CN114036208A

CN114036208A - Model training and sensitivity analysis method, device, equipment and medium

Info

Publication number: CN114036208A
Application number: CN202111317599.8A
Authority: CN
Inventors: 陈嘉瑞
Original assignee: CCB Finetech Co Ltd
Current assignee: CCB Finetech Co Ltd
Priority date: 2021-11-09
Filing date: 2021-11-09
Publication date: 2022-02-11

Abstract

The invention discloses a model training and sensitivity analysis method, a device, equipment and a medium. The original sensitivity analysis model can be trained according to the acquired first basic data of the first time, the daily first behavior data of each fund in a preset second time period before the first time in the sample set and the sensitivity label corresponding to each first behavior data respectively, so that the trained sensitivity analysis model is acquired, and the subsequent sensitivity analysis model completed through training can be organically combined with market risk factors when the fund is subjected to sensitivity analysis, so that the fund is accurately subjected to sensitivity analysis. Moreover, the sensitivity analysis of the fund is not required to be carried out manually by the staff, the manual workload is reduced, the influence of the manual efficiency and the accuracy on the sensitivity analysis of the fund is avoided, and the efficiency and the accuracy of the sensitivity analysis of the fund are improved.

Description

Model training and sensitivity analysis method, device, equipment and medium

Technical Field

The invention relates to the technical field of information processing, in particular to a method, a device, equipment and a medium for model training and sensitivity analysis.

Background

At present, the sensitivity analysis of the fund is of great significance for the expectation of the change of the value of the fund. The sensitivity analysis refers to how much the value (V) of the expected fund changes if one of the market risk factors (f) changes.

In the prior art, the sensitivity of the fund is generally analyzed manually according to the acquired market risk factors. The method has higher requirements on professional knowledge of workers who perform sensitivity analysis, improves the difficulty of sensitivity analysis, and greatly influences the efficiency and accuracy of sensitivity analysis because the efficiency and accuracy of sensitivity analysis completely depend on the efficiency and accuracy of the workers. Therefore, how to quickly and accurately determine the sensitivity of the fund is a technical problem to be solved urgently.

Disclosure of Invention

The embodiment of the invention provides a method, a device, equipment and a medium for model training and sensitivity analysis, which are used for solving the problems of low accuracy and low efficiency of the existing method for determining the sensitivity of fund.

The embodiment of the invention provides a sensitivity analysis model training method, which comprises the following steps:

splicing the acquired first basic data at the first time with the first behavior data of any fund in the sample set at the first time to determine sample input data of the fund; the first quotation data is daily quotation data of the fund in a preset first time period before the first time; the first behavior data corresponds to a sensitivity label which is used for representing the real sensitivity of the fund in a preset second time period after the first time; the first base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

acquiring, by an original sensitivity analysis model, a predicted sensitivity of the fund within a preset second time period after the first time based on the sample input data;

and training the original sensitivity analysis model based on the predicted sensitivity and the corresponding real sensitivity to obtain a trained sensitivity analysis model.

Further, before the stitching the first behavior data and the first basic data and determining the sample input data of the fund, the method further includes:

determining a first high-order index according to the first basic data;

the splicing the first behavior data and the first basic data to determine the sample input data of the fund includes:

and splicing the first behavior data, the first high-order index and the first basic data to determine the sample input data.

Further, the determining a first higher-order exponent according to the first basic data includes:

if the first high-order index includes a stock market volatility index and the first basic data includes a daily testimony index within the preset seventh time period, determining that the stock market volatility index can be determined according to the first basic data by the following formula:

wherein sv is_tRepresenting the stock market volatility index, s_tCollecting price for the number of the upper syndrome indexes in the daily number of upper syndrome indexes within the preset seventh time period, wherein T is more than or equal to 250 and is less than or equal to T, T is the preset seventh time period, and T is a positive integer more than or equal to 250;

if the first high-order index comprises a term difference index, and the first basic data comprises the average daily return rate of the Chinese debt in the preset fifth time period and the daily return rate of the Chinese debt in the preset sixth time period, determining the term difference index according to the average daily return rate of the Chinese debt in the preset fifth time period and the daily return rate of the Chinese debt in the preset sixth time period;

if the first high-order index comprises a banking risk tolerance index, and the first basic data comprises the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period, determining the banking risk tolerance index according to the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period.

Further, before the stitching the first behavior data, the first high-order index and the first basic data and determining the sample input data of the fund, the method further includes:

and normalizing the first basic data and the first high-order index.

The embodiment of the invention also provides a sensitivity analysis method, which comprises the following steps:

acquiring second basic data of a second time and second market data of the fund to be analyzed at the second time; the second market data is daily market data of the fund in a preset first time period before the second time; the second base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

splicing the second basic data and the second market data to determine the input data of the fund to be analyzed;

and acquiring the predicted sensitivity of the fund to be analyzed in a preset second time period after the second time based on the input data through a pre-trained sensitivity analysis model.

The embodiment of the invention also provides a device for training the sensitivity analysis model, which comprises:

the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for splicing acquired first basic data at a first time with first behavior data of any fund in a sample set at the first time to determine sample input data of the fund; the first quotation data is daily quotation data of the fund in a preset first time period before the first time; the first behavior data corresponds to a sensitivity label which is used for representing the real sensitivity of the fund in a preset second time period after the first time; the first base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

the training unit is used for acquiring the predicted sensitivity of the fund in a preset second time period after the first time based on the sample input data through an original sensitivity analysis model; and training the original sensitivity analysis model based on the predicted sensitivity and the corresponding real sensitivity to obtain a trained sensitivity analysis model.

Further, the obtaining unit is further configured to determine a first high-order index according to the first basic data; and splicing the first behavior data, the first high-order index and the first basic data to determine the sample input data.

Further, the obtaining unit is specifically configured to determine, according to the first basic data, that the stock market volatility index is determinable according to the following formula if the first high-order index includes a stock market volatility index and the first basic data includes a daily testimony index within the preset seventh time period:

wherein sv is_tRepresenting the stock market volatility index, s_tTo said isSetting a daily closing price of a proof-giving index T in the proof-giving indexes every day in a seventh time period, wherein T is more than or equal to 250 and is less than or equal to T, T is the preset seventh time period, and T is a positive integer more than or equal to 250; or, if the first high-order index includes a term difference index, and the first basic data includes the average daily return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period, determining the term difference index according to the average daily return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period; or, if the first high-order index includes a banking risk tolerance index, and the first basic data includes the average daily return rate of the national bonds in the preset fifth time period and the average daily return rate of the national bonds in the preset fourth time period, determining the banking risk tolerance index according to the average daily return rate of the national bonds in the preset fifth time period and the average daily return rate of the national bonds in the preset fourth time period.

Further, the obtaining unit is further configured to splice the first behavior data, the first high-order index, and the first basic data, and perform normalization processing on the first basic data and the first high-order index before determining sample input data of the fund.

The embodiment of the invention also provides a sensitivity analysis device, which comprises:

the acquisition module is used for acquiring second basic data of a second time and second market data of the fund to be analyzed at the second time; the second market data is daily market data of the fund in a preset first time period before the second time; the second base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

the splicing module is used for splicing the second basic data and the second market data and determining the input data of the fund to be analyzed;

and the processing module is used for acquiring the predicted sensitivity of the fund to be analyzed in a preset second time period after the second time based on the input data through a pre-trained sensitivity analysis model.

An embodiment of the present invention further provides an electronic device, where the electronic device includes a processor, and the processor is configured to implement, when executing a computer program stored in a memory, any of the steps of the sensitivity analysis model training method described above, or implement the steps of the sensitivity analysis method described above.

An embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program, and the computer program, when executed by a processor, implements the steps of any of the above-mentioned sensitivity analysis model training methods, or implements the steps of the above-mentioned sensitivity analysis method.

Because the embodiment of the invention can acquire the first basic data with the first time in advance, then the original sensitivity analysis model can be trained according to the first basic data, the daily first behavior data of each fund in the sample set in the preset second time period before the first time and the sensitivity label corresponding to each first behavior data respectively so as to acquire the trained sensitivity analysis model, the trained sensitivity analysis model not only can consider the influence of the behavior data of the fund on the sensitivity of the fund, but also can fully consider the influence of the first basic data on the sensitivity of the fund, so that the sensitivity of the fund can be accurately determined by the sensitivity analysis model which is trained in advance, and workers do not need to analyze the sensitivity of the fund manually, thereby reducing the workload of manual work, the influence of manual efficiency and accuracy on the sensitivity analysis of the fund is avoided, and therefore the efficiency and accuracy of the sensitivity analysis of the fund are improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a schematic diagram of a sensitivity analysis model training process according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a training process of a specific sensitivity analysis model according to an embodiment of the present invention;

fig. 3 is a schematic flow chart of acquiring sample input data and a sensitivity label corresponding to the sample input data according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a sensitivity analysis process according to an embodiment of the present invention;

FIG. 5 is a diagram illustrating a specific sensitivity analysis process according to an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of a sensitivity analysis model training apparatus according to an embodiment of the present invention;

fig. 7 is a schematic structural diagram of a sensitivity analysis apparatus according to an embodiment of the present invention;

fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention;

fig. 9 is a schematic structural diagram of another electronic device according to an embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the attached drawings, and it should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

As will be appreciated by one skilled in the art, embodiments of the present invention may be embodied as a system, apparatus, device, method, or computer program product. Thus, the present invention may be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.

In this document, it is to be understood that any number of elements in the figures are provided by way of illustration and not limitation, and any nomenclature is used for differentiation only and not in any limiting sense.

For convenience of understanding, some concepts involved in the embodiments of the present invention are explained below:

deep learning: deep learning is to learn the intrinsic rules and the expression levels of sample input data, and information obtained in the learning process is very helpful for the interpretation of data such as characters, images and sounds. The final aim of the method is to enable the machine to have the analysis and learning capability like a human, and to recognize data such as characters, images and sounds.

Financial assets: the property which is owned by a unit or an individual and exists in the form of value is an intangible right to ask for physical property, and is a general term of all financial tools which can be traded in an organized financial market and have real price and future valuation. The biggest feature of a financial asset is the ability to provide its owner with an on-demand or on-demand monetary deposit flow in a market transaction.

And (3) risk measurement: the risk is quantitatively analyzed and described on the basis of risk identification, namely, on the basis of analysis of past loss data, the probability and mathematical statistics method is used for quantitatively analyzing and predicting the occurrence probability of the risk accident and the severity of the loss possibly caused after the risk accident occurs.

Financial risk: the types can be classified into market risk, credit risk, liquidity risk, operational risk, and the like. Market risk is also referred to as systemic risk, and the measure of systemic risk is generally described in terms of two aspects: sensitivity and volatility.

Sensitivity analysis: refers to how much the value (V) of the portfolio is expected to change if one of the market risk factors (f) changes.

Market risk factors: refers to the variables that exist in the market from which the value of a financial asset can be derived. The major market risk factors include interest rates, credit spreads, stock prices, exchange rates, implied volatility, prices for products in circulation (e.g., gold and oil), and the like.

Example 1:

fig. 1 is a schematic diagram of a sensitivity analysis model training process provided in an embodiment of the present invention, where the process includes:

s101: splicing the acquired first basic data at the first time with the first behavior data of any fund in the sample set at the first time to determine sample input data of the fund; the first quotation data is daily quotation data of the fund in a preset first time period before the first time; the first behavior data corresponds to a sensitivity label which is used for representing the real sensitivity of the fund in a preset second time period after the first time; the first base data includes at least one of: the interest rate of the same industry per day in the third time period is preset, the average return rate of the national-level financial bonds per day in the fourth time period is preset, the average return rate of the national bonds per day in the fifth time period is preset, the return rate of the national bonds per day in the sixth time period is preset, and the accrual index in the seventh time period is preset.

The sensitivity analysis model training method provided by the embodiment of the invention is applied to electronic equipment, and the electronic equipment can be intelligent equipment, such as a computer, a mobile terminal and the like, and can also be a server and the like.

In the practical application process, the market data of the fund before a certain time has a certain influence on determining the sensitivity of the fund after a certain time. Therefore, in the embodiment of the present invention, in order to perform sensitivity analysis on the fund quickly and accurately, a sample set may be generated according to market data (for convenience of description, recorded as first market data) of one or more kinds of funds at different times collected in advance, so as to train the sensitivity analysis model according to the market data of different funds at different times in the sample set, that is, according to daily market data of different funds in a preset time period (for convenience of description, recorded as a preset first time period) before different times, train the sensitivity analysis model, thereby implementing the sensitivity analysis model completed according to the training, and accurately and quickly performing sensitivity analysis on the fund.

In a possible implementation manner, since there are generally a plurality of daily quotation data in a preset first time period before a certain time, the daily quotation data in the preset first time period may be sorted according to the time sequence, and the time sequence formed by the sorted daily quotation data is determined as the first quotation data of the fund at the time.

In order to ensure the accuracy of the sensitivity analysis model, the market data of any fund in the sample set at different times are respectively corresponding to a sensitivity label, and the sensitivity label is used for representing the real sensitivity of the fund in a preset time period (for convenience of description, recorded as a preset second time period) after the time, so that the sensitivity analysis model is trained based on the market data of different funds in the sample set at different times and the corresponding sensitivity labels thereof.

It should be noted that, the process of specifically obtaining the true sensitivity of the fund in the preset second time period after the time belongs to the prior art, and details are not described herein.

In a possible implementation manner, the sensitivity label may be directly represented by a value of the real sensitivity, so that the real sensitivity of a fund in a preset second time period after a certain time can be represented by the sensitivity label.

In another possible implementation manner, in order to reduce the influence of the sensitivity analysis model on the accuracy of the sensitivity analysis of the fund, the sensitivity label may be represented by a sensitivity level corresponding to the real sensitivity, so that the real sensitivity of a fund in a preset second time period after a certain time may be represented by the sensitivity label. When the sensitivity label corresponding to the first action data of a fund at a certain time is determined, according to each value range of the preset sensitivity grade, a target value range in which the real sensitivity of the fund is located in a preset second time period after the time is determined, and the sensitivity grade corresponding to the target value range is determined as the sensitivity label corresponding to the first action data of the fund at the time.

Optionally, the electronic device used for training the sensitivity analysis model may be the same as or different from the electronic device used for performing the sensitivity analysis.

As a possible implementation, the sensitivity of the fund after a certain time may also be affected by market risk factors such as daily equity interest rate, average daily return to maturity for national-level financial bonds, average daily return to maturity for national bonds, daily return to national bonds, and daily upper evidence-based index before that time. Therefore, in order to realize the sensitivity analysis of the fund, the influence of the market risk factor on the sensitivity of the fund can be considered, and in the embodiment of the present invention, basic data (for convenience of description, referred to as first basic data) at different times can be obtained. Wherein the first base data comprises at least one of: the interest rate of the same industry divorced every day in a preset time period (for convenience of description, recorded as a preset third time period), the average income rate of national-level financial bonds due every day in the preset time period (for convenience of description, recorded as a preset fourth time period), the average income rate of national bonds due every day in the preset time period (for convenience of description, recorded as a preset fifth time period), the income rate of national bonds every day in the preset time period (for convenience of description, recorded as a current preset time period), and the index of the same industry divorced every day in the preset time period (for convenience of description, recorded as a preset seventh time period).

For example, the first basic data includes interest rate of same industry split between Shanghai banks, average return rate of national-level financial bonds due in China in one to three years, average return rate of national bonds due in China in one to three years, return rate of national bonds in China in ten years, Shanghai's index, and monthly data of actual effective exchange rate of RMB within one week.

It should be noted that, in the preset first time period to the preset seventh time period, there may be a plurality of time periods with equal time lengths, for example, the preset first time period is the same as the preset second time period, and the time lengths of the time periods may not be equal.

Because the sensitivity analysis model is trained according to the first basic data and the market data of the fund, the sensitivity analysis model completed through training in the follow-up process can organically combine the market data of the fund with the market risk factors when the sensitivity analysis is carried out on the fund, thereby accurately carrying out the sensitivity analysis on the fund and further improving the accuracy of the sensitivity analysis on the fund.

In one possible embodiment, if the daily first basic data cannot be directly acquired, for example, only basic data counted in months or basic data counted in years can be acquired, the acquired basic data may be processed to acquire basic data counted in days, and the processed basic data may be determined as the first basic data, so as to acquire the first basic data counted in days.

For example, if the obtained basic data represents the average daily data condition in a certain time period, the basic data may be directly determined as the first basic data of each day in the time period, so as to obtain the first basic data of each day in the time period.

For example, if the obtained basic data is the average return-to-expiration rate of national-level financial bonds in january, which represents the average return-to-expiration rate of national-level financial bonds in the 1 month, the average return-to-expiration rate of national-level financial bonds in january may be determined as the average return-to-expiration rate of 1 month to 30 days per day.

For another example, if the obtained basic data represents the overall data condition in a certain time period, the ratio of the basic data to the number of days included in the time period may be determined as the first basic data of each day in the time period, so as to obtain the first basic data of each day in the time period.

In a possible embodiment, since the first basic data of any fund at a certain time is generally daily basic data, and the first basic data generally includes a plurality of daily basic data, each daily basic data may be sorted in order of time, and a time series formed by each sorted daily basic data is determined as the first basic data of the time.

In a possible implementation manner, due to different obtaining channels of the first basic data, for example, the first basic data counted by different financial institutions, different rules for counting the first basic data, and the like, value ranges of different first basic data may be different, for example, a value range of some first basic data is between 0 and 1, and a value range of some first basic data is between 0 and 100, so that in the process of training the sensitivity analysis model according to the first basic data in different value ranges, the first basic data with a large value range has a larger influence on the sensitivity analysis model, and the first basic data with a small value range has a smaller influence on the sensitivity analysis model, thereby reducing the accuracy of the sensitivity analysis model obtained by training. Therefore, in order to avoid the influence of training the sensitivity analysis model on the first basic data in different value ranges, in the embodiment of the present invention, normalization processing may be performed on the acquired first basic data. In a specific implementation process, after the first basic data is obtained, normalization processing may be performed on the first basic data according to a preset normalization function, for example, a sigmoid function, so as to limit a value range of the first basic data within a preset unified value range.

After the first behavior data of different funds at different times in the sample set is obtained based on the above embodiment, the first behavior data of any fund at a certain time (for convenience of description, denoted as a first time) may be obtained from the sample set, and the first basic data at the same time, that is, the first basic data at the first time, may be obtained. And splicing the acquired first basic data at the first time with the first behavior data of the fund at the first time in the sample set, so that the spliced data is determined as the sample input data of the fund. And subsequently training the sensitivity analysis model based on the sample input data of the fund and the corresponding sensitivity label to obtain the trained sensitivity analysis model.

S102: acquiring, by an original sensitivity analysis model, a predicted sensitivity of the fund within a preset second time period after the first time based on the sample input data.

After sample input data is acquired based on the above-described embodiments, the sample input data may be input into the raw sensitivity analysis model. And processing the sample input data through the original sensitivity analysis model to obtain the predicted sensitivity of the fund corresponding to the sample input data in a preset second time period after the time.

The original sensitivity analysis model may be a decision tree, a Logistic Regression (LR), a Naive Bayes (NB) classification algorithm, a Random Forest (RF) algorithm, a Support Vector Machine (SVM) classification algorithm, a Histogram of Oriented Gradients (HOG), a deep learning algorithm, or the like. The deep learning algorithm may include a neural Network, a deep neural Network, a Convolutional Neural Network (CNN), and the like.

In a possible implementation manner, in order to reduce the influence of the sensitivity analysis model on the accuracy of the sensitivity analysis of the fund, if the sensitivity label can be represented by the sensitivity level corresponding to the real sensitivity, the original sensitivity analysis model is used to process the sample input data, so as to obtain the sensitivity level of the fund corresponding to the sample input data in a preset second time period after the time, and characterize the predicted sensitivity of the fund in the preset second time period after the time according to the sensitivity level and the value range corresponding to each preset sensitivity level.

S103: and training the original sensitivity analysis model based on the predicted sensitivity and the corresponding real sensitivity to obtain a trained sensitivity analysis model.

After the predicted sensitivity of the fund corresponding to the sample input data in the preset second time period after the first time is determined, because the sensitivity label is corresponding to the first behavior data of the first time contained in the sample input data, namely the true sensitivity is corresponding to the first behavior data, the loss value can be calculated according to the obtained predicted sensitivity and the obtained true sensitivity. And training the original sensitivity analysis model according to the loss value so as to update parameter values of parameters in the original sensitivity analysis model, thereby obtaining the trained sensitivity analysis model.

In specific implementation, when the original sensitivity analysis model is trained according to the loss value, a gradient descent algorithm can be adopted to perform back propagation on the gradient of the parameter in the original sensitivity analysis model, so that the original sensitivity analysis model is trained.

Because the sample set contains a large amount of first behavior data of different funds at different times, the operations are carried out on the first behavior data of each fund at each time, and when a preset convergence condition is met, the training of the sensitivity analysis model is finished.

The condition that the preset convergence condition is satisfied may be that a loss value calculated according to the obtained prediction sensitivity and the true sensitivity is smaller than a set loss value threshold, the number of iterations for training the model reaches a set maximum number of iterations, and the like. The specific implementation can be flexibly set, and is not particularly limited herein.

As a possible implementation manner, when performing model training, the first behavior data in the sample set may be divided into training behavior data and testing behavior data, and the original sensitivity analysis model is trained based on the training behavior data, and then the reliability of the trained sensitivity analysis model is verified based on the testing behavior data.

By training the original sensitivity analysis model through iteration, the trained sensitivity analysis model has higher accuracy and smaller generalization error and higher robustness.

Example 2:

to further improve the accuracy of the sensitivity model, in an embodiment of the present invention based on the above embodiment, before the splicing the first behavior data and the first basic data and determining the sample input data of the fund, the method further includes:

determining a first high-order index according to the first basic data;

In order to further improve the accuracy of the sensitivity model, in the embodiment of the present invention, the first basic data acquired at the first time may be subjected to characterization learning, and a high-order feature included in the first basic data is acquired, so as to improve a utilization value of the first basic data, and the sensitivity model may be trained through the high-order feature.

In a possible implementation manner, in order to accurately acquire the high-order features included in the first basic data, in an embodiment of the present invention, the first high-order index may be determined according to the acquired first basic data.

Due to the fact that the first basic data comprise different data, the first high-order index is determined according to the acquired first basic data. The following exemplifies a case where the first high-order index is determined according to the acquired first basic data:

in case one, if the first high-order index includes a stock market volatility index and the first basic data includes the daily shanghai index within the preset seventh time period, the stock market volatility index may be determined according to the daily shanghai index within the preset seventh time period.

In a possible embodiment, the stock market volatility index is determined according to the daily shanghai index in the preset seventh time period by the following formula:

wherein sv is_tRepresenting the stock market volatility index, s_tAnd closing the price for the daily upper syndrome index T in the daily upper syndrome index in the preset seventh time period, wherein T is more than or equal to 250 and is less than or equal to T, T is the preset seventh time period, and T is a positive integer more than or equal to 250.

And in the second situation, if the first high-order index comprises a term difference index, and the first basic data comprises the daily average return rate of the national debt in the preset fifth time period and the daily return rate of the national debt in the preset sixth time period, determining the term difference index according to the daily average return rate of the national debt in the preset fifth time period and the daily return rate of the national debt in the preset sixth time period.

In a possible implementation manner, a difference (for convenience of explanation, noted as a first difference) between the average daily return rate of the debt due in the preset fifth time period and the daily return rate of the debt in the preset sixth time period may be determined. The first difference is determined as the deadline interest index.

For example, if the average daily return rate of the national debt in the fifth time period is preset to be the average return rate of the national debt in the first to third years in china, and the daily return rate of the national debt in the sixth time period is preset to be the return rate of the national debt in the decade in china, the average return rate of the national debt in the first to third years in china and the first difference between the return rate of the national debt in the decade in china can be determined as the term profit-difference index.

And thirdly, if the first high-order index comprises a banking risk tolerance index, and the first basic data comprises the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period, determining the banking risk tolerance index according to the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period.

In one possible implementation, the difference (for convenience of description, denoted as the second difference) between the daily average return to maturity of the national-level financial bonds in the preset fourth time period and the daily average return to maturity of the national bonds in the preset fifth time period may be determined. And determining the second difference as the banking risk interest difference index.

For example, if the average daily return rate of national-level financial bonds in the fourth time period is the average daily return rate of national-level financial bonds in the first to third years in china, and the average daily return rate of national bonds in the fifth time period is the average daily return rate of national bonds in the first to third years in china, the second difference between the average daily return rate of national-level financial bonds in the first to third years in china and the average daily return rate of national bonds in the first to third years in china is determined as the banking risk difference index.

In a possible implementation manner, due to different acquisition channels of the first basic data, for example, the first basic data counted by different financial institutions, and different rules for counting the first basic data, the value ranges of different first basic data may be different, for example, the value range of some first basic data is between 0 and 1, and the value range of some first basic data is between 0 and 100, so that according to the first basic data of different value ranges, the determined value range of the first high-order index may also be different, resulting in that according to the first high-order index of the different value ranges, in the process of training the sensitivity analysis model, the first high-order index with a large value range has a greater influence on the sensitivity analysis model, and the first high-order index with a small value range has a smaller influence on the sensitivity analysis model, thereby reducing the accuracy of the trained sensitivity analysis model. Therefore, in order to avoid the influence of training the sensitivity analysis model on the first high-order index in different value ranges, in the embodiment of the present invention, normalization processing may be performed on the obtained first high-order index. In a specific implementation process, after the first high-order index is obtained, normalization processing may be performed on the first high-order index according to a preset normalization function, for example, a sigmoid function, so as to limit a value range of the first high-order index within a preset unified value range.

After the first high-order index is obtained based on the above embodiment, the first basic data at the first time, the first high-order index corresponding to the first basic data, and the first behavior data of a fund in the sample set at the first time may be spliced to determine the sample input data of the fund. And then inputting the sample input data into the original sensitivity analysis model for processing, thereby realizing the training of the sensitivity analysis model.

According to the embodiment of the invention, the first high-order index is determined according to the selected first basic data, and the first basic data and the first high-order index are embedded into the first behavior data of a certain fund at the first time by using a character embedding thought, so that the organic combination of the behavior data of the fund and the market risk factor is realized. And training the sensitivity analysis model according to the first basic data of the first time, the corresponding first high-order index and the sample input data determined by the first behavior data of a fund in the sample set at the first time, so that the improvement of the precision of the sensitivity analysis model is facilitated, the subsequent sensitivity analysis of the fund is facilitated through the sensitivity analysis model, and the accuracy of the sensitivity analysis of the fund is improved.

Example 3:

the method for training the sensitivity analysis model provided by the embodiment of the present invention is explained below by a specific embodiment, and fig. 2 is a schematic diagram of a training process of the specific sensitivity analysis model provided by the embodiment of the present invention, where the process includes:

s201: and acquiring sample input data and a corresponding sensitivity label.

For convenience of describing the process of specifically acquiring the sample input data and the sensitivity label corresponding thereto, the detailed description is made with reference to fig. 3. Fig. 3 is a schematic flowchart of a specific process for acquiring sample input data and a sensitivity label corresponding to the sample input data according to an embodiment of the present invention, where the process includes:

s301: basic data of a first time is obtained.

Wherein the base data includes at least one of: the interest rate of the same industry divorced in the third time period is preset, the average return rate of the national-level financial bonds in the fourth time period is preset, the average return rate of the national bonds in the fifth time period is preset, the return rate of the national bonds in the sixth time period is preset, and the upper evidence index in the seventh time period is preset.

S302: and if the basic data of the first time is not the daily basic data of the first time, processing the basic data of the first time to obtain the basic data counted by taking the day as a unit, and determining the processed basic data as the first basic data so as to obtain the first basic data counted by taking the day as a unit.

In a possible implementation manner, if the acquired basic data represents the average daily data condition in a certain time period, the basic data can be directly determined as the first basic data of each day in the time period, so as to acquire the first basic data of each day in the time period.

In another possible implementation, if the acquired basic data represents the overall data condition in a certain time period, the ratio of the basic data to the number of days included in the time period may be determined as the first basic data of each day in the time period, so as to achieve the acquisition of the first basic data of each day in the time period.

S303: and determining a first high-order index according to the acquired first basic data.

The process of determining a high-order index according to the obtained first basic data is explained in the above embodiments, and is not described herein again.

S304: and normalizing the first basic data and the first high-order index.

S305: and splicing the first behavior data, the first high-order index and the first behavior data of any fund in the sample set at the first time to determine sample input data.

S306: and determining a sensitivity label corresponding to the first action data of the fund at the first time.

The steps S301 to S306 may be executed for the first behavior data of each fund in the sample set at the first time.

S202: the acquired sample input data is divided into training sample input data and training sample input data.

S203: and training the original sensitivity analysis model based on the input data of each training sample and the sensitivity labels respectively corresponding to the input data.

Specifically, any training sample input data is obtained. And acquiring the predicted sensitivity of the fund corresponding to the input data of the training sample in a preset second time period after the first time based on the input data of the training sample through an original sensitivity analysis model. And training the original sensitivity analysis model based on the predicted sensitivity and the corresponding real sensitivity to obtain a trained sensitivity analysis model.

S204: and testing the trained sensitivity analysis model based on the input data of each test sample and the sensitivity labels respectively corresponding to the input data.

S205: and determining whether the trained sensitivity analysis model reaches a preset convergence condition or not based on the test result obtained in the step S204, if so, executing the step S206, and otherwise, executing the step S207.

S206: adjusting parameter values of parameters of the trained sensitivity analysis model, updating the original sensitivity analysis model according to the adjusted sensitivity analysis model, and executing S203.

S207: and determining the trained sensitivity analysis model as a trained sensitivity analysis model and storing the trained sensitivity analysis model.

According to the embodiment of the invention, the first basic data with the first time can be obtained in advance, and the first high-order index is determined according to the first basic data, so that the utilization value of the first basic data is improved. And then, according to the first basic data and the corresponding first high-order index thereof, daily first behavior data of each fund in a preset second time period before the first time in the sample set and the sensitivity label corresponding to each first behavior data, training the original sensitivity analysis model to obtain a trained sensitivity analysis model, wherein the trained sensitivity analysis model not only can consider the influence of the behavior data of the fund on the sensitivity of the fund, but also can fully consider the influence of the first basic data on the sensitivity of the fund, so that the organic combination of the behavior data of the fund and the market risk factor is realized. And training the sensitivity analysis model according to the first basic data of the first time, the corresponding first high-order index and the sample input data determined by the first behavior data of a fund in the sample set at the first time, so that the improvement of the precision of the sensitivity analysis model is facilitated, the subsequent sensitivity analysis of the fund is facilitated through the sensitivity analysis model, and the accuracy of the sensitivity analysis of the fund is improved. And the staff does not need to carry out sensitivity analysis on the fund manually, so that the workload of manpower is reduced, the requirement of professional knowledge storage required by the staff when carrying out sensitivity analysis on the fund is lowered, the influence of manual efficiency and accuracy on the sensitivity analysis of the fund is avoided, and the efficiency and accuracy of the sensitivity analysis on the fund are improved. In addition, because of the high uncertainty of the financial market, the sensitivity analysis of the fund generally has time-varying property, namely the sensitivity of one fund is different in different time periods, and the sensitivity analysis model obtained by training through the embodiment of the invention can fully utilize the characteristics of the time dimension in the sample input data, thereby better and accurately determining the sensitivity of the fund.

Example 4:

the embodiment of the present invention further provides a sensitivity analysis method, and fig. 4 is a schematic diagram of a sensitivity analysis process provided by the embodiment of the present invention, where the process includes:

s401: acquiring second basic data of a second time and second market data of the fund to be analyzed at the second time; the second market data is daily market data of the fund in a preset first time period before the second time; the second base data includes at least one of: the interest rate of the same industry per day in the third time period is preset, the average return rate of the national-level financial bonds per day in the fourth time period is preset, the average return rate of the national bonds per day in the fifth time period is preset, the return rate of the national bonds per day in the sixth time period is preset, and the evidence-taking index per day in the seventh time period is preset.

S402: and splicing the second basic data and the second market data to determine the input data of the fund to be analyzed.

S403: and acquiring the predicted sensitivity of the fund to be analyzed in a preset second time period after the second time based on the input data through a pre-trained sensitivity analysis model.

The sensitivity analysis method provided by the embodiment of the invention is applied to electronic equipment, and the electronic equipment can be intelligent equipment such as a robot and the like and can also be a server.

The electronic device for performing sensitivity analysis in the embodiment of the present invention may be the same as or different from the electronic device for performing sensitivity analysis model training.

In one possible embodiment, the sensitivity analysis model is trained in an off-line manner, which is generally used in the process of training the sensitivity analysis model. After the trained sensitivity analysis model is obtained, the sensitivity analysis model can be stored in the electronic equipment for sensitivity analysis, so that the electronic equipment for sensitivity analysis can conveniently perform sensitivity analysis on the fund through the pre-trained sensitivity analysis model.

When a worker wishes to perform sensitivity analysis on a fund, an analysis request for performing sensitivity analysis on the fund can be input through the intelligent device, so that the intelligent device can be controlled to perform sensitivity analysis on the fund through the analysis request. The analysis request carries the identifier of the fund and time (for convenience of description, denoted as second time).

It should be noted that there are many specific ways to input an analysis request, for example, the way to input the analysis request may be input by inputting voice information, or by operating a virtual button displayed on a display screen of the smart device, and the like, and the specific implementation process may be flexibly set according to the requirement, and is not limited specifically herein. After the intelligent device obtains the analysis request, the analysis request can be sent to the electronic device for sensitivity analysis.

After receiving the analysis request, the electronic device performing sensitivity analysis may analyze the analysis request, and obtain the identifier of the fund and the second time carried in the analysis request, so as to determine which fund is currently analyzed for the sensitivity within a preset second time period after which time. And then determining the fund as the fund to be analyzed, acquiring second market data of the fund to be analyzed at a second time, processing the second market data through a pre-trained sensitivity analysis model, and acquiring the predicted sensitivity of the fund to be predicted in a preset second time period after the second time.

It should be noted that the specific training process of the pre-trained sensitivity analysis model has been described in the above embodiments, and repeated details are not repeated.

As a possible implementation, the sensitivity of the fund after a certain time may also be affected by market risk factors such as daily equity interest rate, average daily return to maturity for national-level financial bonds, average daily return to maturity for national bonds, daily return to national bonds, and daily upper evidence-based index before that time. Therefore, in order to realize the sensitivity analysis of the fund, the influence of the market risk factor on the sensitivity of the fund can be considered, and in the embodiment of the present invention, when the sensitivity analysis of the fund to be analyzed is performed, the basic data (for convenience of description, referred to as the second basic data) at the second time can be further acquired. Wherein the second base data comprises at least one of: the interest rate of the same industry per day in the third time period is preset, the average return rate of the national-level financial bonds per day in the fourth time period is preset, the average return rate of the national bonds per day in the fifth time period is preset, the return rate of the national bonds per day in the sixth time period is preset, and the evidence-taking index per day in the seventh time period is preset.

For example, the second basic data includes interest rate of same industry split between Shanghai banks, average return rate of national-level financial bonds due in China in one to three years, average return rate of national bonds due in China in one to three years, return rate of national bonds in China in ten years, Shanghai's index, and monthly data of actual effective exchange rate of RMB within one week.

Because the sensitivity analysis is carried out on the fund to be analyzed according to the second basic data and the second market data of the fund to be analyzed, the market data of the fund and the market risk factors can be organically combined, the sensitivity analysis is accurately carried out on the fund to be analyzed, and the accuracy of the sensitivity analysis on the fund to be analyzed is further improved.

In one possible embodiment, if the second basic data for each day cannot be directly acquired, for example, only basic data for statistics in a month unit or basic data for statistics in a year unit can be acquired, the acquired basic data for the second time may be processed to acquire basic data for statistics in a day unit, and the processed basic data for each day may be determined as the second basic data, thereby acquiring the second basic data for statistics in a day unit.

For example, if the obtained basic data represents the average daily data condition in a certain time period, the basic data may be directly determined as the second basic data for each day in the time period, so as to obtain the second basic data for each day in the time period.

For another example, if the obtained basic data represents the overall data condition in a certain time period, the ratio of the basic data to the number of days included in the time period may be determined as the second basic data for each day in the time period, so as to obtain the second basic data for each day in the time period.

In a possible embodiment, since the second basic data of any fund at the second time is generally daily basic data, and the second basic data generally includes a plurality of daily basic data, each daily basic data may be sorted in time order, and the time series formed by the sorted daily basic data may be determined as the second basic data of the second time.

In a possible implementation manner, due to different obtaining channels of the second basic data, for example, the second basic data counted by different financial institutions, different rules for counting the second basic data, and the like, value ranges of different second basic data may be different, for example, a value range of some second basic data is between 0 and 1, and a value range of some second basic data is between 0 and 100, so that in the process of performing sensitivity analysis on the fund to be analyzed according to the second basic data in different value ranges, the second basic data with a large value range has a larger influence on determining the sensitivity of the fund to be analyzed, and the second basic data with a small value range has a smaller influence on determining the sensitivity of the fund to be analyzed, thereby reducing the accuracy of determining the sensitivity of the fund to be analyzed. Therefore, in order to further accurately determine the sensitivity of the fund to be analyzed, in the embodiment of the present invention, normalization processing may be performed on the acquired second basic data. In a specific implementation process, after the second basic data is obtained, normalization processing may be performed on the second basic data according to a preset normalization function, for example, a sigmoid function, so as to limit a value range of the second basic data within a preset unified value range.

In order to further accurately determine the sensitivity of the fund to be analyzed, in the embodiment of the present invention, the characterization learning may be performed on the acquired second basic data at the second time, and the high-order features included in the second basic data are acquired, so as to improve the utility value of the second basic data, and to accurately determine the sensitivity of the fund to be analyzed through the high-order features.

In a possible implementation manner, in order to accurately acquire the high-order features included in the second basic data, in an embodiment of the present invention, the second high-order index may be determined according to the acquired second basic data.

And determining a second high-order index according to the acquired second basic data due to the difference of the data included in the second basic data. The following exemplifies a case where the second high-order index is determined according to the obtained second basic data:

in case one, if the second high-order index includes a stock market volatility index and the second basic data includes the daily shanghai index within the preset seventh time period, the stock market volatility index may be determined according to the daily shanghai index within the preset seventh time period.

And in the second situation, if the second high-order index comprises a deadline difference index and the second basic data comprises the daily average return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period, determining the deadline difference index according to the daily average return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period.

And in a third case, if the second high-order index comprises a banking risk tolerance index, and the second basic data comprises the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period, determining the banking risk tolerance index according to the daily average return rate from maturity of the national bond in the preset fifth time period and the daily average return rate from maturity of the national-level financial bond in the preset fourth time period.

In a possible implementation manner, due to different acquisition channels of the second basic data, for example, the second basic data counted by different financial institutions, different rules for counting the second basic data, and the like, the value ranges of different second basic data may be different, for example, the value range of some second basic data is between 0 and 1, and the value range of some second basic data is between 0 and 100, so that the value ranges of the determined second high-order index may also be different according to the second basic data of different value ranges, and therefore, according to the second basic data of different value ranges, in the process of performing sensitivity analysis on the fund to be analyzed, the second basic data with a large value range has a larger influence on determining the sensitivity of the fund to be analyzed, and the second basic data with a small value range has a smaller influence on determining the sensitivity of the fund to be analyzed, thereby reducing the accuracy of determining the sensitivity of the fund to be analyzed. Therefore, in order to avoid the influence of the sensitivity on the fund to be analyzed due to the second high-order index in different value ranges, in the embodiment of the present invention, normalization processing may be performed on the obtained second high-order index as well. In a specific implementation process, after the second high-order index is obtained, normalization processing may be performed on the second high-order index according to a preset normalization function, for example, a sigmoid function, so as to limit a value range of the second high-order index within a preset unified value range.

After the second high-order index is obtained based on the above embodiment, the second basic data at the second time, the second high-order index corresponding to the second basic data, and the second market data of the fund to be analyzed may be spliced to determine the input data of the fund to be analyzed. And then inputting the input data into a pre-trained sensitivity analysis model for processing, thereby realizing the sensitivity analysis of the fund to be analyzed.

And processing the input data through the pre-trained sensitivity analysis model, so that the predicted sensitivity of the fund to be analyzed in a preset second time period after the second time can be obtained.

According to the embodiment of the invention, the second basic data with the second time can be obtained in advance, and the second high-order index is determined according to the second basic data, so that the utilization value of the second basic data is improved. And then, the second basic data, the corresponding second high-order index of the second basic data and the second market data of the fund to be analyzed at the second time are processed through a pre-trained sensitivity analysis model, so that the predicted sensitivity of the fund to be analyzed in the preset second time period after the second time can be obtained, and the market data of the fund and the market risk factor are organically combined. And performing sensitivity analysis on the fund to be analyzed according to the second basic data at the second time, the corresponding second high-order index and the input data determined by the second market data of the fund to be analyzed at the second time, so that the accuracy of determining the sensitivity of the fund to be analyzed is improved. And the staff does not need to carry out sensitivity analysis on the fund manually, so that the workload of manpower is reduced, the requirement of professional knowledge storage required by the staff when carrying out sensitivity analysis on the fund is lowered, the influence of manual efficiency and accuracy on the sensitivity analysis of the fund is avoided, and the efficiency and accuracy of the sensitivity analysis on the fund are improved. In addition, because of the high uncertainty of the financial market, the sensitivity analysis of the fund generally has time-varying property, namely the sensitivity of one fund is different in different time periods, and the sensitivity analysis model obtained by training through the embodiment of the invention can fully utilize the characteristics of the time dimension in the sample input data, thereby better and accurately determining the sensitivity of the fund.

Example 5:

the sensitivity analysis method provided by the embodiment of the present invention is described below with reference to fig. 5, where fig. 5 is a schematic diagram of a specific sensitivity analysis process provided by the embodiment of the present invention, where the process includes:

s501: and acquiring a pre-trained sensitivity analysis model.

It should be noted that, the process of specifically training the sensitivity analysis model has been described in the above embodiments, and is not described in detail here.

S502: and acquiring second basic data of a second time and second market data of the fund to be analyzed at the second time.

Wherein the second basic data includes at least one of: the interest rate of the same industry per day in the third time period is preset, the average return rate of the national-level financial bonds per day in the fourth time period is preset, the average return rate of the national bonds per day in the fifth time period is preset, the return rate of the national bonds per day in the sixth time period is preset, and the evidence-taking index per day in the seventh time period is preset.

S503: and determining a second high-order index according to the acquired second basic data.

The process of determining the second high-order index according to the obtained second basic data is explained in the above embodiments, and is not described herein again.

S504: and normalizing the second basic data and the second high-order index.

S505: and splicing the second market data, the second high-order index and the second market data to determine input data.

S506: and acquiring the predicted sensitivity of the fund to be analyzed in a preset second time period after the second time based on the input data through a pre-trained sensitivity analysis model.

Example 6:

the embodiment of the present invention further provides a sensitivity analysis model training device, and fig. 6 is a schematic structural diagram of the sensitivity analysis model training device provided in the embodiment of the present invention, where the device includes:

the acquiring unit 61 is configured to splice the acquired first basic data at the first time with the first behavior data of any fund in the sample set at the first time, and determine sample input data of the fund; the first quotation data is daily quotation data of the fund in a preset first time period before the first time; the first behavior data corresponds to a sensitivity label which is used for representing the real sensitivity of the fund in a preset second time period after the first time; the first base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

a training unit 62, configured to obtain, through an original sensitivity analysis model, a predicted sensitivity of the fund in a preset second time period after the first time based on the sample input data; and training the original sensitivity analysis model based on the predicted sensitivity and the corresponding real sensitivity to obtain a trained sensitivity analysis model.

Further, the obtaining unit 61 is further configured to determine a first high-order index according to the first basic data; and splicing the first behavior data, the first high-order index and the first basic data to determine the sample input data.

Further, the obtaining unit 61 is specifically configured to determine, according to the first basic data, that the stock market volatility index is determinable according to the following formula if the first high-order index includes a stock market volatility index and the first basic data includes a daily shannon index within the preset seventh time period:

wherein sv is_tRepresenting the stock market volatility index, s_tCollecting price for the number of the upper syndrome indexes in the daily number of upper syndrome indexes within the preset seventh time period, wherein T is more than or equal to 250 and is less than or equal to T, T is the preset seventh time period, and T is a positive integer more than or equal to 250; or, if the first high-order index includes a term difference index, and the first basic data includes the average daily return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period, determining the term difference index according to the average daily return rate of the national debt within the preset fifth time period and the daily return rate of the national debt within the preset sixth time period; or, if the first high-order index includes a banking risk tolerance index, and the first basic data includes the average daily return rate of the national bonds in the preset fifth time period and the average daily return rate of the national bonds in the preset fourth time period, determining the banking risk tolerance index according to the average daily return rate of the national bonds in the preset fifth time period and the average daily return rate of the national bonds in the preset fourth time period.

Further, the obtaining unit 61 is further configured to splice the first behavior data, the first high-order index, and the first basic data, and perform normalization processing on the first basic data and the first high-order index before determining the sample input data of the fund.

Example 7:

an embodiment of the present invention further provides a sensitivity analysis apparatus, and fig. 7 is a schematic structural diagram of the sensitivity analysis apparatus provided in the embodiment of the present invention, where the apparatus includes:

the obtaining module 71 is configured to obtain second basic data at a second time and second market data of the fund to be analyzed at the second time; the second market data is daily market data of the fund in a preset first time period before the second time; the second base data includes at least one of: presetting a daily allotment interest rate in a third time period, presetting a daily average return rate of national-level financial bonds in a fourth time period, presetting a daily average return rate of national bonds in a fifth time period, presetting a daily return rate of national bonds in a sixth time period and presetting a daily accrual index in a seventh time period;

the splicing module 72 is configured to splice the second basic data and the second market data, and determine input data of the fund to be analyzed;

and the processing module 73 is configured to obtain, through a pre-trained sensitivity analysis model, a predicted sensitivity of the fund to be analyzed in a preset second time period after the second time based on the input data.

Example 8:

on the basis of the foregoing embodiment, an embodiment of the present invention further provides an electronic device, and fig. 8 is a schematic structural diagram of the electronic device provided in the embodiment of the present invention, as shown in fig. 8, including: the system comprises a processor 81, a communication interface 82, a memory 83 and a communication bus 84, wherein the processor 81, the communication interface 82 and the memory 83 are communicated with each other through the communication bus 84;

the memory 83 has stored therein a computer program which, when executed by the processor 81, causes the processor 81 to perform the steps of:

Further, the processor is further configured to determine a first high-order index according to the first basic data; and splicing the first behavior data, the first high-order index and the first basic data to determine the sample input data.

Further, the processor is specifically configured to determine, according to the first basic data, that the stock market volatility index is determinable according to the following formula if the first high-order index includes a stock market volatility index and the first basic data includes a daily testimony index within the preset seventh time period:

Further, the processor is further configured to splice the first behavior data, the first high-order index, and the first basic data, and perform normalization processing on the first basic data and the first high-order index before determining the sample input data of the fund.

Because the principle of the electronic device for solving the problem is similar to the sensitivity analysis model training method, the implementation of the electronic device can be referred to in embodiments 1-3 of the method, and repeated details are not repeated.

The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus. The communication interface 82 is used for communication between the above-described electronic apparatus and other apparatuses. The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Alternatively, the memory may be at least one memory device located remotely from the processor.

The Processor may be a general-purpose Processor, including a central processing unit, a Network Processor (NP), and the like; but may also be a Digital instruction processor (DSP), an application specific integrated circuit, a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like.

Example 9:

on the basis of the foregoing embodiment, an embodiment of the present invention further provides an electronic device, and fig. 9 is a schematic structural diagram of another electronic device provided in the embodiment of the present invention, as shown in fig. 9, including: the system comprises a processor 91, a communication interface 92, a memory 93 and a communication bus 94, wherein the processor 91, the communication interface 92 and the memory 93 are communicated with each other through the communication bus 94;

the memory 93 has stored therein a computer program which, when executed by the processor 91, causes the processor 91 to perform the steps of:

Since the principle of the electronic device for solving the problem is similar to the sensitivity analysis method, the implementation of the electronic device can be referred to in embodiments 4-5 of the method, and repeated details are not repeated.

The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus. The communication interface 92 is used for communication between the above-described electronic apparatus and other apparatuses. The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Alternatively, the memory may be at least one memory device located remotely from the processor.

Example 10:

on the basis of the foregoing embodiments, the present invention further provides a computer-readable storage medium, in which a computer program executable by a processor is stored, and when the program is run on the processor, the processor is caused to execute the following steps:

determining a first high-order index according to the first basic data;

and normalizing the first basic data and the first high-order index.

Because the principle of solving the problem by the computer-readable storage medium is similar to the sensitivity analysis model training method, the implementation of the computer-readable storage medium can be referred to in embodiments 1-3 of the method, and repeated details are not repeated.

Example 11:

Since the principle of solving the problem of the computer-readable storage medium is similar to that of the sensitivity analysis method, the implementation of the computer-readable storage medium can be referred to in embodiments 4-5 of the method, and repeated details are not repeated.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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 means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.

Claims

1. A sensitivity analysis model training method, the method comprising:

2. The method of claim 1, wherein prior to stitching the first mood data and the first base data to determine the sample input data for the fund, the method further comprises:

determining a first high-order index according to the first basic data;

3. The method of claim 2, wherein said determining a first higher order exponent from said first base data comprises:

4. The method of claim 2, wherein prior to concatenating the first behavioral data, the first higher-order index, and the first base data to determine the sample input data for the fund, the method further comprises:

and normalizing the first basic data and the first high-order index.

5. A sensitivity analysis method, the method comprising:

6. A sensitivity analysis model training apparatus, characterized in that the apparatus comprises:

7. The apparatus of claim 6, wherein the obtaining unit is further configured to determine a first higher-order exponent based on the first base data; and splicing the first behavior data, the first high-order index and the first basic data to determine the sample input data.

8. The apparatus of claim 7, wherein the obtaining unit is specifically configured to determine, according to the first basic data, that the stock market volatility index is determinable according to a formula as follows, if the first high-order index includes a stock market volatility index and the first basic data includes a daily shannon index within the preset seventh time period:

9. The apparatus of claim 7, wherein the obtaining unit is further configured to concatenate the first behavior data, the first higher-order index, and the first basic data, and normalize the first basic data and the first higher-order index before determining the sample input data of the fund.

10. A sensitivity analysis apparatus, characterized in that the apparatus comprises:

11. An electronic device, characterized in that the electronic device comprises a processor for implementing the steps of the sensitivity analysis model training method according to any one of claims 1-4 or the steps of the sensitivity analysis method according to claim 5 when executing a computer program stored in a memory.

12. A computer-readable storage medium, characterized in that it stores a computer program which, when being executed by a processor, carries out the steps of the sensitivity analysis model training method according to any one of claims 1 to 4, or the steps of the sensitivity analysis method according to claim 5.