i.e. .
[0091] It is understood that the number of hidden states, The value of is not specifically limited; the above is merely an example.
[0092] S203. Based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state, generate an initial asset portfolio.
[0093] In this embodiment of the application, based on the market state corresponding to each high-frequency factor data and the factor return rate under the market state, the high-frequency factor data is divided into a training set and a validation set for a sparse jump model, and a test period is determined; based on the training set, the validation set and the test period, an initial asset portfolio is generated.
[0094] In this embodiment, when training the sparse skip model, the data is divided into a training set and a validation set. A relatively long time window is selected as the training set, and the number of hidden states and constraint coefficients are initialized. Based on this, within this time window, an optimization algorithm is used to calculate the center vector of each hidden state and the hidden state corresponding to each time point.
[0095] After obtaining the center vector, online inference is performed on the validation set, and the hidden state at the last time point is used to replace the prediction of the future, thereby generating a view on the future trend of factors.
[0096] For example, in this embodiment of the application, the center vector obtained through training can be fixed, and then the validation set can be divided into a series of test periods, such as a validation set period of 120 trading days and each test period of 30 trading days. The sparse jump model can then be optimized again during the test periods. Unlike the training phase, the process of optimizing the sparse jump model again during the test periods only obtains the hidden state of the high-frequency factors at each time point. Based on this, the hidden state of the last trading day can be used as a basis for future predictions.
[0097] In this embodiment, the purpose of setting up a validation set is to optimize hyperparameters, including the number of hidden states and constraint coefficients. The evaluation criteria for optimization are the return rate and Sharpe ratio obtained from backtesting on the validation set. After optimization, to ensure the stability of the sparse skip model, the optimal hyperparameters and the trained center vector are fixed, and actual backtesting can be performed during the testing period. After the backtesting is completed, the process continues to scroll forward until the entire historical period has been traversed, finally obtaining the backtesting results.
[0098] like Figure 3 As shown in the figure, this figure is a schematic diagram of generating an initial asset portfolio according to an embodiment of this application.
[0099] In the presence of three high-frequency factors, the historical average return of each asset data in the predicted latent state is calculated. Based on this, an initial asset portfolio is generated according to the sparse jump model.
[0100] The three high-frequency factors include a first factor, a second factor, and a third factor. The data of the first factor, the second factor, and the third factor are combined at the same time point, and the combined data is divided into a training set, a validation set, and a test set.
[0101] The model training phase includes initializing hyperparameters, optimizing the sparse skip model on the training set, and outputting the hidden state sequence and center vector after training is completed.
[0102] The model validation phase includes using the model obtained in the model training phase to perform online inference on the validation set, using the prediction results to backtest the policy, and optimizing the hyperparameters again based on the backtest results to obtain the optimized hyperparameters.
[0103] During the model testing phase, the sparse jump model is further optimized based on the optimized hyperparameters over the testing period, and an initial portfolio is generated based on this further optimized sparse jump model.
[0104] S204. Based on the asset price and volume data corresponding to each of the high-frequency factor data, adjust the initial asset portfolio to generate an adjusted asset portfolio.
[0105] In this embodiment of the application, the correlation between assets in the initial asset portfolio is determined based on the asset price and volume data corresponding to each of the high-frequency factor data; based on the correlation between the assets, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0106] Based on the relationships between the assets, the initial asset portfolio is adjusted to generate the adjusted asset portfolio. In the process, the weight of each asset can be determined based on the relationships between the assets. Based on the weights of each asset, the initial asset portfolio is adjusted to generate the adjusted asset portfolio.
[0107] For example, after obtaining an initial portfolio based on high-frequency factor data and a sparse jump model, the initial portfolio can be adjusted using a deep Q-network (DQN) model by combining the trend information and price and volume data of each asset, thereby generating an adjusted portfolio.
[0108] The training process of the DQN model is as follows: Figure 4 As shown in the figure, this figure is a schematic diagram of DQN model training provided in an embodiment of this application.
[0109] The loss function of the DQN model is used to minimize the error between the target Q-value and the current Q-network prediction. Here, s represents the state of each asset; a represents the action given the state (in the DQN algorithm, actions are discrete and can be buying, selling, or holding); and r represents the reward for the action, such as the profit or volatility generated by that action.
[0110] In this embodiment of the application, the status of each asset can be... Figure 5 The network shown is generated. Figure 5 This is a schematic diagram illustrating the generation of asset status as provided in an embodiment of this application.
[0111] like Figure 5 As shown, the price and volume data of each asset (such as the first asset, the second asset, and the third asset) are input into the corresponding feature extraction networks (such as the first feature network, the second feature network, and the third feature network) to obtain the first feature corresponding to the first asset, the second feature corresponding to the second asset, and the third feature corresponding to the third asset.
[0112] By inputting the first feature corresponding to the first asset, the second feature corresponding to the second asset, and the third feature corresponding to the third asset into the cross-asset evaluation network, the correlation between different assets can be extracted through the attention mechanism. The correlation between assets is then added to the state of each asset, resulting in the states of each asset including the correlation (e.g., s1, s2, s3). Based on this, the trained DQN model is obtained.
[0113] Based on the trained DQN model, buy or sell recommendations can be determined for each asset, and the initial portfolio can be adjusted based on the buy or sell recommendations for each asset.
[0114] For example, by determining the weights of each asset, the initial asset portfolio can be adjusted based on these weights to generate an adjusted asset portfolio. For instance, the weight of an asset for which a buy recommendation is given can be increased by a certain percentage, while the weight of an asset for which a sell recommendation is given can be decreased by a certain percentage.
[0115] In summary, this application embodiment constructs high-frequency factor data and uses a sparse jump model to determine the market state of the high-frequency factor data and the factor returns under the market state. This process can capture changes in the market state. Based on this, asset price and volume data are introduced to adjust the initial asset portfolio and generate an adjusted asset portfolio, which can balance risk and return and provide users with a more optimized asset allocation strategy.
[0116] This application provides an asset allocation device, see [link to relevant documentation]. Figure 6The figure is a schematic diagram of the structure of an asset allocation device provided in an embodiment of this application. Its specific implementation method is consistent with the implementation method and the technical effect achieved in the embodiments of the above method. Some contents will not be repeated.
[0117] This application provides an asset allocation device 6100, including:
[0118] The system comprises a first generation unit 6101, an acquisition unit 6102, a second generation unit 6103, and an adjustment unit 6104.
[0119] The first generation unit 6101 is used to generate multiple high-frequency factor data based on asset data of multiple asset classes and original factor data;
[0120] The obtaining unit 6102 is used to input the multiple high-frequency factor data into a sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state.
[0121] The second generation unit 6103 is used to generate an initial asset portfolio based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state;
[0122] The adjustment unit 6104 is used to adjust the initial asset portfolio based on the asset price and volume data corresponding to each of the high-frequency factor data, and generate an adjusted asset portfolio.
[0123] In one possible implementation, the first generating unit is specifically used for:
[0124] Based on the correlation between asset data from multiple asset classes and the original factor data, highly correlated asset data are identified from the asset data of the multiple asset classes.
[0125] Based on the highly correlated asset data and the original factor data, multiple high-frequency factor data are generated.
[0126] In one possible implementation, the first generating unit is specifically used for:
[0127] Using the asset return rate of the highly correlated asset data as the independent variable and the original factor data as the dependent variable, the regression coefficient of each highly correlated asset data is determined.
[0128] Multiple high-frequency factor data are generated based on the regression coefficients of each highly correlated asset data.
[0129] In one possible implementation, the second generating unit is specifically used for:
[0130] Based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state, the high-frequency factor data is divided into a training set and a validation set for the sparse jump model, and the test period is determined.
[0131] Based on the training set, validation set, and test period, an initial asset portfolio is generated.
[0132] In one possible implementation, the adjustment unit is specifically used for:
[0133] Based on the asset price and volume data corresponding to each of the high-frequency factor data, the correlation between the assets in the initial asset portfolio is determined;
[0134] Based on the relationships between the assets, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0135] In one possible implementation, the adjustment unit is specifically used for:
[0136] Based on the relationships between the assets, the weight of each asset is determined.
[0137] Based on the weights corresponding to each asset, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
[0138] In summary, this application embodiment constructs high-frequency factor data and uses a sparse jump model to determine the market state of the high-frequency factor data and the factor returns under the market state. This process can capture changes in the market state. Based on this, asset price and volume data are introduced to adjust the initial asset portfolio and generate an adjusted asset portfolio, which can balance risk and return and provide users with a more optimized asset allocation strategy.
[0139] This application provides a computer device, which includes a processor and a memory;
[0140] The memory is used to store program code and transmit the program code to the processor;
[0141] The processor is used to execute the steps of an asset allocation method as described above, according to the instructions in the program code.
[0142] For example, the processor generates multiple high-frequency factor data based on asset data from multiple asset classes and original factor data; inputs the multiple high-frequency factor data into a sparse jump model to obtain the market state and factor return rate corresponding to each high-frequency factor data; generates an initial asset portfolio based on the market state and factor return rate corresponding to each high-frequency factor data; and adjusts the initial asset portfolio based on the asset price and volume data corresponding to each high-frequency factor data to generate an adjusted asset portfolio.
[0143] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of an asset allocation method as described above.
[0144] This application provides a computer program product that, when run on a computer, executes the steps of an asset configuration method as described above.
[0145] This application provides a chip including a processor coupled to a memory for executing a computer program or instructions stored in the memory, such that the chip implements the steps of an asset allocation method as described above.
[0146] It should be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for the device and apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments. The device and apparatus embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components indicated as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0147] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An asset allocation method, characterized in that, include: Multiple high-frequency factor data are generated based on asset data from multiple asset classes and raw factor data; The multiple high-frequency factor data are input into a sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state. An initial asset portfolio is generated based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state. Based on the asset price and volume data corresponding to each of the high-frequency factor data, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
2. The method according to claim 1, characterized in that, The process generates multiple high-frequency factor data based on asset data from multiple asset classes and raw factor data, including: Based on the correlation between asset data from multiple asset classes and the original factor data, highly correlated asset data are identified from the asset data of the multiple asset classes. Based on the highly correlated asset data and the original factor data, multiple high-frequency factor data are generated.
3. The method according to claim 1, characterized in that, Based on the highly correlated asset data and the original factor data, multiple high-frequency factor data are generated, including: Using the asset return rate of the highly correlated asset data as the independent variable and the original factor data as the dependent variable, the regression coefficient of each highly correlated asset data is determined. Multiple high-frequency factor data are generated based on the regression coefficients of each highly correlated asset data.
4. The method according to claim 1, characterized in that, The process of generating an initial asset portfolio based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state includes: Based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state, the high-frequency factor data is divided into a training set and a validation set for the sparse jump model, and the test period is determined. Based on the training set, validation set, and test period, an initial asset portfolio is generated.
5. The method according to claim 1, characterized in that, The step of adjusting the initial asset portfolio based on the asset price and volume data corresponding to each of the high-frequency factor data to generate an adjusted asset portfolio includes: Based on the asset price and volume data corresponding to each of the high-frequency factor data, the correlation between the assets in the initial asset portfolio is determined; Based on the relationships between the assets, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
6. The method according to claim 5, characterized in that, The step of adjusting the initial asset portfolio based on the relationships between the assets to generate an adjusted asset portfolio includes: Based on the relationships between the assets, the weight of each asset is determined. Based on the weights corresponding to each asset, the initial asset portfolio is adjusted to generate an adjusted asset portfolio.
7. An asset allocation device, characterized in that, include: The system comprises a first generation unit, an acquisition unit, a second generation unit, and an adjustment unit. The first generation unit is used to generate multiple high-frequency factor data based on asset data of multiple asset classes and original factor data; The obtaining unit is used to input the multiple high-frequency factor data into a sparse jump model to obtain the market state corresponding to each high-frequency factor data and the factor return rate under the market state. The second generation unit is used to generate an initial asset portfolio based on the market state corresponding to each of the high-frequency factor data and the factor return rate under the market state; The adjustment unit is used to adjust the initial asset portfolio based on the asset price and volume data corresponding to each of the high-frequency factor data, and generate an adjusted asset portfolio.
8. A computer device, characterized in that, The computer device includes: a processor and a memory; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the steps of an asset allocation method as described in any one of claims 1-6 according to instructions in the program code.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of an asset allocation method as described in any one of claims 1-6.
10. A computer program product, characterized in that, When the computer program product is run on a computer, the computer performs the steps of an asset allocation method as described in any one of claims 1-6.