An asset allocation method, device, storage medium and computer program product

Abstract

Embodiments of the present application provide an asset allocation method and device, a storage medium and a computer program product. The method comprises: obtaining first historical data information corresponding to one or more first assets, and determining predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructing a first Hamiltonian based on the predicted value information corresponding to each second asset and second historical data information corresponding to the second asset; wherein the first Hamiltonian is used at least to determine an optimal asset investment portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; preparing a corresponding first quantum circuit based on the first Hamiltonian, and determining weight information corresponding to each second asset based on the first quantum circuit, to generate an optimal investment strategy based on the weight information, thereby improving asset allocation efficiency.

Description

Technical Field

[0001] This application relates to the field of quantum computing technology, and in particular to an asset allocation method, device, storage medium, and computer program product. Background Technology

[0002] Portfolio optimization is a core issue in finance, aiming to achieve the optimal balance between risk and return through quantitative analysis. With technological advancements, three main approaches have emerged: classical methods are based on the Markowitz mean-variance model, combined with sequential quadratic programming and genetic algorithms to handle high-dimensional nonlinear constraints; quantum computing approaches use the Harrow-Hassidim-Lloyd algorithm (HHL) to solve for the eigenvalues ​​of the covariance matrix, or construct a Quadratic Unconstrained Binary Optimization (QUBO) model to find the optimal solution using quantum annealing; and machine learning approaches, represented by FactorVAE, combine dynamic factor models with variational autoencoders to predict expected stock returns, and then complete the portfolio allocation based on the mean-variance model.

[0003] However, the relevant technologies still have significant drawbacks: classical methods are computationally inefficient when faced with massive amounts of data and complex constraints, and are susceptible to data noise; among mainstream quantum schemes, the HHL algorithm is limited by the scale of data and quantum noise, resulting in insufficient accuracy; the QUBO / Quantum Approximate Optimization Algorithm (QAOA) model relies on historical return data and is less adaptable to future market changes; while FactorVAE, combined with the framework of classical optimization, can improve the accuracy of return prediction, the subsequent allocation process still has not broken through the performance bottleneck of classical computing, making it difficult to meet the needs of large-scale real-time optimization, thus leading to a decline in the efficiency of asset allocation. Summary of the Invention

[0004] This application provides an asset allocation method, device, storage medium, and computer program product that can improve the efficiency and accuracy of asset allocation.

[0005] The technical solution of this application embodiment is implemented as follows: In a first aspect, embodiments of this application provide an asset allocation method, the method comprising: Obtain first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and a first preset algorithm; A first Hamiltonian is constructed based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second asset; wherein, the first Hamiltonian is used at least to determine the optimal asset portfolio, the first asset includes the second asset, and the first historical data information includes the second historical data information; A first quantum circuit is prepared based on the first Hamiltonian, and weight information corresponding to each second asset is determined based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information.

[0006] Secondly, embodiments of this application provide an asset configuration device, the asset configuration device comprising: a processor and a memory; wherein, The memory is used to store computer programs that can run on the processor; The processor is configured to execute the asset configuration method described above when running the computer program.

[0007] Thirdly, embodiments of this application provide a computer-readable storage medium storing computer program code, which, when executed by a computer, implements the asset allocation method described above.

[0008] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the asset allocation method described above.

[0009] This application provides an asset allocation method, device, storage medium, and computer program product. The method includes: acquiring first historical data information corresponding to one or more first assets, and determining predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructing a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; preparing a corresponding first quantum circuit based on the first Hamiltonian, and determining the weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information. Therefore, the embodiments of this application can construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset. That is, the first Hamiltonian constructed in this application is based on the predicted value information of the second asset, rather than the historical value information of the second asset. This can improve the accuracy of subsequent asset allocation. Then, a corresponding first quantum circuit can be prepared based on the first Hamiltonian, and the weight information corresponding to each second asset can be determined based on the first quantum circuit. In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit prepared based on the first Hamiltonian. The optimal investment strategy can be generated based on the weight information. The optimal investment strategy can be calculated by quantum circuit, which can meet the needs of large-scale real-time optimization and efficiently calculate the corresponding portfolio method, greatly accelerating the calculation time and thus improving the efficiency of asset allocation. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the asset allocation method proposed in the embodiments of this application. Figure 1 ; Figure 2 This is a schematic diagram of the FactorVAE architecture proposed in the embodiments of this application; Figure 3 This is a schematic diagram of the ansatz template circuit proposed in the embodiments of this application; Figure 4 This is a schematic diagram of the asset allocation method proposed in the embodiments of this application. Figure 2 ; Figure 5 This is a schematic diagram of the composition structure of the asset configuration equipment proposed in the embodiments of this application. Figure 1 ; Figure 6 This is a schematic diagram of the composition structure of the asset configuration equipment proposed in the embodiments of this application. Figure 2 . Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the relevant application and not for limiting the application. Furthermore, it should be noted that, for ease of description, only the parts related to the relevant application are shown in the accompanying drawings.

[0012] Portfolio optimization is a core issue in finance, aiming to achieve the optimal balance between risk and return through quantitative analysis. With technological advancements, three main approaches have emerged: classical methods are based on the Markowitz mean-variance model, combined with sequential quadratic programming and genetic algorithms to handle high-dimensional nonlinear constraints; quantum computing approaches use the HHL algorithm to solve for the eigenvalues ​​of the covariance matrix, or construct QUBO models to find optimal solutions using quantum annealing; and machine learning approaches, represented by FactorVAE, combine dynamic factor models with variational autoencoders to predict expected stock returns, and then complete the portfolio allocation based on the mean-variance model.

[0013] The relevant technologies still have significant shortcomings: classical methods are computationally inefficient when faced with massive amounts of data and complex constraints, and are susceptible to data noise; among the mainstream quantum schemes, the HHL algorithm is limited by the scale of data and quantum noise, resulting in insufficient accuracy; the QUBO / Quantum Approximate Optimization Algorithm (QAOA) model relies on historical return data and is less adaptable to future market changes; while FactorVAE, combined with the framework of classical optimization, can improve the accuracy of return prediction, the subsequent allocation process still has not broken through the performance bottleneck of classical computing, making it difficult to meet the needs of large-scale real-time optimization, thus leading to a decline in the efficiency of asset allocation.

[0014] To address the problem of low computational efficiency in portfolio optimization methods in related technologies, which leads to decreased efficiency in asset allocation, this application provides an asset allocation method, device, storage medium, and computer program product. The method includes: acquiring first historical data information corresponding to one or more first assets, and determining predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructing a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; preparing a corresponding first quantum circuit based on the first Hamiltonian, and determining weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information. Therefore, the embodiments of this application can construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset. That is, the first Hamiltonian constructed in this application is based on the predicted value information of the second asset, rather than the historical value information of the second asset. This can improve the accuracy of subsequent asset allocation. Then, a corresponding first quantum circuit can be prepared based on the first Hamiltonian, and the weight information corresponding to each second asset can be determined based on the first quantum circuit. In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit prepared based on the first Hamiltonian. The optimal investment strategy can be generated based on the weight information. The optimal investment strategy can be calculated by quantum circuit, which can meet the needs of large-scale real-time optimization and efficiently calculate the corresponding portfolio method, greatly accelerating the calculation time and thus improving the efficiency of asset allocation.

[0015] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0016] This application provides an asset allocation method. Figure 1 This is a schematic diagram of the asset allocation method proposed in the embodiments of this application. Figure 1 ,like Figure 1 As shown, asset allocation methods may include the following steps: Step 101: Obtain first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm.

[0017] In the embodiments of this application, the asset configuration device can acquire first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm.

[0018] It should be noted that, in the embodiments of this application, the asset configuration device can be any terminal or device with storage and communication functions. For example, the asset configuration device can be a personal computer (PC). This application does not specifically limit the type of asset configuration device.

[0019] It should be noted that, in the embodiments of this application, the first asset may include stocks, funds, etc., and this application does not specifically limit the type of asset included in the first asset.

[0020] It should be noted that, in the embodiments of this application, the first historical data information may include the historical pricing of the first asset, asset risk values, etc. The types and quantity of information included in the first historical data information in this application are not specifically limited. It should be noted that, in the embodiments of this application, the first preset algorithm may include the FactorVAE algorithm, and this application does not specifically limit the type of algorithm included in the first preset algorithm.

[0021] It should be noted that, in the embodiments of this application, the predicted value information may include the rate of return information corresponding to the first asset, and this application does not specifically limit the types of information included in the preset value information.

[0022] For example, in an embodiment of this application, when the asset allocation device determines the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm, it can do so through the following steps: First, obtain the historical data information of the first asset (such as the stock to be selected); historical data information of the stock This typically includes historical stock pricing, stock risk metrics, and historical returns. The second step involves using a feature extractor to extract latent features. In the FactorVAE algorithm, feature extraction primarily utilizes gated recurrent units (GRUs) to extract historical time-series features from historical stock data. This process incorporates the historical stock data obtained in the previous step. The first step involves a feature extractor that, relying on the GRU's ability to extract time-series features, extracts latent features *e* from historical stock data. The third step uses an encoder to combine these latent features with known future stock returns to extract a posterior factor. The encoder's role is to obtain a posterior factor by processing the latent features and historical stock returns from the historical data. This posterior factor conforms to ;in, Indicates a Gaussian distribution. Represents the posterior factor The mean of the normal distribution it follows. Represents the posterior factor The variance of the normal distribution it follows; the fourth step is to use a predictor, which inputs the hidden features of the historical data to obtain the prior factor. The predictor only inputs the hidden features e of the historical data. The predictor uses an attention mechanism and finally generates a prior factor. , ;in, Represents prior factors The mean of the normal distribution it follows. It is a priori factor The variance of the normal distribution it follows; the fifth step is to train the prior factors to reduce the loss function between the prior and posterior factors, thus obtaining the trained prior factors; the sixth step is to input the hidden features of the historical data and the trained prior factors into the decoder to obtain the predicted future return (i.e., the predicted value information); entering the decoder stage, here the hidden features of the historical data e and the trained prior factors must be input simultaneously. For the encoder of FactorVAE, it contains Layers and The specific principle formula for the layer is as follows (1).

[0023] (1) in, Indicates the projected rate of return. This represents the prior factors after training. , This indicates the layers contained in the encoder of FactorVAE.

[0024] For example, in the embodiments of this application, Figure 2 This is a schematic diagram of the FactorVAE architecture proposed in the embodiments of this application, as shown below. Figure 2As shown, FactorVAE can contain a Factor encoder and a Factor decoder. The meanings of each part of the Factor encoder are as follows: 1. Input layer: Latent features e: Represents the initial input of the model, usually preprocessed low-level market features, such as stock fundamentals data, technical indicators, etc.; Future returns y: Represents the target variable to be predicted, which is the market return signal that the model needs to learn to capture; 2. Core computation layer: Portfolio Layer: Receives latent features e and outputs portfolio weights a, used to construct the portfolio; Portfolios weights are generated by the Portfolio Layer and determine the proportion of different assets in the portfolio; Mapping Layer: Receives portfolio returns (Portfolios returns) ), outputting the posterior factor z_post, realizing the mapping from portfolio returns to the factor space; 3. Output and intermediate variables: Portfolios returns (Portfolio Returns): Obtained by multiplying portfolio weights and future returns (y), representing the returns of the constructed portfolio; Posterior Factors z_post is the output of the mapping layer, a factor obtained by combining portfolio returns, used for subsequent factor decoding. The meanings of each part of the Factor decoder are explained as follows: 1. Input Layer: Factors z: Posterior factors z_post from the encoder, the initial input to the decoder; Latent features e: Consistent with the latent features input from the encoder; 2. Core Computation Layer: Beta Layer: Receives factors z and latent features e, outputting exposed... This represents the sensitivity of an asset to market factors; Alpha Layer: receives factor z and latent feature e, and outputs alpha. 3. Output and intermediate variable: Exposure Factor exposure: The sensitivity of an asset to market factors, used to measure the extent to which asset returns are affected by market factors; Alpha (Excess Returns): The portion of an asset that generates excess returns; Stock returns (Equity return forecast): Determined by factor exposure ,alpha The result obtained by calculating with factor z (i.e., the posterior factor) is the final stock return prediction output of the model.

[0025] Optionally, in embodiments of this application, after determining the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm, the asset allocation device can perform screening processing on the first asset based on the predicted value information corresponding to the first asset and the first preset threshold to obtain the screened second asset; wherein, the second asset includes the first asset whose predicted value information is greater than or equal to the first preset threshold.

[0026] It should be noted that, in the embodiments of this application, the first preset threshold can be customized according to the actual application scenario, for example, it can be set to 0.01. This application does not specifically limit the size of the first preset threshold.

[0027] For example, in the embodiments of this application, the predicted value information (such as the rate of return) of the first asset can be compared with a first preset threshold, and the first asset whose predicted value information (such as the rate of return) is greater than or equal to the first preset threshold can be selected as the second asset after screening. In this way, the first asset can be cleaned and the accuracy of the data can be improved.

[0028] Step 102: Construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset; wherein, the first Hamiltonian is used at least to determine the optimal asset portfolio, the first asset includes the second asset, and the first historical data information includes the second historical data information.

[0029] In the embodiments of this application, after the asset allocation device acquires first historical data information corresponding to one or more first assets and determines the predicted value information corresponding to the first assets based on the first historical data information and the first preset algorithm, it can construct a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second asset.

[0030] It should be noted that, in the embodiments of this application, the second asset may be a portion of the first asset, and this application does not specifically limit the quantity of the second asset.

[0031] Optionally, in embodiments of this application, when the asset allocation device constructs a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset, it can calculate a first matrix based on each second historical data information and a mean vector; wherein, the mean vector contains the average value of the historical value information corresponding to each second asset, and the first matrix is ​​used to characterize the correlation risk between the second assets; then, the predicted value information corresponding to each second asset can be numbered, and a value information vector can be constructed based on the number; wherein, the number corresponds to the ground state of the qubit; and then, the first Hamiltonian can be constructed based on the first matrix and the value information vector.

[0032] It should be noted that, in the embodiments of this application, the first matrix may include a covariance matrix, and this application does not specifically limit the matrix type included in the first matrix.

[0033] It should be noted that, in the embodiments of this application, the value information vector may include a rate of return vector, and this application does not specifically limit the vector type included in the value information vector.

[0034] For example, in an embodiment of this application, when the asset allocation device calculates the first matrix based on each second historical data information and the mean vector, it can calculate the average value of the historical return of each stock according to the historical data information to obtain the average value vector (i.e., the mean vector). Then, the difference between each historical data point and the corresponding mean can be calculated. Then, the value of each element of the covariance matrix is ​​calculated according to formula (2) to obtain the covariance matrix (i.e. the first matrix). The calculation method is shown in the following formula (2).

[0035] (2) in, This represents the element in the i-th row and j-th column of the covariance matrix, reflecting the coordinated change relationship between the returns of stock i and stock j. If this value is positive, the probability that the returns of the two stocks move in the same direction is high; if it is negative, the probability that they move in opposite directions is high. This represents the number of historical data points representing the covariance. This represents the historical return of stock i at time point k. This represents the average historical return of stock i, reflecting its average historical return level. This represents the historical return of stock j at time point k. This represents the average historical return of stock j.

[0036] For example, in an embodiment of this application, when the asset allocation device assigns a number to the predicted value information corresponding to each second asset and constructs a value information vector based on the number, it can arrange the future rate of return (i.e., predicted value information) of the second asset, assign a number to each stock (i.e., the second asset), the number will correspond to each ground state in the quantum state, and then construct an expected rate of return vector (i.e., value information vector) according to the numbering order. .

[0037] Optionally, in embodiments of this application, when the asset allocation device constructs a first Hamiltonian based on a first matrix and a value information vector, it can determine a second Hamiltonian based on the first matrix and a first Pauli matrix; wherein the second Hamiltonian is used at least to characterize the risk information of the asset portfolio, and the first Pauli matrix is ​​used to characterize the ground state of the qubit; then a third Hamiltonian can be determined based on the value information vector and the second Pauli matrix; wherein the third Hamiltonian is used to characterize the value information of the asset portfolio; and further, the first Hamiltonian can be determined based on the second Hamiltonian and the third Hamiltonian.

[0038] For example, in an embodiment of this application, when the asset allocation device determines the second Hamiltonian based on the first matrix and the first Pauli matrix, the covariance matrix (i.e., the first matrix) represents the risk factors of the stock in a practical sense, and can be converted into a quadratic form using the Pauli matrix, as shown in the following formula (3).

[0039] (3) in, This represents the second Hamiltonian. Describes the first matrix. This represents the first Pauli matrix, corresponding to the measurement basis (i.e., the ground state) of the qubit.

[0040] For example, in the embodiments of this application, when the asset allocation device determines the third Hamiltonian based on the value information vector and the second Pauli matrix, it can convert the linear part of the future return vector (i.e., the value information vector) into a linear pattern using the Pauli matrix, as shown in the following formula (4).

[0041] (4) in, This represents the third Hamiltonian. Represents a value information vector. This represents the second Pauli matrix.

[0042] It should be noted that, in the embodiments of this application, after the asset configuration device determines the second Hamiltonian and the third Hamiltonian, it can determine the first Hamiltonian based on the second Hamiltonian and the third Hamiltonian.

[0043] For example, in an embodiment of this application, when the asset configuration device determines the first Hamiltonian based on the second Hamiltonian and the third Hamiltonian, it can use the following formula (5).

[0044] (5) in, This represents the first Hamiltonian. This represents the second Hamiltonian. This represents the third Hamiltonian. Describes the first matrix. This represents the first Pauli matrix, corresponding to the measurement basis (i.e., the ground state) of the qubit. Represents a value information vector. This represents the second Pauli matrix.

[0045] It should be noted that, in the embodiments of this application, the asset allocation device can determine the first Hamiltonian based on the second Hamiltonian and the third Hamiltonian. That is, the first Hamiltonian not only considers the risk information of the asset portfolio, but also the value information of the asset portfolio (such as the rate of return). In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian, thereby facilitating the acquisition of a better portfolio more quickly and efficiently in the future.

[0046] Step 103: Prepare the corresponding first quantum circuit based on the first Hamiltonian, and determine the weight information corresponding to each second asset based on the first quantum circuit, so as to generate the optimal investment strategy based on the weight information.

[0047] In the embodiments of this application, after the asset allocation device constructs a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset, it can prepare a corresponding first quantum circuit based on the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit, so as to generate the optimal investment strategy based on the weight information.

[0048] For example, in the embodiments of this application, when the asset configuration device prepares the corresponding first quantum circuit based on the first Hamiltonian, it can automatically convert the Hamiltonian (i.e. the first Hamiltonian) in the physical system into a quantum circuit form that can be executed on a quantum computer by using tools such as OpenFermion, Qiskit, and PennyLane. That is, it determines which quantum gate operations to apply to which qubits. This application does not specifically limit the method of preparing the quantum circuit.

[0049] It should be noted that, in the embodiments of this application, after the asset allocation device prepares the corresponding first quantum circuit based on the first Hamiltonian, it can add an ansatz circuit (i.e., a second quantum circuit) to the first quantum circuit. The second quantum circuit is used at least for training the first quantum circuit. The ansatz circuit is a hardware-efficient ansatz template circuit (i.e., a simplified_TwoDesign circuit layer). This circuit layer can perform calculations more efficiently on a quantum computer. This application does not specifically limit the type of circuit added to the first quantum circuit.

[0050] For example, in the embodiments of this application, Figure 3 This is a schematic diagram of the ansatz template circuit proposed in the embodiments of this application, as shown below. Figure 3 As shown, each column from left to right in the diagram represents a layer of quantum gate operations. Each qubit in each layer has a Ry gate. The lines connecting adjacent columns represent entanglement gates (such as controlled-NOT gates (CNOT)), which are responsible for establishing quantum entanglement between qubits. The parameter set of the entire circuit is the parameter set of all Ry gates. (i.e., the first parameter).

[0051] Optionally, in an embodiment of this application, when the asset allocation device determines the weight information corresponding to each second asset based on the first quantum circuit, it can optimize the first parameter of the second quantum circuit to obtain an optimized first quantum circuit; wherein, the first parameter includes the initial parameter of the second quantum circuit; then, it can perform measurement processing based on the optimized first quantum circuit to obtain the amplitude probability corresponding to each ground state; and then, it can determine the weight information based on the amplitude probability corresponding to each ground state.

[0052] It should be noted that, in the embodiments of this application, the first parameter includes the initial parameters of the second quantum circuit, such as the parameters of all Ry gates. A set of randomly initialized values; this application does not specify the type of the initial parameters.

[0053] Optionally, in an embodiment of this application, when the asset allocation device optimizes the first parameter of the second quantum circuit to obtain the optimized first quantum circuit, it can perform measurement processing on the first quantum circuit to obtain an initial expected value; wherein, the initial expected value is used to characterize the comprehensive value information in the current quantum state; then, iterative optimization processing of the first parameter of the second quantum circuit can be performed based on a preset optimization algorithm until the initial expected value is less than a second preset threshold, at which point the iterative optimization stops, and the optimized first quantum circuit is obtained.

[0054] It should be noted that, in the embodiments of this application, the initial expected value is used to characterize the comprehensive value information in the current quantum state. For example, the initial expected value of the current quantum state under the first Hamiltonian H. Since the first Hamiltonian comprehensively considers the risk (derived from the covariance matrix) and return (future return vector) factors of the stock in this investment model, the expected value reflects the initial parameters of the variable quantum circuit ansatz in the current quantum state. Under the decision, the average situation after comprehensively considering risk and return factors (i.e., comprehensive value information) is achieved by continuously optimizing the initial parameters of the variable quantum circuit during the training process. This is done to ensure that the expected value meets specific requirements in order to find a better portfolio solution.

[0055] It should be noted that, in the embodiments of this application, the preset optimization algorithm may include Adaptive Moment Estimation (ADAM), Stochastic Gradient Descent (SGD), Constrained Optimization by Linear Approximation (COBYLA), etc. This application does not specifically limit the types of algorithms included in the preset optimization algorithm.

[0056] For example, in an embodiment of this application, the asset configuration device performs measurement processing on the first quantum circuit, and the resulting initial expected value can be as shown in the following formula (6).

[0057] (6) in, This indicates a quantum state. The initial expected value under the first Hamiltonian, This represents the quantum state output by the ansatz circuit. This represents the first Hamiltonian.

[0058] Optionally, in embodiments of this application, when the asset allocation device iteratively optimizes the first parameter of the second quantum circuit based on a preset optimization algorithm, it can optimize the first parameter based on optimization algorithms such as ADAM, SGD, and COBYLA. For example, it can continuously iteratively calculate the parameters on the variable quantum circuit. The value (i.e., the first parameter) is iteratively optimized until the initial expected value is less than a certain threshold (i.e., the second preset threshold), thereby obtaining the optimized first quantum circuit. This application does not specifically limit the size of the second preset threshold.

[0059] It should be noted that, in the embodiments of this application, after the asset allocation device optimizes the first parameter of the second quantum circuit to obtain the optimized first quantum circuit, it can perform measurement processing based on the optimized first quantum circuit to obtain the amplitude probability corresponding to each ground state; and then determine the weight information based on the amplitude probability corresponding to each ground state.

[0060] It should be noted that, in the embodiments of this application, in the portfolio scenario of quantum computing, the base (i.e., the ground state) can be represented by a binary string to indicate different stock portfolio states. Each binary bit corresponds to a stock, with a binary bit of 1 indicating investment in that stock and a binary bit of 0 indicating no investment in that stock. Taking two stocks as an example, all possible base combinations are 00 (no investment), 01 (invest only in the second stock), 10 (invest only in the first stock), and 11 (invest in both stocks). If there are four stocks, the base consists of 16 combinations from 0000 to 1111, with each string corresponding to a specific investment choice.

[0061] For example, in the embodiments of this application, when the asset allocation device performs measurement processing based on the optimized first quantum circuit to obtain the amplitude probability corresponding to each ground state, taking two stocks as an example, the amplitude probability corresponding to each ground state can be: 00 corresponds to an amplitude probability of 0.3, 01 corresponds to an amplitude probability of 0.2, 10 corresponds to an amplitude probability of 0.1, and 11 corresponds to an amplitude probability of 0.4. This application does not specifically limit the magnitude of the amplitude probability corresponding to the ground state.

[0062] Optionally, in the embodiments of this application, when the asset allocation device determines the weight information based on the amplitude probability corresponding to each ground state, it can determine the investment probability corresponding to each second asset based on the amplitude probability corresponding to each ground state; then, it can normalize the investment probability corresponding to each second asset to obtain the weight information corresponding to each second asset.

[0063] For example, in an embodiment of this application, when the asset allocation device determines the investment probability of each second asset based on the amplitude probability corresponding to each ground state, it can multiply the measured probability (i.e., amplitude probability) of each ground state by the binary bit (0 or 1) of the corresponding stock in that ground state, and then add the results of all ground states to obtain the total investment probability of the stock. For example, taking two stocks (i.e., second assets) as an example, the first stock: 00 (probability 0.3) × 0 + 01 (probability 0.2) × 0 + 10 (probability 0.1) × 1 + 11 (probability 0.4) × 1 = 0.5; the second stock: 00 (probability 0.3) × 0 + 01 (probability 0.2) × 1 + 10 (probability 0.1) × 0 + 11 (probability 0.4) × 1 = 0.6. Thus, the investment probability of the first stock is 0.5, and the investment probability of the second stock is 0.6.

[0064] For example, in an embodiment of this application, when the asset allocation device normalizes the investment probability corresponding to each second asset to obtain the weight information corresponding to each second asset, taking two stocks (i.e., second assets) as an example, it can calculate the total probability of the investment probability of the first stock and the investment probability of the second stock. The total probability is 0.5 + 0.6 = 1.1. Then the weight of the first stock is 0.5 ÷ 1.1 ≈ 45.5%, and the weight of the second stock is 0.6 ÷ 1.1 ≈ 54.5%. In this way, the weight information corresponding to each second asset can be obtained.

[0065] Optionally, in the embodiments of this application, after obtaining the weight information of each stock, it can be combined with the total investable funds, and the funds can be allocated according to the weight information of each stock (i.e., the optimal investment strategy). Then, it can be checked whether the combination result can obtain a high rate of return on funds. That is, the embodiments of this application can convert the probability result of quantum measurement into the expected proportion of each stock being selected, and then standardize it into the actual investment proportion.

[0066] It should be noted that in the embodiments of this application, a quantum algorithm—the Quantum Variational Quantum Eigensolver (VQE)—is used when calculating the portfolio. Its core idea is based on the variational principle in quantum mechanics: by constructing a quantum circuit with adjustable parameters (i.e., the first quantum circuit), its energy expectation (i.e., the initial expectation value) is measured on a quantum device, and the parameters are continuously adjusted by a classical optimizer (i.e., a preset optimization algorithm) to reduce the energy, eventually approximating the true ground state. Compared with classical algorithms, this algorithm can obtain the portfolio result faster and more efficiently, is better able to handle real-time portfolio optimization calculations, and can serve some financial service scenarios more quickly. This application provides an asset allocation method, which includes: acquiring first historical data information corresponding to one or more first assets, and determining predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructing a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; preparing a corresponding first quantum circuit based on the first Hamiltonian, and determining the weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information. Therefore, the embodiments of this application can construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset. That is, the first Hamiltonian constructed in this application is based on the predicted value information of the second asset, rather than the historical value information of the second asset. This can improve the accuracy of subsequent asset allocation. Then, a corresponding first quantum circuit can be prepared based on the first Hamiltonian, and the weight information corresponding to each second asset can be determined based on the first quantum circuit. In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit prepared based on the first Hamiltonian. The optimal investment strategy can be generated based on the weight information. The optimal investment strategy can be calculated by quantum circuit, which can meet the needs of large-scale real-time optimization and efficiently calculate the corresponding portfolio method, greatly accelerating the calculation time and thus improving the efficiency of asset allocation.

[0067] Based on the above embodiments, another embodiment of this application provides an asset allocation method, which can... Based on historical data of the candidate stocks (i.e., the first asset), the optimal portfolio plan (i.e., the optimal investment strategy) is calculated to better provide investment advice to users. This asset allocation method mainly consists of two parts. The first part is the classical part, which mainly consists of the FactorVAE algorithm. This part extracts time-series information from the historical data of the candidate stocks (i.e., the first asset) and then combines it with a dynamic factor model to use the VAE algorithm to predict the expected return of the corresponding stock (i.e., predicted value information). The second part is the quantum part, which mainly consists of the Variational Quantum Circuit (VQC) (i.e., the first quantum circuit). Based on the expected return (i.e., predicted value information) obtained from FactorVAE and the Sharpe ratio formula, a Hamiltonian (i.e., the first Hamiltonian) is constructed. The Hamiltonian is then encoded into the quantum circuit (i.e., the first quantum circuit) using Hamiltonian simulation technology. After that, an ansatz circuit (i.e., the second quantum circuit) is added for training, and finally the allocation weight of each candidate stock (i.e., the second asset) is obtained. By allocating the weights, the investment funds that each candidate stock can obtain can be obtained.

[0068] It should be noted that, in the embodiments of this application, Figure 4 This is a schematic diagram of the asset allocation method proposed in the embodiments of this application. Figure 2 ,like Figure 4 As shown, this asset allocation method mainly consists of two parts. The first part is the classic part, which mainly uses the FactorVAE algorithm to predict the expected return of stocks. This part is explained step by step: First, obtain the historical data information of the primary asset (e.g., the stocks to be selected); historical data information of the stocks... This typically includes historical stock pricing, stock risk metrics, and historical returns. The second step involves using a feature extractor to extract latent features. In the FactorVAE algorithm, feature extraction primarily uses GRU to extract historical time-series features from historical stock data. This process combines the historical stock data obtained in the previous step with... The first step involves a feature extractor that, relying on the GRU's ability to extract time-series features, extracts latent features *e* from historical stock data. The third step uses an encoder to combine these latent features with known future stock returns to extract a posterior factor. The encoder's role is to obtain a posterior factor by processing the latent features and historical stock returns from the historical data. This posterior factor conforms to ;in, Indicates a Gaussian distribution. Represents the posterior factor The mean of the normal distribution it follows. Represents the posterior factor The variance of the normal distribution it follows; the fourth step is to use a predictor, which inputs the hidden features of the historical data to obtain the prior factor. The predictor only inputs the hidden features e of the historical data. The predictor uses an attention mechanism and finally generates a prior factor. , ;in, Represents prior factors The mean of the normal distribution it follows. It is a priori factor The variance of the normal distribution it follows; the fifth step is to train the prior factors to reduce the loss function between the prior and posterior factors, thus obtaining the trained prior factors; the sixth step is to input the hidden features of the historical data and the trained prior factors into the decoder to obtain the predicted future return (i.e., the predicted value information); entering the decoder stage, here the hidden features of the historical data e and the trained prior factors must be input simultaneously. For the encoder of FactorVAE, it contains Layers and The first layer, specifically the principle formula is the above formula (1). The second part is the quantum part, which can construct the return vector (i.e., the value information vector) and the covariance matrix (i.e., the first matrix). Then, based on the first matrix and the value information vector, the first Hamiltonian can be constructed. Then, based on the first Hamiltonian, a variable quantum circuit (i.e., the first quantum circuit) can be built. Then, the first quantum circuit can be trained and optimized to obtain the investment weights of the candidate stocks (i.e., the second asset), thereby obtaining the portfolio scheme (i.e., the optimal investment strategy).

[0069] Optionally, in the embodiments of this application, before the asset allocation device builds a variable quantum circuit (i.e., the first quantum circuit) to obtain the investment weight of the candidate stocks and thus obtain the portfolio plan, the principle of the plan can be explained first: First, the Sharpe ratio is a classic indicator for measuring the risk-adjusted return of a portfolio. The formula for the Sharpe ratio is as follows (7). When the Sharpe ratio is higher, it means that the return is higher. The goal is to obtain a portfolio with a higher Sharpe ratio. It is difficult to directly use the formula for the Sharpe ratio for modeling. We can make the Sharpe ratio higher, which means that the numerator is higher and the denominator is lower. The following formula (8) can be constructed.

[0070] (7) in, Indicates the Sharpe ratio. Indicates the expected rate of return of the investment portfolio. Indicates the risk-free interest rate. This represents the standard deviation of portfolio returns.

[0071] (8) in, This represents a factor that adjusts the relationship between portfolio returns and variance.

[0072] Optionally, in the embodiments of this application, maximizing the above formula (8) can achieve the same effect, because It is a constant and can be ignored. However, when applied to quantum mechanics, this formula is usually rearranged into a Hamiltonian. The minimum value of the Hamiltonian is usually sought. A symbol can be added before the formula to transform the formula into a Hamiltonian, as shown in the following formula (9). Then, by implementing such a Hamiltonian on the quantum circuit, the modeling from the classical model to the quantum model can be initially achieved.

[0073] (9) in, Represents the Hamiltonian. This represents the covariance between stock i and stock j. This indicates the weight information of stock i and stock j in the investment portfolio. This represents the expected rate of return for stock i.

[0074] It should be noted that, in the embodiments of this application, the quantum part is described step by step below: Step 1: Cleaning out stocks with low returns. In the FactorVAE part, the future expected return (predicted value information) of each candidate stock (i.e., the first asset) is obtained. Some stocks may have relatively low returns, close to 0. In such cases, these data are preferentially removed from the dataset. For example, a threshold (i.e., the first preset threshold) can be set, assuming it is 0.01. When the return... The first step is to eliminate the corresponding stock and its return rate. The second step is to construct a return rate vector (i.e., a value information vector) for the remaining stocks (i.e., the second asset) and calculate the covariance matrix (i.e., the first matrix). The future returns of the remaining stocks can be arranged and each stock can be numbered, with each number corresponding to a ground state in the quantum state. Then, according to the numbering order, an expected return rate vector (i.e., a value information vector) can be constructed. The calculation of the covariance matrix needs to be combined with historical data (i.e., second historical data information). First, calculate the average historical return of each stock based on the historical data to obtain the average value vector (i.e., the mean vector). Calculate the difference between each historical data point and the corresponding mean, and then calculate the value of each element of the covariance matrix according to the above formula (2) to obtain the covariance matrix; the third step: construct the Hamiltonian according to the above description (i.e., formula (7)-formula (9)) and construct the Hamiltonian on the quantum circuit. In this part, the composition of the Hamiltonian is explained in more detail; the above covariance matrix (i.e., the first matrix) represents the risk factors of the stock in a practical sense. Convert it into a quadratic form using the Pauli matrix, as shown in the above formula (3), to obtain the second Hamiltonian; the linear part of the future return vector (i.e., the value information vector) can be converted into a quadratic form using the Pauli matrix. The linear mode of the li matrix, as shown in the above formula (4), can yield the third Hamiltonian; therefore, the Hamiltonian represented by the pauli matrix (i.e., the first Hamiltonian) is as shown in the above formula (5); then, based on the simulation technique of the Hamiltonian (i.e., the first Hamiltonian), the above Hamiltonian is prepared onto the circuit (i.e., the first quantum circuit), thus completing the preparation process of the Hamiltonian; the fourth step: add ansatz (i.e., the second quantum circuit) to the quantum circuit (i.e., the first quantum circuit) and train it. This part involves adding ansatz circuits to the quantum circuit for training. The ansatz circuit used in this part is a hardware-efficient ansatz template circuit, i.e., the simplified_TwoDesign circuit layer. This circuit layer is a simplified hardware-efficient type of ansatz circuit, which can perform calculations more efficiently on a quantum computer; then, the quantum circuit can be trained. First, the parameters of the variable quantum circuit are initialized. Generally, random initialization is performed, followed by a round of measurement of the circuit. The expected measurement value is shown in formula (6) above. During the training process, a classic optimizer (i.e., a preset optimization algorithm) can be selected to calculate the gradient descent, such as ADAM, SGD, COBYLA, etc. The training process is to iteratively calculate the parameters on the variable quantum circuit. value, The process is iterated until the expected value E is less than a certain threshold. The fifth step involves measuring the quantum circuit (i.e., the optimized first quantum circuit) to obtain the probability distribution of each ground state. Based on these probabilities, the investment weight of each stock (i.e., the second asset) is calculated. After the entire quantum circuit has been trained, measurements are taken to obtain an amplitude probability corresponding to each ground state. These probabilities are then converted into a weight vector. The weight vector is the expected weight result; Step 6: Verify the results. After obtaining the weight of each stock, combine it with the total investable funds to allocate funds and check whether the portfolio result can obtain a high rate of return.

[0075] For example, in an embodiment of this application, when the asset allocation device determines the investment probability of each second asset based on the amplitude probability corresponding to each ground state, it can multiply the measured probability (i.e., amplitude probability) of each ground state by the binary bit (0 or 1) of the corresponding stock in that ground state, and then add the results of all ground states to obtain the total investment probability of the stock. For example, taking two stocks (i.e., second assets) as an example, the first stock: 00 (probability 0.3) × 0 + 01 (probability 0.2) × 0 + 10 (probability 0.1) × 1 + 11 (probability 0.4) × 1 = 0.5; the second stock: 00 (probability 0.3) × 0 + 01 (probability 0.2) × 1 + 10 (probability 0.1) × 0 + 11 (probability 0.4) × 1 = 0.6. Thus, the investment probability of the first stock is 0.5, and the investment probability of the second stock is 0.6.

[0076] For example, in an embodiment of this application, when the asset allocation device normalizes the investment probability corresponding to each second asset to obtain the weight information corresponding to each second asset, taking two stocks (i.e., second assets) as an example, it can calculate the total probability of the investment probability of the first stock and the investment probability of the second stock. The total probability is 0.5 + 0.6 = 1.1. Then the weight of the first stock is 0.5 ÷ 1.1 ≈ 45.5%, and the weight of the second stock is 0.6 ÷ 1.1 ≈ 54.5%. In this way, the weight information corresponding to each second asset can be obtained.

[0077] Optionally, in the embodiments of this application, after obtaining the weight information of each stock, it can be combined with the total investable funds, and the funds can be allocated according to the weight information of each stock (i.e., the optimal investment strategy). Then, it can be checked whether the combination result can obtain a high rate of return on funds. That is, the embodiments of this application can convert the probability result of quantum measurement into the expected proportion of each stock being selected, and then standardize it into the actual investment proportion.

[0078] It should be noted that in the embodiments of this application, the relevant quantum algorithms applied in the financial field (such as the QAOA algorithm) directly use historical data of financial stocks for their datasets. However, this application uses predicted stock returns to calculate a better investment portfolio, resulting in higher accuracy. Compared to classical algorithms, the embodiments of this application utilize the characteristics of quantum computing to construct a Hamiltonian, enabling faster and more efficient acquisition of a better investment portfolio.

[0079] This application provides an asset allocation method, which includes: acquiring first historical data information corresponding to one or more first assets, and determining predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructing a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; preparing a corresponding first quantum circuit based on the first Hamiltonian, and determining the weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information. Therefore, the embodiments of this application can construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset. That is, the first Hamiltonian constructed in this application is based on the predicted value information of the second asset, rather than the historical value information of the second asset. This can improve the accuracy of subsequent asset allocation. Then, a corresponding first quantum circuit can be prepared based on the first Hamiltonian, and the weight information corresponding to each second asset can be determined based on the first quantum circuit. In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit prepared based on the first Hamiltonian. The optimal investment strategy can be generated based on the weight information. The optimal investment strategy can be calculated by quantum circuit, which can meet the needs of large-scale real-time optimization and efficiently calculate the corresponding portfolio method, greatly accelerating the calculation time and thus improving the efficiency of asset allocation.

[0080] Based on the above embodiments, this application provides an asset configuration device. Figure 5 Schematic diagram of the composition structure of equipment for asset allocation Figure 1 ,like Figure 5 As shown, the asset configuration device 10 includes: an acquisition unit 11, a construction unit 12, a preparation unit 13, and a determination unit 14; wherein, The acquisition unit 11 is used to acquire first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm. The construction unit 12 is used to construct a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second asset; wherein, the first Hamiltonian is used at least to determine the optimal asset portfolio, the first asset includes the second asset, and the first historical data information includes the second historical data information; The preparation unit 13 is used to prepare a corresponding first quantum circuit based on the first Hamiltonian. The determining unit 14 is used to determine the weight information corresponding to each of the second assets based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information.

[0081] In the embodiments of this application, further, Figure 6 Schematic diagram of the composition structure of equipment for asset allocation Figure 2 ,like Figure 6 As shown, the asset configuration device 10 proposed in this application embodiment may further include a processor 15, a memory 16 storing instructions executable by the processor 15, and further, the asset configuration device 10 may further include a communication interface 17 and a bus 18 for connecting the processor 15, the memory 16 and the communication interface 17.

[0082] In the embodiments of this application, the processor 15 can be at least one of the following: Application-Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field-Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that for different devices, the electronic device used to implement the above-mentioned processor function can also be other types, and this application embodiment does not specifically limit this. The asset configuration device 10 may also include a memory 16, which can be connected to the processor 15. The memory 16 is used to store executable program code, which includes computer operation instructions. The memory 16 may include high-speed RAM memory and may also include non-volatile memory, such as at least two disk drives.

[0083] In embodiments of this application, bus 18 is used to connect communication interface 17, processor 15, and memory 16, as well as the mutual communication between these devices.

[0084] In embodiments of this application, memory 16 is used to store instructions and data.

[0085] Furthermore, in the embodiments of this application, the processor 15 is configured to acquire first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; construct a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; prepare a corresponding first quantum circuit based on the first Hamiltonian, and determine the weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information.

[0086] In practical applications, the aforementioned memory 16 can be volatile memory, such as random-access memory (RAM); or non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid-state drive (SSD); or a combination of the above types of memory, and provide instructions and data to the processor 15.

[0087] This application provides an asset allocation device that acquires first historical data information corresponding to one or more first assets, and determines the predicted value information corresponding to the first assets based on the first historical data information and a first preset algorithm; constructs a first Hamiltonian based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second assets; wherein the first Hamiltonian is used at least to determine the optimal asset portfolio, the first assets include the second assets, and the first historical data information includes the second historical data information; prepares a corresponding first quantum circuit based on the first Hamiltonian, and determines the weight information corresponding to each second asset based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information. Therefore, the embodiments of this application can construct a first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset. That is, the first Hamiltonian constructed in this application is based on the predicted value information of the second asset, rather than the historical value information of the second asset. This can improve the accuracy of subsequent asset allocation. Then, a corresponding first quantum circuit can be prepared based on the first Hamiltonian, and the weight information corresponding to each second asset can be determined based on the first quantum circuit. In other words, the embodiments of this application can use the characteristics of quantum computing to construct the first Hamiltonian and determine the weight information corresponding to each second asset based on the first quantum circuit prepared based on the first Hamiltonian. The optimal investment strategy can be generated based on the weight information. The optimal investment strategy can be calculated by quantum circuit, which can meet the needs of large-scale real-time optimization and efficiently calculate the corresponding portfolio method, greatly accelerating the calculation time and thus improving the efficiency of asset allocation.

[0088] This application provides a computer-readable storage medium storing a program that, when executed by a processor, implements the asset allocation method described above.

[0089] Specifically, the program instructions corresponding to an asset configuration method in this embodiment can be stored on storage media such as optical discs, hard disks, and USB flash drives. When the program instructions corresponding to an asset configuration method in the storage media are read or executed by an electronic device, the following steps are included: Obtain first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and a first preset algorithm; A first Hamiltonian is constructed based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second asset; wherein, the first Hamiltonian is used at least to determine the optimal asset portfolio, the first asset includes the second asset, and the first historical data information includes the second historical data information; A first quantum circuit is prepared based on the first Hamiltonian, and weight information corresponding to each second asset is determined based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information.

[0090] This application also provides a computer program product, including a computer program that can be executed by the processor 15 of the asset configuration device 10 to perform the steps described in any of the foregoing methods.

[0091] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0092] This application is described with reference to schematic and / or block diagrams of implementations of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the schematic and / or block diagrams can be implemented by computer program instructions, and combinations of blocks in the schematic and / or block diagrams can be implemented. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the schematic and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0093] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in the implementation flow diagram. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0095] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. An asset allocation method, characterized in that, The method includes: Obtain first historical data information corresponding to one or more first assets, and determine the predicted value information corresponding to the first asset based on the first historical data information and a first preset algorithm; A first Hamiltonian is constructed based on the predicted value information corresponding to each second asset and the second historical data information corresponding to the second asset; wherein, the first Hamiltonian is used at least to determine the optimal asset portfolio, the first asset includes the second asset, and the first historical data information includes the second historical data information; A first quantum circuit is prepared based on the first Hamiltonian, and weight information corresponding to each second asset is determined based on the first quantum circuit, so as to generate an optimal investment strategy based on the weight information.

2. The method according to claim 1, characterized in that, After determining the predicted value information corresponding to the first asset based on the first historical data information and the first preset algorithm, the method further includes: Based on the predicted value information corresponding to the first asset and a first preset threshold, the first asset is filtered to obtain the filtered second asset; wherein... The second asset includes the first asset whose predicted value information is greater than or equal to the first preset threshold.

3. The method according to claim 1, characterized in that, The construction of the first Hamiltonian based on the predicted value information and the second historical data information corresponding to each second asset includes: A first matrix is ​​calculated based on each of the second historical data and the mean vector; wherein the mean vector contains the average value of the historical value information corresponding to each of the second assets, and the first matrix is ​​used to characterize the correlation risk between the second assets; Each predicted value information corresponding to the second asset is assigned a number, and a value information vector is constructed based on the number; wherein, the number corresponds to the ground state of the qubit; The first Hamiltonian is constructed based on the first matrix and the value information vector.

4. The method according to claim 3, characterized in that, The construction of the first Hamiltonian based on the first matrix and the value information vector includes: The second Hamiltonian is determined based on the first matrix and the first Pauli matrix; wherein the second Hamiltonian is used at least to characterize the risk information of the asset portfolio, and the first Pauli matrix is ​​used to characterize the ground state of the quantum bit; The third Hamiltonian is determined based on the value information vector and the second Pauli matrix; wherein the third Hamiltonian is used to characterize the value information of the asset portfolio. The first Hamiltonian is determined based on the second Hamiltonian and the third Hamiltonian.

5. The method according to claim 1, characterized in that, The first quantum circuit includes a second quantum circuit, the second quantum circuit being used at least for training the first quantum circuit, and the step of determining the weight information corresponding to each second asset based on the first quantum circuit includes: The first parameter of the second quantum circuit is optimized to obtain the optimized first quantum circuit; wherein the first parameter includes the initial parameter of the second quantum circuit. Based on the optimized first quantum circuit, the amplitude probability corresponding to each ground state is obtained by performing measurement processing. The weight information is determined based on the amplitude probability corresponding to each ground state.

6. The method according to claim 5, characterized in that, The optimization of the first parameter of the second quantum circuit to obtain the optimized first quantum circuit includes: The first quantum circuit is subjected to measurement processing to obtain an initial expected value; wherein, the initial expected value is used to characterize the comprehensive value information in the current quantum state; The first parameter of the second quantum circuit is iteratively optimized based on a preset optimization algorithm until the initial expected value is less than a second preset threshold. Then the iterative optimization stops, and the optimized first quantum circuit is obtained.

7. The method according to claim 5, characterized in that, Determining the weight information based on the amplitude probability corresponding to each ground state includes: The investment probability corresponding to each of the second assets is determined based on the amplitude probability corresponding to each of the ground states; The investment probability corresponding to each of the second assets is normalized to obtain the weight information corresponding to each of the second assets.

8. An asset allocation device, characterized in that, The asset configuration device includes: a processor and a memory; wherein... The memory is used to store computer programs that can run on the processor; The processor is configured to perform the method as described in any one of claims 1-7 when running the computer program.

9. A computer-readable storage medium, characterized in that, The storage medium stores computer program code, which, when executed by a computer, performs the method described in any one of claims 1-7.

10. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the method according to any one of claims 1-7.

Images