CN116523539A

CN116523539A - Option pricing method, system, equipment and storage medium based on neural network

Info

Publication number: CN116523539A
Application number: CN202310283130.XA
Authority: CN
Inventors: 张莲民; 张功球; 许天阳; 王树; 邓建辉; 彭洋洋; 罗敏
Original assignee: Shenzhen Research Institute of Big Data SRIBD
Current assignee: Shenzhen Research Institute of Big Data SRIBD
Priority date: 2023-03-15
Filing date: 2023-03-15
Publication date: 2023-08-01

Abstract

The embodiment of the application provides an option pricing method, system, equipment and storage medium based on a neural network, and belongs to the field of option pricing. The method comprises the following steps: acquiring target data related to options and used for pricing the options; preprocessing the target data to obtain target input data for neural network model input; inputting the target input data into a pre-trained option pricing model to obtain a target option price predicted value; the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model. The method and the device can improve option pricing efficiency while processing multidimensional option pricing problems.

Description

Option pricing method, system, equipment and storage medium based on neural network

Technical Field

The present application relates to the field of option pricing, and in particular, to a method, a system, a device, and a storage medium for option pricing based on a neural network.

Background

Option pricing is an important activity in a financial market, wherein the option pricing is to analyze market variation situations of simulated option prices through an appropriate mathematical model according to factors influencing the option prices, and finally obtain reasonable theoretical prices.

In the related art, a random differential equation model, a tree model, a PDE method and a Monte Carlo method are often utilized to obtain an analytic solution of the option price, however, the random differential equation model often cannot analyze more complex financial derivatives, the tree model and the PDE method have dimension disasters and can only process the low-dimension problem, while the Monte Carlo method is used as a method with the widest application range in the option pricing, and the real expectation is estimated by using an average value through simulating a plurality of paths, however, the calculation efficiency is low because tens of thousands of paths need to be simulated.

Disclosure of Invention

The embodiment of the application mainly aims to provide a method, a system, equipment and a storage medium for option pricing based on a neural network, which can improve the option pricing efficiency while processing the multidimensional option pricing problem.

To achieve the above object, a first aspect of an embodiment of the present application proposes an option pricing method based on a neural network, the method including: acquiring target data related to options and used for pricing the options; preprocessing the target data to obtain target input data for neural network model input; inputting the target input data into a pre-trained option pricing model to obtain a target option price predicted value; the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model.

In some embodiments, the option pricing model is trained by the steps of: acquiring a sample data set, and acquiring a training sample set from the sample data set; preprocessing the training sample set to obtain training sample input data for neural network model input; calculating expected values of the training sample input data to obtain expected values of option prices of the training samples; inputting the training sample input data into the option pricing model, and predicting the option price of the training sample input data through the option pricing model to obtain a training sample option price prediction value; and calculating the option loss value of the option pricing model according to the predicted value of the option price of the training sample and the expected value of the option price of the training sample, and adjusting the parameters of the option pricing model according to the option loss value to obtain the trained option pricing model.

In some embodiments, the method further comprises; acquiring a sample data set, and acquiring a test sample set from the sample data set; preprocessing the test sample set to obtain test sample input data for neural network model input; calculating expected values of the input data of the test sample to obtain expected values of the test sample; inputting the test sample input data into the option pricing model, and predicting the option price of the test sample input data through the option pricing model to obtain a test sample option price prediction value; and calculating the option mean square error according to the test sample option price predicted value and the test sample expected value, and obtaining an evaluation result of the option pricing model performance according to the option mean square error.

In some embodiments, the training sample set includes a training sample time set and a target asset price limit set; generating a first normal variable from preset standard normal distribution, performing solution differentiation processing according to the first normal variable, a first current time selected from the training sample time set and a first price selected from the target asset price limit set to obtain a first target asset price, and taking the first current time, the first price and the first target asset price as training sample input data under an European option; or generating a second normal variable from the preset standard normal distribution, performing solution differentiation processing according to the second normal variable, a second current time selected from the training sample time set and a second price selected from the target asset price limit set to obtain a second target asset price, generating a first uniform variable from the preset uniform distribution, taking the first uniform variable as a first candidate value, obtaining a second candidate value according to the second price, selecting the maximum value from the first candidate value and the second candidate value, obtaining a second target asset history highest price, and taking the second current time, the second price, the second target asset price and the second target asset history highest price as training sample input data under a presupposition option; or generating a third normal variable and a fourth normal variable from the preset standard normal distribution, performing solution differentiation processing according to the third normal variable, the fourth normal variable, the third current time selected from the training sample time set, the third price selected from the target asset price limit set and the fourth price selected from the target asset price limit set to obtain a third target asset price and a fourth target asset price, and taking the third current time, the third price, the fourth price, the third target asset price and the fourth target asset price as training sample input data under a difference price option.

In some embodiments, the calculating the expected value of the training sample input data to obtain a training sample option price expected value includes obtaining a first execution price, a first risk-free interest rate and a first expiration time, obtaining a third candidate value according to a difference value between the first target asset price and the first execution price, selecting a maximum value from the third candidate value and 0 as a return value of the European style option, and calculating the European style option expected value according to the first risk-free interest rate, the first expiration time, the first current time, the first price and the return value of the European style option to obtain the training sample option price expected value under the European style option; or, obtaining a second risk-free rate and a second expiration time, obtaining a return value of a return floating notice drop option according to the difference value of the second target asset price and the highest second target asset history price, and performing return notice option expected calculation according to the second risk-free rate, the second expiration time, the second current time, the second price and the return value of the return floating notice drop option to obtain a training sample option price expected value under the return notice option; or obtaining a third risk-free rate and a third expiration time, obtaining a fourth candidate value according to the difference value of the third target asset price and the fourth target asset price, selecting the maximum value from the fourth candidate value and 0 as a return value of the spread option, and performing spread option expected calculation according to the third risk-free rate, the third expiration time, the third current time, the third price, the fourth price and the return value of the spread option to obtain a training sample option price expected value under the spread option.

In some embodiments, the deriving the target option price forecast comprises: in the forward propagation process of the option pricing model, carrying out linear processing on the target input data through each layer of a network, and obtaining a target data linear processing value according to the weight, bias and number of each layer of nodes in the linear processing process; performing activation processing on the target data linear processing values in each layer to obtain target data activation values output by each layer; and obtaining a target option price predicted value corresponding to the target input data according to the training sample linear processing value and the training sample activation value corresponding to each layer.

In some embodiments, the parameters of the option pricing model include weights and biases for the various tier nodes; the adjusting parameters of the option pricing model according to the option loss value includes: in the back propagation process of the option pricing model, determining a weight gradient descent direction according to the option loss value and the weights before updating the nodes of each layer, and determining a bias gradient descent direction according to the option loss value and the bias before updating the nodes of each layer; according to the weight before updating, the preset learning rate and the weight gradient descending direction, calculating to obtain updated weights of nodes of each layer; and calculating to obtain the updated bias of each layer of nodes according to the pre-update bias, the preset learning rate and the gradient descent direction of the bias.

To achieve the above object, a second aspect of the embodiments of the present application proposes an option pricing system based on a neural network, the system comprising: the system comprises: the acquisition module is used for acquiring target data related to options and used for pricing the options; the preprocessing module is used for preprocessing the target data to obtain target input data for inputting a neural network model; the option pricing module is used for inputting the target input data into a pre-trained option pricing model to obtain a target option price predicted value; the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model.

To achieve the above object, a third aspect of the embodiments of the present application provides an electronic device, where the electronic device includes a memory and a processor, where the memory stores a computer program, and the processor implements the method for visualizing and configuring the chip test parameters according to the embodiment of the first aspect when executing the computer program.

To achieve the above object, a fourth aspect of the embodiments of the present application proposes a storage medium, which is a computer-readable storage medium, storing a computer program, where the computer program is executed by a processor to implement the method for visualizing configuration of chip test parameters according to the embodiment of the first aspect.

The embodiment of the application provides a method, a system, equipment and a storage medium for option pricing based on a neural network, wherein the method for option pricing based on the neural network can be applied to an option pricing system based on the neural network. In the option pricing method based on the neural network, firstly, target data which are related to options and used for option pricing of different option types are acquired; then, preprocessing the target data to obtain target input data for inputting a neural network model; finally, inputting the target input data into a pre-trained option pricing model to price options under various option types to obtain target option price predicted values of the corresponding option types; the option pricing model is a neural network model, efficiency of option pricing is improved, the option pricing model is obtained through training of sample option price expected values and sample option price predicted values under corresponding types of options, the sample option price expected values are obtained through expected value calculation according to sample data, and the sample option price predicted values are obtained after the sample data are input into the option pricing model.

According to the method and the system, the target data are obtained, the target input data obtained by preprocessing the target data are input into the trained option pricing model, the target option price predicted value is finally obtained, the target option price predicted value can guide the option related activities of the user, the option pricing model is a neural network model, and the efficiency of option pricing is improved while the multidimensional option pricing problem is well processed.

Drawings

FIG. 1 is a block diagram of a neural network-based option pricing system provided by embodiments of the present application;

FIG. 2 is an alternative flow chart of an option pricing method based on neural networks provided by embodiments of the present application;

FIG. 3 is another alternative flow chart of a neural network-based option pricing method provided by embodiments of the present application;

FIG. 4 is a flow chart of yet another alternative option pricing method based on neural networks provided by embodiments of the present application;

FIG. 5 is a further alternative flow chart of a neural network-based option pricing method provided by embodiments of the present application;

FIG. 6 is a flow chart of one implementation in step S202 of FIG. 3;

FIG. 7 is a flow chart of one implementation in step S203 of FIG. 3;

FIG. 8 is a schematic view of a curved surface of expected values of expected price of a training sample return option provided in an embodiment of the present application;

FIG. 9 is a schematic diagram of a curved surface of a predicted value of a return option price of a training sample according to an embodiment of the present application;

FIG. 10 is a schematic diagram of a training sample spread option price expectation curve provided in an embodiment of the present application;

FIG. 11 is a schematic diagram of a training sample spread option price predictor surface provided in an embodiment of the present application;

FIG. 12 is a flow chart of one implementation in step S103 of FIG. 2;

FIG. 13 is a schematic diagram of an option pricing model training process provided by embodiments of the present application;

FIG. 14 is a flow chart of one implementation in step S205 of FIG. 3;

FIG. 15 is a functional block diagram of an option pricing system based on neural networks provided in an embodiment of the present application;

fig. 16 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

It should be noted that although functional block division is performed in a device diagram and a logic sequence is shown in a flowchart, in some cases, the steps shown or described may be performed in a different order than the block division in the device, or in the flowchart. The terms first, second and the like in the description and in the claims and in the above-described figures, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the present application.

First, several nouns referred to in this application are parsed:

option contracts are a type of finance-derived product. The option contracts were generated in the chicago option exchange in 1973 and are trading contracts with financial derivative products as the option varieties. Refers to the right to trade a certain number of varieties at a certain price in a certain time.

Option pricing is a fixed price agreed upon in an option contract by which an option holder purchases or sells a subject asset. Factors influencing the option price are the market price and option execution price of the contractual asset, the expiration date of the option, the risk-free interest rate level, the volatility of the target asset price, etc.

Option execution price, also known as agreement price, option agreement price, exercise price, refers to the price that the option trading parties agree to execute the purchase and sale contracts in a specified future period of time. After the price determination is performed, the seller of the option must fulfill the obligation at the price whenever the buyer of the option asks to perform the option, regardless of price fluctuations, within the time period specified by the option contract.

The term of an option contract refers to the length of time remaining from the expiration date of the option contract. Under the condition that other factors are unchanged, the longer the term validity period is, the greater the time value of the term is. The longer the term buyer is, the more room is selected, the greater the likelihood that the future price will change in the direction desired by the buyer, and the more opportunities the term buyer will exercise the term, the greater the likelihood of earning. Conversely, the shorter the term validity, the lower the time value of the term.

The option due date, also known as the option line date, before this day comes, the investor must conduct line contact with this option contract.

Risk-free interest rate refers to the interest rate that can be obtained by investing funds in an investment object without any risk.

The volatility of a target asset is an important factor affecting option prices. In calculating the theoretical price of an option, the historical volatility of the target asset is typically employed: the larger the fluctuation rate, the higher the theoretical price of the option; conversely, the smaller the volatility, the lower the theoretical price of the option.

The european option refers to an option that a party who pays for the option must exercise on the date of the option due.

The option is returned, and the return is dependent on the maximum or minimum price reached by the asset of the option validity period. The return to the European presbyopia option is equal to the amount by which the last target asset price exceeds the minimum price reached by the asset of the option expiration date. The revenue for a return option is determined in part by the minimum or maximum price of the subject asset during the expiration date of the option.

Spread options refer to buying and selling the same kind of options at the same time with different execution prices.

The BlackScholes-Merton, BSM for short, is a mathematical model for pricing financial derivative tools such as rights or certificates, and was first developed by the United states economist Michelens Scholes and Fei Xue Blacks and completed by Robert Mo Du (Robert C.Merton). The model is named after milan schuls and Fei Xue blake.

Option pricing can help investors better control investment risk, and market changes can also be predicted by option pricing to better govern investment strategies, resulting in greater return on investment. Generally, a model is used for option pricing, at present, a random differential equation model, a tree model, a PDE method and a Monte Carlo method are often used for obtaining an analytic solution of option price, however, the random differential equation model often cannot analyze more complex financial derivatives, the tree model and the PDE method have dimension disasters, only can handle low-dimension problems, and the Monte Carlo method is used as a method with the widest application range in option pricing, and estimates real expectations by simulating a plurality of paths and using average values, however, the calculation efficiency is low because tens of thousands of paths need to be simulated.

Based on the above, the embodiments of the present application provide a method, a system, a device and a storage medium for option pricing based on a neural network, which can improve the option pricing efficiency while processing the multidimensional option problem.

The method, the system, the electronic device and the storage medium for option pricing based on the neural network provided by the embodiment of the application are specifically described through the following embodiments, and first, the system framework of the option pricing system based on the neural network in the application is introduced.

Illustratively, as shown in fig. 1, the option pricing system based on the neural network in the embodiment of the application includes a terminal and a client. The terminal can receive various input information of a user by setting the client and display the generated component through the client; or the client and the terminal can be independent logic individuals respectively, the client and the terminal can mutually receive and send data through communication connection, the client can send the configuration information of the user and other input information to the terminal after receiving the configuration information of the user and the other input information, the terminal processes based on the configuration information of the user and the other input information to generate or update the component, and the component is displayed in an interactive interface of the terminal, or the generated component is sent to the client to be displayed through the client.

Embodiments of the present application may also be attributed to the field of machine learning, where related data may be acquired and processed based on artificial intelligence techniques. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results.

Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.

The embodiment of the application provides an option pricing method based on a neural network, and relates to the technical field of artificial intelligence. The option pricing method based on the neural network can be applied to a terminal, a server side or software running in the terminal or the server side. In some embodiments, the terminal may be a tablet, notebook, desktop, or the like; the server side can be configured as an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, and a cloud server for providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, basic cloud computing services such as big data and artificial intelligent platforms and the like; the software may be an application or the like that implements the option pricing method based on neural networks, but is not limited to the above.

The subject application is operational with numerous general purpose or special purpose computer system environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The option pricing method based on the neural network in the embodiment of the application can be illustrated by the following embodiment.

It should be noted that, in each specific embodiment of the present application, when related processing needs to be performed according to data related to a user identity or a characteristic, such as user information, user behavior data, user history data, user location information, etc., permission or consent of the user is obtained first, for example, when data stored by the user and a request for accessing cached data of the user are obtained first; when acquiring the option related information of the user, the embodiment of the application can firstly acquire the permission or consent of the user. Moreover, the collection, use, and processing of such data, etc., complies with relevant national and regional laws and regulations. In addition, when the embodiment of the application needs to acquire the sensitive personal information of the user, the independent permission or independent consent of the user is acquired through a popup window or a jump to a confirmation page or the like, and after the independent permission or independent consent of the user is explicitly acquired, necessary user related data for enabling the embodiment of the application to normally operate is acquired.

As shown in fig. 2, fig. 2 is an optional flowchart of an option pricing method based on a neural network according to an embodiment of the present application, and the method in fig. 2 may include, but is not limited to, steps S101 to S103.

Step S101, acquiring target data which is related to options and is used for pricing the options;

step S102, preprocessing target data to obtain target input data for neural network model input;

step S103, inputting target input data into a pre-trained option pricing model to obtain a target option price predicted value; the option pricing model is a neural network model, the option pricing model is obtained through training of a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by calculating the expected value according to sample data, and the sample option price predicted value is obtained after the sample data is input into the expiration option pricing model.

In some embodiments, the target data can be obtained through a related interface and related open source data disclosed by a related platform, and the target data is an important basis for option pricing.

In some embodiments, the obtained target data cannot be directly input into the option pricing model, the target data needs to be preprocessed to obtain target input data which can be used for inputting the option pricing model, the target input data is input into the option pricing model, a target option price predicted value can be obtained, the option price predicted value plays a role of reference direction for a user, and the user is better helped to make related decisions of the option activity.

In some embodiments, the option pricing model is trained from sample option price expectations and sample option price predictions of sample data, and the sample data required for training includes sample data for different types of options. It can be appreciated that the option pricing model can address the option pricing problem for multiple options after training of sample data for different types of options.

The method comprises the steps of obtaining Europeanism sample data, preprocessing the sample data to obtain Europism sample input data capable of inputting an option pricing model, and inputting the preprocessed Europism sample input data to the option pricing model to perform model training. On one hand, the option price predicted value is obtained after the option sample input data passes through the option pricing model, on the other hand, the option price expected value is obtained by calculating the option sample input data, the option loss value is calculated through the option price predicted value and the option price expected value, and the option pricing model is adjusted according to the option loss value, so that the option loss value of the option pricing model after continuous adjustment is smaller and smaller, and the purpose is to enable the option predicted value output by the option pricing model after training to be closer and closer to the option expected value expected by the user, so that the option pricing model has practical significance, and the output option target predicted value can accord with the expected output value of the user when the user inputs related option target input data, and the option activity of the user is brought with a reality guiding function.

In some embodiments, a termination condition for stopping the option pricing model training may be preset, for example, the termination condition may be set to 1000 iterations, if the number of times the option pricing model training does not reach 1000 times, training sample data will be continuously input to train the option pricing model, and the option pricing model will be continuously adjusted according to the calculated option loss value, and if the number of times the option pricing model training reaches 1000 times, training the option pricing model is stopped, which indicates that the option pricing model training is completed.

It should be noted that, the termination condition for stopping the option pricing model training may be that the option loss value is lower than a certain threshold, or other termination conditions may be set according to the specific situation, which is not limited in this application.

It should be noted that, the option pricing model may also train the option pricing model through option sample data such as a return option and a difference option, and the option pricing model trained by different option sample data may process option input data corresponding to different types and output option prediction prices corresponding to different types.

In some embodiments, each neural network layer of the trained option pricing model can perform linear processing and activation operations on the input option target input data to finally obtain the option price prediction value. It can be understood that the option pricing model is trained by using sample data of different option types and a final option pricing model is obtained, and the option pricing model is a neural network model, so that the neural network has strong self-learning property, strong robustness, strong comprehensive processing capacity and the like, the option pricing model can process multidimensional option pricing problems of different option types, and the efficiency of option pricing is improved through strong calculation of the neural network.

As shown in fig. 3, fig. 3 is another alternative flowchart of an option pricing method based on a neural network according to an embodiment of the present application, and the method in fig. 3 may include, but is not limited to, steps S201 to S205.

Step S201, acquiring a sample data set, and acquiring a training sample set from the sample data set;

step S202, preprocessing a training sample set to obtain training sample input data for neural network model input;

Step S203, expected value calculation is carried out on training sample input data to obtain a training sample option price expected value;

step S204, inputting training sample input data into an expiration pricing model, and predicting the option price of the training sample input data through the option pricing model to obtain a training sample option price prediction value;

step S205, according to the predicted value of the option price of the training sample and the expected value of the option price of the training sample, the option loss value of the option pricing model is calculated, and the parameters of the option pricing model are adjusted according to the option loss value, so as to obtain the trained option pricing model.

In some embodiments, the sample data set may be obtained from the relevant interface and relevant open source data disclosed by the relevant platform, and the training sample set is divided from the sample data set, and generally, the training sample set is divided by 80% according to the size of the sample data set, and is used for training the neural network model, and when training of the neural network model is completed, the option pricing model may be obtained.

It should be noted that, the training sample set of the sample data set may be divided according to a required proportion, so long as the purpose of training the neural network model can be achieved, and the specific requirement is not made in the application.

In some embodiments, the obtained training sample set is information related to options, and the option pricing model cannot be directly input, and the training sample set needs to be preprocessed to obtain training sample input data conforming to the input option pricing model.

In some embodiments, on one hand, expected value calculation needs to be performed on training sample input data to obtain a training sample price expected value, on the other hand, the training sample input data needs to be input into an option pricing model to perform option price prediction to obtain a training sample option price predicted value, wherein the training sample price expected value represents an option price value of a training sample expected by a user, the training sample option price predicted value represents an option predicted result obtained after processing by the option pricing model, the training sample option price predicted value and the training sample option price expected value are subjected to loss value calculation to obtain an option loss value, and parameter adjustment is performed by using the option loss value to obtain a trained option pricing model.

As shown in fig. 4, fig. 4 is a flowchart of still another alternative option pricing method based on neural networks provided in an embodiment of the present application, and the method in fig. 4 may include, but is not limited to including, step S301 to step S305.

Step S301, a sample data set is obtained, and a test sample set is obtained from the sample data set;

step S302, preprocessing a test sample set to obtain test sample input data for neural network model input;

step S303, calculating expected values of input data of the test sample to obtain expected values of the test sample;

step S304, inputting the test sample input data into an expiration price model, and predicting the option price of the test sample input data through the option price model to obtain a test sample option price prediction value;

and step S305, calculating the option mean square error according to the test sample option price predicted value and the test sample expected value, and obtaining an evaluation result of the option pricing model performance according to the option mean square error.

In some embodiments, the sample data set may be obtained from the relevant interface and relevant open source data disclosed by the relevant platform, and the test sample set is divided from the sample data set, and generally, the test sample set is divided by 20% according to the size of the sample data set, and the test sample set is used for testing the option pricing model obtained by training, so as to evaluate the training result of the option pricing model.

It should be noted that, the test sample set of the sample data set may be divided according to a required proportion, so long as the purpose of testing the trained option pricing model can be achieved, and the specific requirement is not made in the present application.

In some embodiments, the obtained test sample set is information related to options, and the option pricing model cannot be directly input, and the test sample set needs to be preprocessed to obtain test sample input data conforming to the input option pricing model.

In some embodiments, on one hand, expected value calculation needs to be performed on test sample input data to obtain a test sample price expected value, on the other hand, the test sample input data needs to be input into an option pricing model to perform option price prediction to obtain a test sample option price predicted value, where the test sample price expected value represents an option price value of a test sample expected to be obtained by a user, the test sample option price predicted value represents an option predicted result obtained after processing by the option pricing model, mean square error calculation is performed on the test sample option price predicted value and the test sample option price expected value to obtain a mean square error value, and the trained option pricing model is evaluated by using the mean square error.

As shown in fig. 5, fig. 5 is a further alternative flow chart of the option pricing method based on neural networks provided in the embodiments of the present application. In some embodiments, the option pricing process may proceed as follows:

step1, according to different types of options, sampling related data (such as time, target asset price and the like) as input of a neural network model;

step2, initializing training parameters (such as weights among nerve nodes, node bias, algorithm termination iteration times and the like) of the nerve network model;

step3, according to the training sample set size: test dataset size=8:2, dividing the sample dataset (Input, output) into training sample sets (Input) _train ,Output _train ) And a test dataset (Input _test ,Output _test ) Will (Input) _train ,Output _train ) Inputting a neural network model for training, wherein a training sample set contains M pieces of sample data;

step4, according to forward propagation, by Input _train Calculating the input and output of each layer of the neural network model to obtain the option price predictive value of the training sample

Step5, judging whether the number of times of ending the iteration is 1000 times or not, if so, turning to Step7; otherwise, go to Step6;

step6, according to the back propagation, the neural network model parameters are adjusted, and Step4 is carried out;

Step7, obtaining a final trained option pricing model, inputting related information of options into the network, and obtaining option pricing at corresponding moments;

step8, will (Input _test ,Output _test ) Inputting the trained option pricing model to obtainThe mean square error (Mean Square Error, MSE) of the test sample option price forecast and the training sample option price forecast is calculated as follows:

wherein, the mean square error is used for evaluating the algorithm performance, the smaller the MSE, the better the network effect.

As shown in fig. 6, fig. 6 is a flowchart of one implementation of step S202 of fig. 3, and in some embodiments, step S202 may include steps S401 to S403:

step S401, generating a first normal variable from preset standard normal distribution, performing solution differentiation processing according to the first normal variable, a first current time selected from a training sample time set and a first price selected from a target asset price limit set to obtain a first target asset price, and taking the first current time, the first price and the first target asset price as training sample input data under the European option;

step S402, or, generating a second normal variable from a preset standard normal distribution, performing solution differentiation processing according to the second normal variable, a second current time selected from a training sample time set and a second price selected from a target asset price limit set to obtain a second target asset price, generating a first uniform variable from a preset uniform distribution, taking the first uniform variable as a first candidate value, obtaining a second candidate value according to the second price, selecting the maximum value from the first candidate value and the second candidate value, obtaining a second target asset history highest price, and taking the second current time, the second price, the second target asset price and the second target asset history highest price as training sample input data under a return option;

In step S403, or, generating a third normal variable and a fourth normal variable from the preset standard normal distribution, performing a solution differentiation process according to the third normal variable, the fourth normal variable, the third current time selected from the training sample time set, the third price selected from the target asset price limit set, and the fourth price selected from the target asset price limit set, to obtain a third target asset price and a fourth target asset price, and taking the third current time, the third price, the fourth price, the third target asset price, and the fourth target asset price as training sample input data under the differential option.

In some embodiments, the training sample set includes a training sample time set and a target asset price limit set, wherein the training sample time set is uniformly distributed U [0, T]T represents the option expiration time, and the set of target asset price limits is evenly distributed SRepresenting the lower bound of the price of the target asset, +.>Representing the upper bound of the target asset price.

In some embodiments, from U [0, T]Generates a first current time t _i From the slaveGenerates a first price x _i I=1, …, M, where M represents the number of training samples, generating a first normal variable Z from a standard normal distribution _i -N (0, 1), i=1, …, M, at which time a de-differentiation process is performed to obtain a first target asset price ∈ ->The method comprises the following steps:

wherein r is ₁ Representing a first risk-free interest rate, sigma ₁ Representing a first rate of fluctuation.

The calculated first current time, first price and firstThe price of a target asset is used as training sample input data under European option, namely

In some embodiments, from U [0, T]Generates a second current time t _i From the slaveGenerates a second price x _i I=1, …, M, where M represents the number of training samples, generating a second normal variable Z from the standard normal distribution _i -N (0, 1), i=1, …, M, at which time a de-differentiation process is performed to obtain a second target asset price +.>The method comprises the following steps:

wherein r is ₂ Representing a second risk-free interest rate, sigma ₂ Representing a second rate of fluctuation.

Generating a first uniform variable u from a uniform distribution _i U (0, 1), where the second target asset history maximum price isWherein (1)>

The calculated second current time, second price, second target asset price and second target asset historical highest price are used as training sample input data under the hopeback option, namely

In some embodiments, from U [0, T]Generates a third current time t _i From the slaveGenerates a third price x _1,i And x _2,i Generating a third normal variable Z from the normal distribution _1,i And a fourth normal variable Z _2,i At this time, a solution differentiation process is performed to obtain a third target asset price +.>The method comprises the following steps:

fourth target asset priceThe method comprises the following steps:

wherein r is ₃ Representing a third risk-free interest rate, sigma ₃ Represents the third fluctuation rate, r ₄ Representing a third risk-free interest rate, sigma ₄ Representing a third rate of fluctuation.

Inputting the calculated third current time, third price, fourth price, third target asset price and fourth target asset price as training samples under the spread option, namely

As shown in fig. 7, fig. 7 is a flowchart of one implementation of step S203 of fig. 3, and in some embodiments, step S203 may include steps S501 to S503:

step S501, obtaining a first execution price, a first risk-free rate and a first expiration time, obtaining a third candidate value according to the difference between a first target asset price and the first execution price, selecting the maximum value from the third candidate value and 0 as a return value of the European style expansion option, and performing European style option expected calculation according to the first risk-free rate, the first expiration time, the first current time, the first price and the return value of the European style expansion option to obtain a training sample option price expected value under the European style option;

Step S502, or, obtaining a second risk-free interest rate and a second expiration time, obtaining a return value of a return floating notice option according to a difference value between a second target asset price and a second target asset history highest price, and performing return notice option expected calculation according to the second risk-free interest rate, the second expiration time, a second current time, the second price and the return value of the return floating notice option to obtain a training sample option price expected value under the return notice option;

step S503, or, obtaining a third risk-free rate and a third expiration time, obtaining a fourth candidate value according to the difference between the third target asset price and the fourth target asset price, selecting the maximum value from the fourth candidate value and 0 as the return value of the spread option, and performing the expected calculation of the spread option according to the third current risk-free rate, the third expiration time, the third current time, the third price, the fourth price and the return value of the spread option to obtain the expected value of the training sample option under the spread option.

In some embodiments, for an European option, the option holder can only conduct the option at the first expiration time T. According to the Black-Scholes-Merton (BSM) model, the first target asset price { X _t T=1, …, T } meets the differential equation:

dX _t ＝rX _t dt+σX _t dW _t

wherein, under the Euro option in the embodiment of the application, r represents a first risk-free interest rate, sigma represents a first fluctuation rate, W _t Is brownian motion under a risk neutral measure. The price X of the first target asset of the person when the return of European style expansion option is the first expiration time _T The maximum value between the difference value from the first execution price K and 0, namely:

F(X _T )＝max(X _T -K,0)

thus, the Black-Scholes formula of European expansion options is obtained:

wherein N (·) is the cumulative probability distribution function of the standard normal distribution, anThe first equation represents the price X of the first subject asset at time T _T Under the condition of =x, the price of the European option is the return F (X _T ) According to a first risk-free interest rate r ₁ Discount the price to the first current time; the second equation represents an analytical solution for obtaining the option pricing according to the BSM model, i.e. the expected value of the option price of the training sample under the option.

In some embodiments, the return on the return option is determined by a target asset history maximum or minimum, for example, the return on the floating drop option is the second target asset history maximum priceAnd a second target asset price X _T The difference is:

Wherein the second target asset price { X } _t T=1, …, T } conforms to the BSM model;representing the maximum value of the second target asset price over the 0-t period. The analysis solution of the option value is as follows:

wherein, r represents a second risk-free interest rate under the hope-back option in the embodiment of the applicationSigma represents a second volatility, and the first equation represents the price X of the second target asset at the second current time T _T =x, second target asset historic highest priceUnder the condition of (1), the price of the return option is the return option of the return floating option +.>Discount the price to the second current time t according to the second risk-free interest rate r; the second equation represents an analytic solution for obtaining the option price of the hopeback according to the BSM model, namely a training sample option price expected value under the hopeback option. Furthermore:

where τ=t-T.

As shown in fig. 8 and fig. 9, fig. 8 is a schematic diagram of a training sample return option price expected value curved surface provided by an embodiment of the present application, and fig. 9 is a schematic diagram of a training sample return option price predicted value curved surface provided by an embodiment of the present application. From this neural network model, expected and predicted values of the return option can be derived, and thus a return option price function surface, which is illustrated, for example, in figures 8 and 9, S＝70，K=100, σ=0.03, r=0.02, m=5000, toFor x and y axes, the option price is taken as a vertical axis, and an option expected value and a predicted value are drawn as shown in fig. 8 and 9, it is understood that a green point in fig. 8 represents a desired value of the option price, a red point in fig. 9 represents a predicted value of the option price, and the shape of a green point curved surface in fig. 8 is matched with the shape of a red point curved surface in fig. 9, and the method can obtain a function curved surface of the option price.

In some embodiments, two stock X following BSM models under risk neutral measures are considered _j,t J=1, 2, the process of which can be expressed as differential equation dX _j,t ＝rX _j,t dt+σ _i X _j,t dW _j,t ,i＝1,2,W _1,t And W is _2,t Is a BSM procedure with two correlation coefficients p. Based on these two assets, the return for the spread option is the maximum between the difference of the two stocks and 0:

F(X _1,T ,X _2,T )＝max(X _1,T -X _2,T ,0)

the solution of the spread option price is known as the Margrane formula:

wherein R represents a third risk-free interest rate, σ represents a third volatility, and the first equation represents the price X of the subject asset at time T _1,T ＝x ₁ Price X of target asset at time T _2,T ＝x ₂ Under the condition that the price of the spread option is the return of the spread optionThe price up to the third current time t is discounted according to the third risk-free interest rate r; the second equation represents an analytical solution for obtaining the price of the spread option according to the BSM model, i.e. a training sample option price expected value under the spread option. Furthermore:

As shown in fig. 10 and 11, fig. 10 is a schematic diagram of a training sample option price expected value curved surface provided by an embodiment of the present application, and fig. 11 is a schematic diagram of a training sample option price predicted value curved surface provided by an embodiment of the present application. From this neural network model, it is possible to derive expected and predicted values of the spread options and thereby derive a spread option price function surface, which is illustrated, for example, in fig. 10 and 11,S＝70，k=100, σ=0.03, r=0.02, ρ=0.03, m=5000, at +.>For x and y axes, the option price is taken as a vertical axis, and a spread option expected value and a predicted value are drawn as shown in fig. 10 and 11, it can be understood that a green point in fig. 10 represents a return option price expected value, a red point in fig. 11 represents a return option price predicted value, and the shape of a green point curved surface in fig. 10 is matched with the shape of a red point curved surface in fig. 11, and the spread option price function curved surface can be obtained by the method.

As shown in fig. 12, fig. 12 is a flowchart of one implementation of step S103 of fig. 1, and in some embodiments, step S103 may include steps S601 to S603:

step S601, in the forward propagation process of the option pricing model, performing linear processing on target input data through each layer of the network, and obtaining a target data linear processing value according to the weight, bias and number of each layer of nodes in the linear processing process;

Step S602, performing activation processing on the target data linear processing values in each layer to obtain target data activation values output by each layer;

step S603, obtaining a target option price predicted value corresponding to target input data according to the target data linear processing value and the target data activation value corresponding to each layer.

As shown in fig. 13, fig. 13 is a schematic diagram of an option pricing model training process provided by embodiments of the present application, in some embodiments, input represents training sample Input data, output represents training sample option price expectations,the training sample option price forecast is represented and the option pricing model includes a plurality of neural network layers, each including a plurality of neural nodes thereon. Input can be subjected to linear processing through forward propagation of option pricing models, and a linear processing result is obtained, wherein the specific formula is as follows: />

Wherein, the liquid crystal display device comprises a liquid crystal display device,representing each layer of linear function to scale y from d _i-1 Dimension mapping to d _i I represents the ith layer, w, of the option pricing model _i Representing the weight between the (i-1) layer node and the i layer node, b _i Indicating the bias of the i-th layer.

Illustratively, the option pricing model in FIG. 4 has a total of L neural network layers, the weights and biases between the (i-1) th and i-th neural network layers are represented as shown in FIG. 4, and the 1 st neural node of the (i-1) th neural network layer and the 1 st of the i-th neural network layer ₁ 、A ₂ 、B ₁ 、B ₂ 、C ₁ 、C ₂ Specifically, A ₁ ＝w _i ^(1,1) 、Wherein d _i Indicating the number of nodes of the i-th layer, it being understood that +.>Representing the weight between the z-th node of the i-1 th layer and the j-th node of the i-th layer,/->Representing the bias of the j-th node of the i-th layer.

The specific implementation of the weights and biases of the 2 nd neural node of the (i-1) th neural network layer and the 1 st neural node of the i-1 th neural network layer are substantially the same as the specific embodiment of the 1 st neural node of the (i-1) th neural network layer described above, and likewise, the specific implementation of the weights and biases of the 1 st neural node of the (i-2) th neural network layer and the 1 st of the (i-1) th neural network layer are substantially the same as the specific embodiment of the 1 st neural node of the (i-1) th neural network layer described above.

In some embodiments, the option pricing model can process the resulting linear results/ _i And performing activation processing to obtain an activation value, wherein the specific formula is as follows:

the option pricing model is based on the linear processing result l of each neural network layer _i And an activation value sigma _i Calculating to obtain the option price predictive value of the training sample The specific calculation formula is as follows:

wherein, the liquid crystal display device comprises a liquid crystal display device,multiplication representing a mapping actually refers to a composite mapping of two mappings.

As shown in fig. 14, fig. 14 is a flowchart of one implementation of step S205 of fig. 3, and in some embodiments, step S205 may include steps S701 to S703:

step S701, in the back propagation process of the option pricing model, determining a weight gradient descent direction according to the option loss value and the weights before updating the nodes of each layer, and determining a bias gradient descent direction according to the option loss value and the bias before updating the nodes of each layer;

step S702, calculating to obtain updated weights of nodes of each layer according to the weights before updating, the preset learning rate and the weight gradient descending direction;

step S703, calculating to obtain the updated bias of each layer node according to the pre-update bias, the preset learning rate and the bias gradient descent direction.

In some embodiments, option price predictors can be based on training samplesAnd training a sample option price expected value Output, wherein the option pricing model can calculate an option loss value of the option pricing model according to back propagation, and adjust parameters of the option pricing model according to the option loss value to obtain a trained option pricing model, and a specific loss value calculation formula is as follows: / >

In some embodiments, to minimize the loss function, the inter-neural node weights w of the neural network are adjusted according to a gradient descent method _i (i=1, 2, …, L) and b _i The (i=1, 2, …, L) node is biased, and the weight adjustment formula is that W is the updated weight _i For pre-update weights, η is learning rate, +.>In the direction of gradient descent, i.e. Loss function Loss with respect to w _i Is a negative gradient direction of (c). Similarly, the bias adjustment formula is +.>

It should be noted that, the number of neural network layers and the number of neural nodes of each neural network layer of the option pricing model may be set according to actual needs, and the embodiment of the present application is not limited specifically.

It can be understood that the option pricing model needs to input a large amount of training sample input data to perform model training, and continuously performs model adjustment through the output option loss value, the option pricing model after each adjustment is closer to the ideal option pricing model than before the last adjustment, the option pricing model after training can perform target option price prediction on the input target data, the obtained target option price prediction value is close to the ideal option price value, and practical reference significance can be brought to a user.

As shown in fig. 15, fig. 15 is a schematic diagram of a functional module of a option pricing system based on a neural network provided in an embodiment of the present application, and the embodiment of the present application further provides a option pricing system based on a neural network, which may implement the option pricing method based on a neural network, where the option pricing system based on a neural network includes:

an acquisition module 801, configured to acquire target data related to options and used for option pricing;

a preprocessing module 802, configured to preprocess target data to obtain target input data for inputting a neural network model;

option pricing module 803 is configured to input target data into a pre-trained option pricing model to obtain a target option price prediction value; the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model.

In some embodiments, the obtained target data is preprocessed, the preprocessed target data is input into an option pricing model, the option pricing model trained by the training sample set can conduct price prediction processing on the target data, and finally an option price predicted value is obtained, the option price predicted value plays a role of reference direction for a user, and the user is better helped to make related decisions of option activities.

The specific implementation of the option pricing system based on the neural network is basically the same as the specific embodiment of the option pricing method based on the neural network, and will not be described herein. On the premise of meeting the requirements of the embodiment of the application, the option pricing system based on the neural network can be provided with other functional modules so as to realize the option pricing method based on the neural network in the embodiment.

The embodiment of the application also provides electronic equipment, which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the option pricing method based on the neural network when executing the computer program. The electronic equipment can be any intelligent terminal including a tablet personal computer, a vehicle-mounted computer and the like.

As shown in fig. 16, fig. 16 illustrates a hardware structure of an electronic device of another embodiment, the electronic device including:

the processor 901 may be implemented by a general purpose CPU (central processing unit), a microprocessor, an application specific integrated circuit (ApplicationSpecificIntegratedCircuit, ASIC), or one or more integrated circuits, etc. for executing related programs to implement the technical solutions provided by the embodiments of the present application;

The memory 902 may be implemented in the form of read-only memory (ReadOnlyMemory, ROM), static storage, dynamic storage, or random access memory (RandomAccessMemory, RAM). Memory 902 may store an operating system and other application programs, and when implementing the technical solutions provided by the embodiments of the present disclosure through software or firmware, relevant program codes are stored in memory 902 and invoked by processor 901 to perform the option pricing method based on the neural network of the embodiments of the present disclosure;

an input/output interface 903 for inputting and outputting information;

the communication interface 904 is configured to implement communication interaction between the device and other devices, and may implement communication in a wired manner (e.g. USB, network cable, etc.), or may implement communication in a wireless manner (e.g. mobile network, WIFI, bluetooth, etc.);

a bus 905 that transfers information between the various components of the device (e.g., the processor 901, the memory 902, the input/output interface 903, and the communication interface 904);

wherein the processor 901, the memory 902, the input/output interface 903 and the communication interface 904 are communicatively coupled to each other within the device via a bus 905.

Embodiments of the present application also provide a computer readable storage medium storing a computer program that when executed by a processor implements the neural network-based option pricing method described above.

The memory, as a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. In addition, the memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory remotely located relative to the processor, the remote memory being connectable to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The embodiments described in the embodiments of the present application are for more clearly describing the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application, and as those skilled in the art can know that, with the evolution of technology and the appearance of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

It will be appreciated by those skilled in the art that the technical solutions shown in the figures do not constitute limitations of the embodiments of the present application, and may include more or fewer steps than shown, or may combine certain steps, or different steps.

The above described apparatus embodiments are merely illustrative, wherein the units illustrated as separate components may or may not be physically separate, i.e. may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Those of ordinary skill in the art will appreciate that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof.

The terms "first," "second," "third," "fourth," and the like in the description of the present application and in the above-described figures, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that embodiments of the present application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

It should be understood that in this application, "at least one (item)" and "a number" mean one or more, and "a plurality" means two or more. "and/or" for describing the association relationship of the association object, the representation may have three relationships, for example, "a and/or B" may represent: only a, only B and both a and B are present, wherein a, B may be singular or plural. The character "/" generally indicates that the context-dependent object is an "or" relationship. "at least one of" or the like means any combination of these items, including any combination of single item(s) or plural items(s). For example, at least one (one) of a, b or c may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.

In the several embodiments provided in this application, it should be understood that the disclosed systems and methods may be implemented in other ways. For example, the system embodiments described above are merely illustrative, e.g., the division of the above elements is merely a logical functional division, and there may be additional divisions in actual implementation, e.g., multiple elements or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including multiple instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the various embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing a program.

Preferred embodiments of the present application are described above with reference to the accompanying drawings, and thus do not limit the scope of the claims of the embodiments of the present application. Any modifications, equivalent substitutions and improvements made by those skilled in the art without departing from the scope and spirit of the embodiments of the present application shall fall within the scope of the claims of the embodiments of the present application.

Claims

1. A method for option pricing based on a neural network, the method comprising:

acquiring target data related to options and used for pricing the options;

preprocessing the target data to obtain target input data for neural network model input;

inputting the target input data into a pre-trained option pricing model to obtain a target option price predicted value;

the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model.

2. The option pricing method based on neural network of claim 1, wherein the option pricing model is trained by the steps of:

acquiring a sample data set, and acquiring a training sample set from the sample data set;

preprocessing the training sample set to obtain training sample input data for neural network model input;

calculating expected values of the training sample input data to obtain expected values of option prices of the training samples;

inputting the training sample input data into the option pricing model, and predicting the option price of the training sample input data through the option pricing model to obtain a training sample option price prediction value;

and calculating the option loss value of the option pricing model according to the predicted value of the option price of the training sample and the expected value of the option price of the training sample, and adjusting the parameters of the option pricing model according to the option loss value to obtain the trained option pricing model.

3. The neural network-based option pricing method of claim 2, wherein the method further comprises;

Acquiring a sample data set, and acquiring a test sample set from the sample data set;

preprocessing the test sample set to obtain test sample input data for neural network model input;

calculating expected values of the input data of the test sample to obtain expected values of the test sample;

inputting the test sample input data into the option pricing model, and predicting the option price of the test sample input data through the option pricing model to obtain a test sample option price prediction value;

and calculating the option mean square error according to the test sample option price predicted value and the test sample expected value, and obtaining an evaluation result of the option pricing model performance according to the option mean square error.

4. The neural network-based option pricing method of claim 2, wherein the training sample set comprises a training sample time set and a target asset price limit set;

the preprocessing the training sample set to obtain training sample input data for neural network model input comprises the following steps:

generating a first normal variable from preset standard normal distribution, performing solution differentiation processing according to the first normal variable, a first current time selected from the training sample time set and a first price selected from the target asset price limit set to obtain a first target asset price, and taking the first current time, the first price and the first target asset price as training sample input data under an European option;

Or generating a second normal variable from the preset standard normal distribution, performing solution differentiation processing according to the second normal variable, a second current time selected from the training sample time set and a second price selected from the target asset price limit set to obtain a second target asset price, generating a first uniform variable from the preset uniform distribution, taking the first uniform variable as a first candidate value, obtaining a second candidate value according to the second price, selecting the maximum value from the first candidate value and the second candidate value, obtaining a second target asset history highest price, and taking the second current time, the second price, the second target asset price and the second target asset history highest price as training sample input data under a presupposition option;

or generating a third normal variable and a fourth normal variable from the preset standard normal distribution, performing solution differentiation processing according to the third normal variable, the fourth normal variable, the third current time selected from the training sample time set, the third price selected from the target asset price limit set and the fourth price selected from the target asset price limit set to obtain a third target asset price and a fourth target asset price, and taking the third current time, the third price, the fourth price, the third target asset price and the fourth target asset price as training sample input data under a difference price option.

5. The option pricing method based on neural network according to claim 2 or 4, wherein the calculating the expected value of the training sample input data to obtain the expected value of the training sample option price comprises:

acquiring a first execution price, a first risk-free interest rate and a first expiration time, obtaining a third candidate value according to the difference value of the first target asset price and the first execution price, selecting the maximum value from the third candidate value and 0 as a return value of the European style expansion option, and performing European style option expected calculation according to the first risk-free interest rate, the first expiration time, the first current time, the first price and the return value of the European style expansion option to obtain a training sample option price expected value under the European style option;

or, obtaining a second risk-free rate and a second expiration time, obtaining a return value of a return floating notice drop option according to the difference value of the second target asset price and the highest second target asset history price, and performing return notice option expected calculation according to the second risk-free rate, the second expiration time, the second current time, the second price and the return value of the return floating notice drop option to obtain a training sample option price expected value under the return notice option;

Or obtaining a third risk-free rate and a third expiration time, obtaining a fourth candidate value according to the difference value of the third target asset price and the fourth target asset price, selecting the maximum value from the fourth candidate value and 0 as a return value of the spread option, and performing spread option expected calculation according to the third risk-free rate, the third expiration time, the third current time, the third price, the fourth price and the return value of the spread option to obtain a training sample option price expected value under the spread option.

6. The neural network-based option pricing method of claim 1, wherein the deriving the target option price forecast comprises:

in the forward propagation process of the option pricing model, carrying out linear processing on the target input data through each layer of a network, and obtaining a target data linear processing value according to the weight, bias and number of each layer of nodes in the linear processing process;

performing activation processing on the target data linear processing values in each layer to obtain target data activation values output by each layer;

and obtaining a target option price predicted value corresponding to the target input data according to the target data linear processing value and the target data activation value corresponding to each layer.

7. The neural network-based option pricing method of claim 2, wherein the parameters of the option pricing model include weights and biases for each tier node;

the adjusting parameters of the option pricing model according to the option loss value includes:

in the back propagation process of the option pricing model, determining a weight gradient descent direction according to the option loss value and the weights before updating the nodes of each layer, and determining a bias gradient descent direction according to the option loss value and the bias before updating the nodes of each layer;

according to the weight before updating, the preset learning rate and the weight gradient descending direction, calculating to obtain updated weights of nodes of each layer;

and calculating to obtain the updated bias of each layer of nodes according to the pre-update bias, the preset learning rate and the gradient descent direction of the bias.

8. An option pricing system based on a neural network, the system comprising:

the acquisition module is used for acquiring target data related to options and used for pricing the options;

the preprocessing module is used for preprocessing the target data to obtain target input data for inputting a neural network model;

The option pricing module is used for inputting the target input data into a pre-trained option pricing model to obtain a target option price predicted value; the option pricing model is a neural network model, the option pricing model is trained through a sample option price expected value and a sample option price predicted value under corresponding types of options, the sample option price expected value is obtained by carrying out expected value calculation according to sample data, and the sample option price predicted value is obtained after the sample data is input into the option pricing model.

9. An electronic device comprising a memory storing a computer program and a processor that when executing the computer program implements the neural network-based option pricing method of any of claims 1 to 7.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the neural network-based option pricing method of any of claims 1 to 7.