A distributed quantitative strategy development platform based on cloud computing

Abstract

The present invention relates to a distributed quantitative strategy development platform based on cloud computing; the distributed quantitative strategy development platform consists of a factor development and analysis module, a machine learning model development and analysis module, a machine learning model integration module, an asset portfolio strategy development and analysis optimization module, a distributed cloud computing engine module and a management module, has high reliability and scalability, is based on distributed cloud computing technology, combines microservice architecture and containerization technology to support high-concurrency data processing and rapid execution of complex algorithms, covers factor development and analysis functions, machine learning model development, analysis and integration functions, asset portfolio strategy development and analysis optimization functions, distributed computing engine functions and performance evaluation functions, and can effectively improve the development efficiency and execution performance of asset investment strategies.

Description

Technical Field

[0001] The present invention relates to the field of quantitative strategy development platforms, and in particular to a distributed quantitative strategy development platform based on cloud computing. Background Art

[0002] In the financial field, millions of transactions are carried out around the world every second, and the fluctuations and changes in the financial market affect the trend of the global economy. In this highly competitive and dynamically changing environment, investors and financial institutions face huge challenges and opportunities. In this context, quantitative investment has become one of the important trends in the financial field.

[0003] Quantitative investment companies use mathematics, statistics and computer technology to analyze market data and execute trading decisions by developing complex algorithms and models. This investment strategy based on data and scientific methods can not only improve investment returns, but also reduce investment risks and quickly adjust investment portfolios in a short period of time. Therefore, more and more investors and financial institutions are turning to quantitative investment to gain competitive advantages and stable returns.

[0004] In the field of quantitative investment, quantitative strategy development platforms play a key role. These platforms provide quantitative investors with basic tools and resources, enabling them to develop, test and execute quantitative investment strategies. Through these platforms, investors can quickly analyze market data, develop new trading strategies, and perform backtesting and optimization to manage portfolios more effectively and achieve robust investment returns.

[0005] In the field of quantitative strategy development, platforms can be roughly divided into two categories: online and offline. Both types of platforms have unique functions and service features to meet the needs of different users.

[0006] For online quantitative strategy development platforms, this type of platform provides users with an instant access environment without the need to install any software locally.

[0007] The characteristics of this type of platform include:

[0008] (1) Instant access and collaboration: Users can access the platform instantly through the Internet to write, test, and optimize strategies. This instant feature facilitates team collaboration, allowing multiple users to work simultaneously under the same strategy. In addition, the platform usually provides real-time communication tools to facilitate communication and exchanges between team members.

[0009] (2) Integrated data sources. The platform integrates a rich set of financial data sources, including historical data and real-time data. Users can directly obtain the required data on the platform to conduct strategy backtesting and analysis without having to purchase additional data licenses or manually download data. This feature of integrating financial data sources greatly simplifies the user's workflow and improves work efficiency.

[0010] (3) Ease of use and flexibility. The platform provides an intuitive user interface and rich strategy development tools, and supports popular programming languages ​​such as Python. Users can easily get started regardless of whether they have a solid programming foundation. At the same time, users can flexibly adjust the platform settings and tools according to their needs in order to achieve a personalized development environment.

[0011] As for offline quantitative strategy development platforms, such platforms focus more on providing a research and development toolset that can run in a local environment.

[0012] The characteristics of this type of platform include:

[0013] (1) To run in a local environment, users need to install the software package and necessary dependencies on their local computers. Although the platform requires a certain amount of configuration work in the early stages of use, it also gives users greater flexibility and control, especially when dealing with large amounts of data and complex calculations.

[0014] (2) Data security: For users with high data sensitivity and security requirements, the platform can ensure that all data and policies are stored locally, greatly reducing the risk of data leakage; and users can manage data storage and access rights themselves to ensure data security and privacy.

[0015] (3) Customized development environment. The platform allows users to customize the development environment according to their own needs, including selecting specific versions of data analysis libraries, machine learning frameworks, etc. Users can flexibly configure the development environment according to project requirements and personal preferences to improve work efficiency.

[0016] (4) No need to rely on the Internet. Strategy development and testing can be carried out even in an environment without an Internet connection. This is suitable for work scenarios that require working in a closed or secure environment. Users do not have to worry about the risk of unstable network connection or data leakage, and can focus more on strategy development and research.

[0017] In view of this, online quantitative strategy development platforms and offline quantitative strategy development platforms each have their own advantages, and users can choose the appropriate platform type according to their specific needs and working environment.

[0018] However, as traditional quantitative strategy development platforms, both online quantitative strategy development platforms and offline quantitative strategy development platforms have some shortcomings, such as:

[0019] (1) Lack of machine learning model functions. Traditional quantitative strategy development platforms lack machine learning models. As the number of factors increases, it is not possible to quickly and easily integrate existing factors, which is not conducive to the development of corresponding work.

[0020] (2) There is a lack of flexible multi-level quantitative process support and customized task scheduling mode. Traditional quantitative strategy development platforms do not support the corresponding developers to be independently responsible for their own work, and do not support the function of allowing other process developers to use their own output results without leaking the source code.

[0021] (3) Lack of performance appraisal function necessary for management teams. In traditional quantitative strategy development platforms, there is a lack of effective team performance appraisal function, which may lead to team members lacking clear goals and motivation, thus affecting the overall efficiency and work results of the team. In addition, unclear goal setting and evaluation results, lack of clear performance goals and indicators, may lead to team members not being clear about work priorities and expected results, making it difficult to evaluate individual and team performance. At the same time, the incentive mechanism is insufficient, and team members may lack enthusiasm and motivation, resulting in unstable work efficiency and quality.

[0022] (4) Computing resources cannot be easily expanded, and the platform cannot be privately deployed. As the number of users increases, the required resources cannot be easily expanded. In addition, some peak demands cannot be easily increased according to actual needs, and there is a lack of the ability to privately deploy.

[0023] In summary, there are many problems with traditional quantitative strategy development platforms, including but not limited to the lack of machine learning model functions, the inability to use computing resources on demand, and the lack of flexible multi-level quantitative processes. These problems seriously affect the development process of quantitative strategies and the optimization process of work efficiency. Therefore, how to solve the above problems through technical improvements and platform optimization is a technical problem that currently needs to be solved urgently by technical personnel in this field.

[0024] To this end, the present invention provides a distributed quantitative strategy development platform based on cloud computing. Summary of the invention

[0025] Based on this, it is necessary to provide a distributed quantitative strategy development platform based on cloud computing to address the above technical issues.

[0026] In order to achieve the above object, the technical solution of the present invention is as follows.

[0027] A distributed quantitative strategy development platform based on cloud computing, the distributed quantitative strategy development platform comprises a factor development and analysis module, a machine learning model development and analysis module, a machine learning model integration module, an asset portfolio strategy development and analysis optimization module, a distributed cloud computing engine module and a management module, wherein: the factor development and analysis module is used to construct, manage, test and evaluate the factors required for the quantitative investment strategy, and generate effective factors; the machine learning model development and analysis module is used to construct, manage, test and evaluate the machine learning model adapted to the financial field, and generate effective machine learning models; the machine learning model integration module is used to construct, manage, test and evaluate the machine learning model integration scheme; the asset portfolio strategy development and analysis optimization module is used to formulate, manage, test and evaluate the asset composition strategy; the distributed cloud computing engine module is used to manage and allocate computing resources, adjust the execution order of computing tasks according to task priorities, deploy a unified operating environment for computing tasks using containerization technology, and expand additional computing capacity during resource demand peaks based on computing task requirements and cloud service technology; the management module is used to query log records, manage personnel accounts, evaluate work performance and obtain user feedback.

[0028] In a possible implementation, the factor development and analysis module includes a factor upload and editing unit, a factor library management unit and a factor testing and evaluation unit, wherein the factor upload and editing unit is used to obtain and edit factors required for quantitative investment strategies according to an online code editor; the factor library management unit is used to store factors in a factor library, and to perform version control management and classification management on the factors; the factor testing and evaluation unit is used to test and evaluate a single factor according to an evaluation indicator, generate a test report corresponding to the single factor, and test and evaluate the factor according to the generated test report and a preset benchmark machine learning model, generate a factor evaluation result for evaluating the relative advantages and disadvantages of the factors, and preset a factor basic evaluation scoring threshold, compare the factor evaluation result with the factor basic evaluation scoring threshold, and select the factor whose factor evaluation result meets the factor basic evaluation scoring threshold as the valid factor.

[0029] In a possible implementation, the version control management and classification management of factors specifically include: marking the name, author, usage rights and classification attributes of the factor during the process of storing the factor in the factor library; associating the corresponding version number based on the name and author of the factor; configuring the corresponding level authority based on the usage rights of the factor; and configuring the corresponding factor classification label based on the classification attributes of the factor.

[0030] In one possible implementation, the machine learning model development and analysis module includes a model building unit, a model library management unit, and a test verification and model evaluation unit, wherein the model building unit is used to develop and optimize the machine learning model according to the online code editor; the model library management unit is used to store the machine learning model in the model library, and to perform version control management, permission management, and label classification management on the machine learning model; the test verification and model evaluation unit is used to test and evaluate the machine learning model based on a preset benchmark factor combination according to the single factor testing technology, obtain and store the capability evaluation result of the corresponding machine learning model, and preset the basic evaluation scoring threshold of the machine learning model, compare the capability evaluation result with the basic evaluation scoring threshold of the machine learning model, and select the machine learning model whose capability evaluation result meets the basic evaluation scoring threshold of the machine learning model as the valid machine learning model.

[0031] In one possible implementation, the machine learning model integration module includes a machine learning model integration solution creation unit, a machine learning model integration solution library management unit, and a machine learning model integration solution testing, verification, and evaluation unit, wherein the machine learning model integration solution creation unit is used to obtain effective factors and effective machine learning models, and construct a machine learning model integration solution based on the time series dimension and the cross-sectional dimension according to the particularity of financial data; the machine learning model integration solution library management unit is used to store and manage the constructed machine learning model integration solution; the machine learning model integration solution testing, verification, and evaluation unit is used to test and evaluate the machine learning model integration solution according to the single factor testing technology, and obtain and store the test results of the corresponding machine learning model integration solution's prediction ability.

[0032] In a possible implementation, the particularities of the financial data include extensiveness, comprehensiveness, and continuity.

[0033] In a possible implementation, the asset portfolio strategy development and analysis optimization module includes an optimization algorithm uploading functional unit, an asset portfolio strategy construction unit, a portfolio backtesting and analysis unit, and an asset portfolio strategy management unit, wherein the optimization algorithm uploading functional unit is used to obtain and edit the optimization algorithm source code required for the asset portfolio strategy according to the online code editor; the asset portfolio strategy construction unit is used to select an existing optimization algorithm according to different needs, configure the constraints of the asset portfolio, set the corresponding objective function and penalty terms, and filter out the corresponding machine learning model integration scheme from the machine learning model integration scheme library management unit to obtain the asset portfolio strategy; the portfolio backtesting and analysis unit is used to use the backtesting tool to perform backtesting analysis on the asset portfolio configuration and generate corresponding backtesting results; the asset portfolio strategy management unit is used to store the asset portfolio strategy in the asset portfolio strategy library, and perform version control management and classification management on the asset portfolio strategy.

[0034] In one possible implementation, the distributed cloud computing engine module includes a resource management unit, a priority management unit, a unified operating environment deployment unit, and a cloud service dynamic integration unit, wherein the resource management unit is used to use task queues and scheduling algorithms to perform time-sharing sharing processing on computing resources; the priority management unit is used to allocate computing resources accordingly according to the priority ranking of computing tasks; the unified operating environment deployment unit is used to deploy a unified operating environment using containerization technology; and the cloud service dynamic integration unit is used to configure additional computing resources through cloud service technology during the peak period of resource demand according to the needs of computing tasks.

[0035] In one possible implementation, according to the needs of the computing task, during the peak period of resource demand, additional computing resources are configured through cloud service technology, specifically including: calculating the number of additional computing resources required for the current task based on the existing computing resources and the computing resources required for the current task execution, and judging whether the existing computing resources meet the number of additional computing resources required for the current task. If not, additional computing resources are configured through cloud service technology; if so, there is no need to configure additional computing resources through cloud service technology.

[0036] In one possible implementation, the management module includes a personnel account management unit, a performance evaluation unit, a log recording and auditing unit, and a user feedback and support unit, wherein the personnel account management unit is used to manage user accounts and assign different roles and different access rights to different users according to different application usage environments; the performance evaluation unit is used to track and evaluate the work performance of team members based on the completion, quality and efficiency of each team member's work tasks; the log recording and auditing unit is used to view and record operation logs; the user feedback and support unit is used to obtain and respond to questions, suggestions or opinions raised by users, and optimize the distributed quantitative strategy development platform according to the corresponding questions, suggestions or opinions.

[0037] The advantages and beneficial effects of the present invention are:

[0038] The present invention provides a distributed quantitative strategy development platform based on cloud computing. The distributed quantitative strategy development platform consists of a factor development and analysis module, a machine learning model development and analysis module, a machine learning model integration module, an asset portfolio strategy development and analysis optimization module, a distributed cloud computing engine module and a management module. It has high reliability and scalability. It is based on distributed cloud computing technology, combined with microservice architecture and containerization technology to support high-concurrency data processing and rapid execution of complex algorithms. It covers factor development and analysis functions, machine learning model development, analysis and integration functions, asset portfolio strategy development and analysis optimization functions, distributed computing engine functions and performance evaluation functions, which can effectively improve the development efficiency and execution performance of asset investment strategies. By providing an integrated operating environment, users can complete the design, testing, optimization and deployment of strategies under the same interface, greatly simplifying the research and development process of quantitative investment; in addition, through access rights and level permission setting technology, it can protect the intellectual property rights of developers, protect data security and the reliability of platform operation, thereby effectively preventing intellectual property leakage and data security issues, and can further enhance the intelligence level of the platform through artificial intelligence and machine learning technologies, thereby providing users with higher quality services; and, through the elastic computing resource allocation mechanism based on cloud computing, users are allowed to dynamically allocate and adjust computing resources according to actual needs, so as to maximize the utilization of computing resources, making quantitative strategy development flexible and scalable. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The figure is a schematic diagram of the structure of a distributed quantitative strategy development platform based on cloud computing in an embodiment of the present invention.

[0040] Figure 2 Schematic diagram of the structure of the factor development and analysis module in an embodiment of the present invention.

[0041] Figure 3 Schematic diagram of the structure of the machine learning model development and analysis module in an embodiment of the present invention.

[0042] Figure 4 Schematic diagram of the structure of the machine learning model integration module in an embodiment of the present invention.

[0043] Figure 5 It is a schematic diagram of the structure of the asset portfolio strategy development and analysis optimization module in an embodiment of the present invention.

[0044] Figure 6 Schematic diagram of the structure of a distributed cloud computing engine module in an embodiment of the present invention.

[0045] Figure 7 Schematic diagram of the structure of the management module in the embodiment of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] In view of the many problems existing in traditional quantitative strategy development platforms, the present invention provides a distributed quantitative strategy development platform based on cloud computing, which is an efficient and reliable tool for quantitative investors to help them optimize their resource investment portfolios in order to achieve a robust return on investment.

[0048] See also Figure 1-7 The distributed quantitative strategy development platform includes factor development and analysis module, machine learning model development and analysis module, machine learning model integration module, asset portfolio strategy development and analysis optimization module, distributed cloud computing engine module and management module.

[0049] In this embodiment, the factor development and analysis module is used to construct, manage, test and evaluate the factors required for quantitative investment strategies, and generate effective factors.

[0050] Furthermore, the factor development and analysis module includes a factor uploading and editing unit, a factor library management unit and a factor testing and evaluation unit, wherein the factor uploading and editing unit is used to obtain and edit factors required for quantitative investment strategies according to an online code editor; the factor library management unit is used to store factors in a factor library, and to perform version control management and classification management on factors; the factor testing and evaluation unit is used to test and evaluate a single factor according to an evaluation indicator, generate a test report corresponding to the single factor, and test and evaluate the factor according to the generated test report and a preset benchmark machine learning model, generate a factor evaluation result for evaluating the relative advantages and disadvantages of the factors, and preset a factor basic evaluation scoring threshold, compare the factor evaluation result with the factor basic evaluation scoring threshold, and select the factor whose factor evaluation result meets the factor basic evaluation scoring threshold as the valid factor.

[0051] Furthermore, in the process of storing the factors in the factor library, the name, author, usage rights and classification attributes of the factors are marked; based on the name and author of the factors, the corresponding version number is associated; based on the usage rights of the factors, the corresponding level permissions are configured; based on the classification attributes of the factors, the corresponding factor classification labels are configured.

[0052] Among them, the factor uploading and editing unit involved in the present invention uses an online code editor to upload and edit the factors required by the quantitative strategy, and provides functions such as syntax highlighting and code verification. In the framework of the online code editor, the CodeMirror front-end component is integrated into the framework project, and the corresponding language mode is configured, which can make the coding process more intuitive and efficient, thereby greatly improving the development efficiency and code quality; the python flake8 library is used to check the code and list all the code style problems and potential errors. At the same time, the output information of the python flake8 library also includes the file where the error occurred, the line number and the error description, so as to realize the logical function of code verification.

[0053] Among them, the factor library management unit involved in the present invention, as a private database of the user, provides a safe and reliable environment to store and manage factor codes, and at the same time, can perform version control management and classification management on factors. The supported version control management function allows users to manage different versions of factors. Each factor has an associated version number in the factor library, which is used to identify different versions. By recording the name and author of the factor and associating it with the corresponding version number, the historical changes of the factor can be traced. In addition, according to the use rights of the factor, the corresponding level authority is configured, and the level authority is provided externally to manage the use rights of the factor. The level authority includes two levels of public and private authority. The public level authority of the factor can ensure that other users can call and view the results of the shared factor, but cannot view the source code of the factor. The private level authority of the factor can prevent other users from calling the corresponding factor. Changes to the factor only exist in the account permissions of the administrator and the author. In addition, the factor library management unit also has the function of storing factors using multiple storage methods to protect data security. In addition, by marking factor classification labels, users can quickly find the required factors. For example, when classifying and labeling according to different sources, the classification labels of corresponding factors can be marked according to fundamental factors, quantity and price factors, alternative factors, etc.; when classifying and labeling according to different data frequencies, the classification labels of corresponding factors can be marked according to high-frequency factors, low-frequency factors, market factors, etc.

[0054] Among them, the factor testing and evaluation unit involved in the present invention, by adopting single factor testing technology, determines whether a certain factor has a corresponding correlation with the actual stock return rate, so as to determine whether the factor has a certain predictive ability, that is, whether the factor can be used to characterize the stock's return performance in the future period, and use this as the basis for building an investment strategy; the test uses a variety of indicators to evaluate the performance of the factor, including but not limited to IC / IR indicators, grouped IC / IR, market value stratified IC / IR and other multi-dimensional indicators, providing users with a comprehensive factor performance analysis. After testing and evaluating the factors using a variety of indicators, a corresponding test report is generated. The test report has an automatic storage function, which can ensure that the results of each factor test are recorded and reviewed, which helps users quickly identify and optimize factor performance. It should be noted here that the distributed quantitative strategy development platform involved in the present invention focuses on two dimensions for judging the results of factor testing, including the accuracy of stock return prediction and the stability of prediction. At the same time, the test report needs to include the performance of each factor in different time periods, as well as the corresponding evaluation indicators, and then select a preset machine learning model as a benchmark. This preset benchmark machine learning model can select existing machine learning algorithms, including but not limited to linear regression, random forest, etc. The specific selection principle depends on the application scenario and the characteristics of the factor. Use the benchmark machine learning model to perform single factor regression analysis on each factor to obtain the regression coefficient of each factor. The relative effect of the factor is judged according to the size of the regression coefficient, and the scoring threshold is evaluated based on the preset factor basis, and the factor with a larger coefficient is selected as the effective factor.

[0055] In this embodiment, the present invention can provide users with a flexible factor input interface by setting up a factor development and analysis module, so that users can quickly develop and test various quantitative investment factors, and perform single factor regression analysis on the factors through a benchmark machine learning model to obtain effective factors, thereby achieving the purpose of expanding and maintaining the factor pool, and providing users with a comprehensive and efficient factor development and analysis tool.

[0056] In this embodiment, the machine learning model development and analysis module is used to build, manage, test and evaluate machine learning models that are suitable for the financial field, and to generate effective machine learning models.

[0057] Furthermore, the machine learning model development and analysis module includes a model building unit, a model library management unit, and a test verification and model evaluation unit, wherein the model building unit is used to develop and optimize the machine learning model according to the online code editor; the model library management unit is used to store the machine learning model in the model library, and perform version control management, permission management and label classification management on the machine learning model; the test verification and model evaluation unit is used to test and evaluate the machine learning model based on a preset benchmark factor combination according to the single factor testing technology, obtain and store the capability evaluation result of the corresponding machine learning model, and preset the basic evaluation scoring threshold of the machine learning model, compare the capability evaluation result with the basic evaluation scoring threshold of the machine learning model, and select the machine learning model whose capability evaluation result meets the basic evaluation scoring threshold of the machine learning model as the valid machine learning model.

[0058] Among them, the model building unit involved in the present invention optimizes the development of the machine learning model using an online code editor, and provides syntax highlighting and code verification functions. Using the syntax highlighting and code verification functions integrated in the online code editor, it is possible to provide users with an intuitive and efficient model building environment, so that the entire process from model building to training becomes smoother and more efficient. At the same time, the syntax highlighting and code verification functions can provide technical support for a variety of machine learning algorithms, making it convenient for users to package and develop related machine learning algorithms already available on the market, and further improve the development efficiency of machine learning models. In addition, the developed machine learning model has data processing capabilities, and developers can choose corresponding appropriate data preprocessing methods in the process of developing different machine learning models, including but not limited to using steps such as cleaning, removing outliers, and filling missing values ​​to implement data preprocessing processes to ensure data quality.

[0059] It should be understood that the machine learning model stored in the model library management unit involved in the present invention represents a machine learning model developed and optimized using a model building unit, and the development and optimization process of the machine learning model includes a data processing process for input and output data and a call to an algorithm library component; wherein the algorithm library component refers to a program code library of an existing machine learning basic component, such as: Scikit-Learn, PyTorch, TensorFlow, etc.

[0060] Among them, the model library management unit involved in the present invention realizes the storage and management of machine learning models through version control, authority management and label classification functions. The model library management unit provides a safe and reliable environment to store and manage the developed machine learning models, and at the same time, can perform version control management and classification management on the machine learning models; each machine learning model has an associated version number in the model library, which is used to identify different versions. By recording the name and author of the machine learning model and associating it with the corresponding version number, the historical changes of the machine learning model can be traced. In addition, according to the use rights of the machine learning model, the corresponding level rights are configured, and the level rights are provided externally to manage the use rights of the machine learning model. The level rights include public and private level rights. The public level rights of the machine learning model can ensure that other users can call and view the results of the shared machine learning model, but cannot view the source code constructed by the machine learning model. The private level rights of the machine learning model can prevent other users from calling the corresponding machine learning model. At the same time, according to the machine learning model has different classification attributes such as version information, author information, and use rights, the present invention also provides a label function so that users can filter by classification labels.

[0061] Among them, the test verification and model evaluation unit involved in the present invention is used to test and verify the following process: according to historical market data and preset training set time period and test set time period, calculate the training data and test data corresponding to the effective factor, input the training data into the machine learning model for training, and obtain the parameters of the trained machine learning model; it is expected that through the aforementioned training process, the prediction ability of the machine learning model and the parameters obtained by training on the test data is improved. Through the built-in single factor testing function of the distributed quantitative strategy development platform, it is possible to provide a verification tool for the developed machine learning model and support a variety of evaluation indicators so as to reveal the performance and potential of the machine learning model in multiple dimensions. Among them, the single factor test parameters include but are not limited to Sharpe ratio, maximum drawdown, return volatility and win rate. After the test verification and model evaluation unit tests the machine learning model, it will generate a corresponding test result report, and use the test result report as the basis for the ability evaluation of the machine learning model. The specific evaluation method includes: testing and evaluating different machine learning models respectively through a preset benchmark factor combination, and distinguishing the advantages and disadvantages between machine learning models according to the feedback of the test result report and the preset machine learning model basic evaluation score threshold. During the testing process, it is necessary to accurately evaluate the machine learning model according to the complexity and variability of financial market data, including: the selection process of the machine learning model, the cross-validation process, the tuning process of the machine learning model, and the real-time monitoring and updating process. Among them, the appropriate machine learning model is selected according to the characteristics of the data. The machine learning model here includes traditional statistical models and deep learning models; the cross-validation process is used to verify the generalization ability and stability of the machine learning model; the performance of the machine learning model can be improved by optimizing the parameters of the model to tune the machine learning model; by establishing and implementing a monitoring system, the purpose of real-time monitoring and updating can be achieved, which is used to regularly update the machine learning model to adapt to market changes. In addition, the model parameters of the machine learning model are continuously adjusted through a preset benchmark factor combination, which can complete the iteration and update of the machine learning model, and the test results will also be saved, so that developers can view historical results conveniently and quickly, thereby improving development efficiency, and providing managers with a basis for each developer (i.e. team member) to evaluate work performance.

[0062] Among them, the Sharpe Ratio measures the rate of return per unit risk and is a commonly used indicator for evaluating the risk-adjusted return of an investment portfolio.

[0063] Among them, Max Drawdown: evaluates the maximum loss of an investment portfolio in a specific period, reflecting the riskiness and risk resistance of the investment strategy.

[0064] Among them, return volatility (Volatility): measures the volatility of an investment portfolio or asset price, and is one of the important indicators for assessing investment risk.

[0065] Among them, Win Rate: the ratio of the number of profitable times of an investment portfolio or strategy to the total number of transactions, which is used to evaluate the stability and profitability of the strategy.

[0066] In this embodiment, by setting up a machine learning model development and analysis module, at the machine learning model development level, not only a variety of classic and advanced machine learning algorithms and models are provided to help investors analyze market data and predict future market trends, but also historical market data can be used to train and verify these models, and the performance of the models can be evaluated through single-factor testing techniques such as cross-validation. The machine learning models include but are not limited to linear regression, support vector machines, random forests, neural networks, etc.; in addition, a model tuning function is also provided to facilitate users to adjust the parameters of the model in order to improve the predictive ability and robustness of the model.

[0067] In this embodiment, the machine learning model integration module is used to build, manage, test and evaluate machine learning model integration solutions.

[0068] Furthermore, the machine learning model integration module includes a machine learning model integration scheme creation unit, a machine learning model integration scheme library management unit, and a machine learning model integration scheme testing, verification, and evaluation unit, wherein the machine learning model integration scheme creation unit is used to obtain effective factors and effective machine learning models, and construct a machine learning model integration scheme based on the time series dimension and the cross-sectional dimension according to the particularity of financial data; the machine learning model integration scheme library management unit is used to store and manage the constructed machine learning model integration scheme; the machine learning model integration scheme testing, verification, and evaluation unit is used to test and evaluate the machine learning model integration scheme according to the single factor testing technology, and obtain and store the test results of the corresponding machine learning model integration scheme's prediction ability.

[0069] Among them, the machine learning model integration scheme creation unit involved in the present invention obtains effective factors and effective machine learning models, and constructs a machine learning model integration scheme from the two perspectives of time series and cross-section according to the particularity of financial data, which refers to the unique attributes and characteristics of data in the financial field, including but not limited to extensiveness, comprehensiveness and continuity. In the process of constructing the machine learning model integration scheme, the user can start from the characteristics of the model and the diversity of the data, and use different financial data and machine learning models to pair each other, so as to achieve the purpose of using the most suitable machine learning model to process the corresponding data, and finally integrate the prediction capabilities of all machine learning models together, thereby improving the prediction capabilities of stock returns. For example: for time series data of different frequencies, taking minute frequency, hour frequency, and biweekly frequency as examples, the user can respectively use machine learning models suitable for different processing time series data to process the aforementioned data, such as: using LSTM model to process minute frequency data, using Transformer model to process hour frequency data, using GRU model to process biweekly frequency data, and generating corresponding processing results, and then assigning different weight values ​​to each processing result, and constructing a machine learning model integration scheme that meets the expected needs of the user through weight adjustment.

[0070] Among them, the breadth: covers various types of financial data, including market data (such as stock prices, trading volume), financial data (such as company financial reports), macroeconomic data, etc.

[0071] Among them, comprehensiveness: financial data is usually composed of data from multiple different sources and of different types, and it is necessary to comprehensively consider multiple data to obtain more comprehensive information.

[0072] Among them, continuity: Financial data is usually collected continuously in chronological order, including time series data and cross-sectional data. Therefore, the time dimension plays a vital role in analysis and modeling. In particular, time series data is heterogeneous and has different characteristics at different time nodes or frequencies.

[0073] Among them, the cross-sectional dimension refers to the data corresponding to different individuals (such as stocks and companies) at a specific point in time. Cross-sectional data usually reflects the characteristics of each individual at a certain point in time.

[0074] Among them, the machine learning model integration solution library management unit involved in the present invention realizes the storage and management of the machine learning model integration solution through version control and label classification functions. The machine learning model integration solution library provides a safe and reliable environment to store and manage the developed machine learning model integration solution, and at the same time, it can perform version control management and classification management on the machine learning model integration solution; each machine learning model integration solution has an associated version number in the machine learning model integration solution library, which is used to identify different versions, and the name and author of the machine learning model integration solution are recorded and associated with the corresponding version number, so as to trace the historical changes of the machine learning model integration solution. In addition, according to the use rights of the machine learning model integration solution, the corresponding level rights are configured, and the level rights are provided externally to manage the use rights of the machine learning model integration solution. The level rights include public and private level rights. The public level rights of the machine learning model integration solution can ensure that other users can view the results of the shared machine learning model integration solution, but can view the construction process and detailed processing process of the machine learning model integration solution. The private level rights of the machine learning model integration solution can prevent other users from obtaining the corresponding machine learning model integration solution results. At the same time, according to the machine learning model integration scheme with different version information, author information, usage factors, model information, configuration parameter information and other classification attributes, the present invention also provides a labeling function so that users can filter by classification labels.

[0075] Among them, the machine learning model integration scheme test verification and evaluation unit involved in the present invention tests and evaluates the constructed machine learning model integration scheme to enhance the stability of the prediction ability of the machine learning model integration scheme. Through the built-in single-factor testing function of the distributed quantitative strategy development platform, it is possible to provide a verification tool for the machine learning model integration scheme and support a variety of evaluation indicators to reveal the performance and potential of the machine learning model integration scheme in multiple dimensions. Among them, the single-factor test parameters include but are not limited to the Sharpe ratio, maximum drawdown, return volatility and win rate. By evaluating the machine learning model integration scheme, a machine learning model integration scheme with higher prediction accuracy and higher prediction stability for stock returns can be obtained.

[0076] In this embodiment, the present invention sets up a machine learning model integration module, adopts effective factors and effective machine learning models, and constructs a machine learning model integration solution from the two perspectives of time series and cross-section according to the particularity of financial data, thereby achieving the purpose of improving the stock return prediction ability and providing users with a comprehensive and efficient machine learning model integration solution construction and analysis tool.

[0077] In this embodiment, the asset portfolio strategy development and analysis optimization module is used to formulate, manage, test and evaluate asset composition strategies.

[0078] Furthermore, the asset portfolio strategy development and analysis optimization module includes an optimization algorithm uploading functional unit, an asset portfolio strategy construction unit, a portfolio backtesting and analysis unit, and an asset portfolio strategy management unit, wherein the optimization algorithm uploading functional unit is used to obtain and edit the optimization algorithm source code required for the asset portfolio strategy according to the online code editor; the asset portfolio strategy construction unit is used to select existing optimization algorithms according to different needs, configure asset portfolio constraints, set corresponding objective functions and penalty items, and filter out corresponding machine learning model integration solutions from the machine learning model integration solution library management unit to obtain asset portfolio strategies; the portfolio backtesting and analysis unit is used to use backtesting tools to perform backtesting analysis on asset portfolio configurations and generate corresponding backtesting results; the asset portfolio strategy management unit is used to store asset portfolio strategies in the asset portfolio strategy library, and to perform version control management and classification management on asset portfolio strategies.

[0079] Among them, the optimization algorithm uploading function unit involved in the present invention can enable users to upload and edit the optimization algorithm source code required for resource combination optimization by setting up an online code editor, so as to enhance the user's coding experience; at the same time, the online code editor integrates syntax highlighting and code verification functions, which can help users quickly locate problems and optimize algorithms.

[0080] Among them, the asset portfolio strategy construction unit involved in the present invention can realize the functions of selecting the uploaded optimization algorithm, setting corresponding constraints, objective functions and penalty items, and screening out the corresponding machine learning model integration scheme from the machine learning model integration scheme library management unit, so that users can flexibly configure asset portfolio strategies according to different needs such as individual stock preferences, industry preferences, index preferences, risk control, etc., reflecting the user orientation and flexibility of the distributed quantitative strategy development platform, so as to meet the personalized needs of different users in the asset portfolio strategy development process.

[0081] Furthermore, different needs include but are not limited to the following situations: (1) For the use scenario of individual stock allocation, limit investment in certain specific stocks or industries, such as only investing in leading companies in a specific industry or stocks with stable growth potential; (2) For the use scenario of industry allocation, limit investment in specific industries or industry weights, such as controlling the weight of certain industries to not exceed a certain proportion, or adjusting the industry allocation ratio according to the industry business cycle; (3) For the use scenario of index allocation, set relevant constraints for tracking the index, such as the tracking error does not exceed a certain level, or the weight of the index constituent stocks matches the index; (4) For the use scenario of risk control, set risk control constraints, such as maximum risk exposure, maximum single asset or industry weight, maximum volatility, etc., to ensure that the risk of the investment portfolio is controllable under different market conditions.

[0082] Among them, the combined backtesting and analysis unit involved in the present invention utilizes a backtesting tool to generate the asset configuration of the next holding period before each position change date within the selected backtesting period according to a specific asset portfolio configuration, and then uses the backtesting tool to perform backtesting analysis on the asset configuration of each holding period in history, and finally obtains the backtesting result, which is used to simulate the performance of the corresponding asset portfolio strategy on historical data, evaluate and quantify the return performance and risk characteristics of the asset portfolio strategy under different market conditions, and can provide users with decision support tools to help users scientifically formulate and adjust asset portfolio strategies.

[0083] In this embodiment, by setting up an asset portfolio strategy development and analysis optimization module, flexible asset portfolio strategy construction tools and backtesting tools are provided, so that users can build asset portfolio strategies according to different needs and simulate the performance of the asset portfolio strategy on historical data, providing users with important decision support tools.

[0084] In this embodiment, the distributed cloud computing engine module is used to manage and allocate computing resources, adjust the execution order of computing tasks according to task priorities, use containerization technology to deploy a unified operating environment for computing tasks, and expand additional computing power during peak resource demand periods based on the needs of computing tasks and cloud service technology.

[0085] Furthermore, the distributed cloud computing engine module includes a resource management unit, a priority management unit, a unified operating environment deployment unit and a cloud service dynamic integration unit, wherein the resource management unit is used to use task queues and scheduling algorithms to perform time-sharing sharing processing on computing resources; the priority management unit is used to allocate computing resources accordingly according to the priority ranking of computing tasks; the unified operating environment deployment unit is used to deploy a unified operating environment for computing tasks using containerization technology; and the cloud service dynamic integration unit is used to configure additional computing resources through cloud service technology according to the needs of computing tasks during the peak period of resource demand.

[0086] The resource management unit of the present invention can realize the alternate execution of tasks on existing resources through task queues and scheduling algorithms. When multiple users submit multiple tasks, these tasks will be added to the task queue. According to the types of resources required for the tasks and the number of existing resources, the preset scheduling algorithm determines the execution order of the tasks in order to maximize the utilization of resources. In particular, when the user is in a rest cycle, the tasks in the task queue can be continuously executed so that the computing tasks can be completed and the computing results can be provided before the next working cycle of the user arrives, so as to fully utilize the computing resources.

[0087] The priority management unit involved in the present invention, when performing task scheduling management, comprehensively considers the urgency of the task and the needs of the user, and intelligently allocates computing resources to ensure that high-priority tasks can be executed in a timely manner.

[0088] Among them, the unified operating environment deployment unit involved in the present invention utilizes containerization technology to provide a unified operating environment for all computing tasks, which not only simplifies the environment configuration process, but also ensures the uniformity of the operating environment of computing tasks on different computing nodes, thereby improving the deployment efficiency; for example, the algorithm library component involved in the present invention also needs to be pre-installed in a unified operating environment for use, so as to provide users with convenient computing services.

[0089] Among them, the cloud service dynamic integration unit involved in the present invention, when it is monitored that the existing resources cannot meet the needs of the task, configures additional computing resources through cloud service technology during the peak period of resource demand, and automatically triggers the expansion mechanism. When the resource demand decreases, when the computing resources of the local computing node are sufficient to support the computing task, it is necessary to release the previously additionally expanded cloud service computing resources. It should be noted here that a unified environment deployment method must also be used for the expanded cloud service computing resources in order to improve the user experience.

[0090] In this embodiment, by setting up a distributed cloud computing engine module, efficient computing task execution and resource management functions are provided. By utilizing resource management, priority management, unified operating environment deployment and cloud service dynamic integration technology, resource utilization is greatly optimized, operating costs are reduced, and the consistency and stability of computing tasks in different environments can be ensured.

[0091] In this embodiment, the management module is used to query log records, manage personnel accounts, evaluate work performance, and obtain user feedback.

[0092] Furthermore, the management module includes a personnel account management unit, a performance evaluation unit, a log recording and auditing unit, and a user feedback and support unit, wherein the personnel account management unit is used to manage user accounts and assign different roles and different access rights to different users according to different application usage environments; the performance evaluation unit is used to track and evaluate the work performance of team members based on the completion, quality and efficiency of each team member's work tasks; the log recording and auditing unit is used to view and record operation logs; the user feedback and support unit is used to obtain and respond to questions, suggestions or opinions raised by users, and optimize the distributed quantitative strategy development platform according to the corresponding questions, suggestions or opinions.

[0093] The personnel account management unit involved in the present invention provides user account management functions, including but not limited to adding new user accounts, deleting user accounts, modifying user permissions, etc. Through the personnel account management unit, the administrator can manage all users on the distributed quantitative strategy development platform to ensure that everyone has appropriate access rights and role allocations.

[0094] The performance evaluation unit involved in the present invention supports administrators to track and evaluate the work performance of team members by viewing the completion status, quality and efficiency of each team member's work tasks. Through the distributed quantitative strategy development platform, which records the completion status, quality and efficiency of each team member's work tasks, administrators can fully understand the overall performance of the team and take corresponding management measures to improve the overall productivity and efficiency of the team.

[0095] The log recording and auditing unit involved in the present invention is used to record all operation logs, including but not limited to user login, data access, task execution and other operation logs. These operation logs are used for auditing and tracking to ensure the security and compliance of the platform. Administrators can timely discover and resolve potential security issues or abnormal situations by viewing the records of the operation logs.

[0096] Among them, the user feedback and support unit involved in the present invention is used to provide users with a way to feedback issues, so that users can ask questions, suggestions or opinions to administrators through the user feedback and support unit. At the same time, administrators can respond to user feedback in a timely manner and improve and optimize the platform according to user needs to enhance user experience and satisfaction.

[0097] In this embodiment, in view of the shortcomings that some platforms cannot support multi-level quantitative processes, which is not conducive to teamwork and task management, in addition to setting up a management module to solve the aforementioned problems, the present invention can also be implemented by introducing a workflow designer and a task management and collaboration platform. In the process of introducing the workflow designer, an intuitive and flexible workflow designer is introduced to allow users to customize and adjust the quantitative process according to their specific quantitative strategy development needs, and the workflow designer is configured with a visual interface to support users to easily add, delete and adjust the order and dependencies of each task, as well as specify the input and output of the task; in the process of introducing the task management and collaboration platform, an integrated task management and collaboration platform is introduced, which enables team members to view the task list, assign tasks, update the task status and other operations in the same interface, as well as communicate and collaborate in real time, and the task management and collaboration platform is configured with functional modules such as authority management, notification reminders, and real-time chat, so that team members can efficiently complete the quantitative strategy development tasks.

[0098] In this embodiment, in view of the shortcoming that some platforms lack the performance appraisal management required by the management team, in addition to setting up a management module to solve the aforementioned problem, the present invention can also be achieved by introducing an existing performance management tool or platform and introducing a clear performance appraisal process. In the process of introducing the existing performance management tool or platform, a functionally rich and easy-to-use performance management software or online platform is introduced and integrated into the distributed quantitative strategy development platform as the main tool for performance appraisal of the management team, and the performance management tool or platform is configured with functional modules such as supporting the setting and tracking of performance goals, conducting regular evaluation and feedback, and generating performance reports to meet the management team's needs for performance appraisal; in the process of introducing a clear performance appraisal process, a clear and standardized performance appraisal process is preset and incorporated into the workflow of the distributed quantitative strategy development platform for use. The process can clearly specify the setting and evaluation standards of performance goals, regular performance evaluation and feedback mechanisms, and performance reward and punishment systems, etc., to ensure that team members can clearly understand and comply with the regulations on performance appraisal, thereby improving overall work efficiency and quality stability.

[0099] In this embodiment, by setting up a management module, core support can be provided for the operation and management of the distributed quantitative strategy development platform, including personnel account management, performance evaluation, logging and auditing, user feedback and support functions. By integrating these functions together, not only the security, compliance and stability of the distributed quantitative strategy development platform are guaranteed, but also convenient account management and work performance evaluation tools can be provided for team members.

[0100] In summary, the present invention provides a distributed quantitative strategy development platform based on cloud computing. The distributed quantitative strategy development platform is to solve the problem that users often face the problem of being unable to conveniently adjust resources according to computing needs when using traditional quantitative strategy development platforms. A cloud server (i.e., an elastic computing resource allocation mechanism based on cloud computing) is proposed to facilitate users to dynamically allocate and adjust computing resources according to actual needs, that is, users can quickly expand computing resources when demand increases, and reduce the use of computing resources accordingly when demand decreases; at the same time, the cloud server can also provide a wealth of computing resource options, including multiple computing types such as CPU and GPU, as well as a large amount of storage and network resources, so as to meet the needs of quantitative models of different complexities. In addition, the cloud server also has a high degree of automation and management functions, and can realize automatic expansion and contraction of resources through API or management console without manual intervention, which not only improves resource utilization efficiency, but also ensures that computing tasks can run efficiently and stably. At the same time, with the help of cloud servers, flexible adjustment of resource management methods can be achieved, not only can computing resources be dynamically adjusted according to the actual operation of the quantitative model, but also rapid response can be made in the event of emergencies or rapid market changes, ensuring timely adjustment and execution of quantitative strategies, thereby providing users with an efficient and flexible quantitative strategy development environment.

[0101] In addition, the distributed quantitative strategy development platform has built a comprehensive quantitative development process, which effectively promotes collaboration among team members while ensuring the security and independence of development results. Under this framework, each team member can undertake corresponding tasks according to his or her expertise and responsibilities, which not only allows team members to deeply tap their potential and focus on their best areas, but also greatly improves work efficiency and the overall quality of the project. In addition, by setting up a management module (i.e., a flexible, multi-level quantitative process support mechanism), it is possible to achieve a high degree of decoupling and free combination of various functional modules. This mechanism not only strengthens the collaboration among team members, but also makes the entire development process more transparent and standardized. At the same time, through a flexible multi-level quantitative process support mechanism, users are supported to freely combine and use each function of quantitative development to meet the needs of different teams and projects, so that users can independently take charge of their own part of the work according to their roles and responsibilities, and more conveniently collaborate and communicate with other team members, thereby improving team collaboration efficiency and project management flexibility. In addition, through access rights and level permission setting technology, it is possible to protect the intellectual property rights of developers, protect data security and the reliability of platform operation, thereby effectively preventing intellectual property leakage and data security issues; in addition, the distributed quantitative strategy development platform also provides real-time task monitoring functions, supporting users to view the execution status, progress and log information of tasks at any time, and using instant feedback mechanisms, greatly enhancing the user's control over task execution.

[0102] Obviously, it should be understood by those skilled in the art that the above-mentioned various steps of the present invention can be performed in a manner different from the present invention, and the simulation method and experimental equipment include but are not limited to the above description. The above-mentioned various steps of the present invention can be performed in an order different from that here in some cases, and the steps shown or described above can be performed separately. Therefore, the present invention is not limited to any specific combination of hardware and software.

[0103] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A distributed quantitative strategy development platform based on cloud computing, characterized in that: The distributed quantitative strategy development platform includes factor development and analysis module, machine learning model development and analysis module, machine learning model integration module, asset portfolio strategy development and analysis optimization module, distributed cloud computing engine module and management module, among which: The factor development and analysis module is used to construct, manage, test and evaluate factors required for quantitative investment strategies, and to generate effective factors; The machine learning model development and analysis module is used to build, manage, test and evaluate machine learning models that are suitable for the financial field, and to generate effective machine learning models; The machine learning model integration module is used to build, manage, test and evaluate machine learning model integration solutions; The asset portfolio strategy development and analysis optimization module is used to formulate, manage, test and evaluate asset composition strategies; The distributed cloud computing engine module is used to manage and allocate computing resources, adjust the execution order of computing tasks according to task priorities, deploy a unified operating environment for computing tasks using containerization technology, and expand additional computing capacity during peak resource demand periods based on computing task requirements and cloud service technology; The management module is used to query log records, manage personnel accounts, evaluate work performance, and obtain user feedback; The factor development and analysis module includes a factor upload and editing unit, a factor library management unit and a factor testing and evaluation unit, wherein the factor upload and editing unit is used to obtain and edit factors required for quantitative investment strategies according to an online code editor; the factor library management unit is used to store factors in a factor library, and to perform version control management and classification management on factors; the factor testing and evaluation unit is used to test and evaluate a single factor according to an evaluation index, generate a test report corresponding to a single factor, and test and evaluate the factor according to the generated test report and a preset benchmark machine learning model, generate a factor evaluation result for evaluating the relative advantages and disadvantages of factors, and preset a factor basic evaluation scoring threshold, compare the factor evaluation result with the factor basic evaluation scoring threshold, and select the factor whose factor evaluation result meets the factor basic evaluation scoring threshold as a valid factor; The version control management and classification management of the factors specifically include: in the process of storing the factors in the factor library, marking the name, author, usage authority and classification attribute of the factors; associating the corresponding version number based on the name and author of the factors; configuring the corresponding level authority based on the usage authority of the factors; configuring the corresponding factor classification label based on the classification attribute of the factors; The machine learning model development and analysis module includes a model construction unit, a model library management unit, and a test verification and model evaluation unit, wherein the model construction unit is used to develop and optimize the machine learning model according to the online code editor; the model library management unit is used to store the machine learning model in the model library, and perform version control management, permission management, and tag classification management on the machine learning model; the test verification and model evaluation unit is used to test and evaluate the machine learning model based on a preset benchmark factor combination according to the single factor testing technology, obtain and store the capability evaluation result of the corresponding machine learning model, and preset the basic evaluation scoring threshold of the machine learning model, compare the capability evaluation result with the basic evaluation scoring threshold of the machine learning model, and select the machine learning model whose capability evaluation result meets the basic evaluation scoring threshold of the machine learning model as the valid machine learning model; The machine learning model integration module includes a machine learning model integration solution creation unit, a machine learning model integration solution library management unit, and a machine learning model integration solution testing, verification, and evaluation unit, wherein the machine learning model integration solution creation unit is used to obtain effective factors and effective machine learning models, and construct a machine learning model integration solution based on the time series dimension and the cross-sectional dimension according to the particularity of financial data; the machine learning model integration solution library management unit is used to store and manage the constructed machine learning model integration solution; the machine learning model integration solution testing, verification, and evaluation unit is used to test and evaluate the machine learning model integration solution according to the single factor testing technology, and obtain and store the test results of the prediction ability of the corresponding machine learning model integration solution; the particularity of the financial data includes extensiveness, comprehensiveness, and continuity; The asset portfolio strategy development and analysis optimization module includes an optimization algorithm uploading functional unit, an asset portfolio strategy construction unit, a portfolio backtesting and analysis unit, and an asset portfolio strategy management unit, wherein the optimization algorithm uploading functional unit is used to obtain and edit the optimization algorithm source code required for the asset portfolio strategy according to the online code editor; the asset portfolio strategy construction unit is used to select an existing optimization algorithm according to different needs, configure the constraint conditions of the asset portfolio, set the corresponding objective function and penalty term, and filter out the corresponding machine learning model integration scheme from the machine learning model integration scheme library management unit to obtain the asset portfolio strategy; the portfolio backtesting and analysis unit is used to use the backtesting tool to perform backtesting analysis on the asset portfolio configuration and generate corresponding backtesting results; the asset portfolio strategy management unit is used to store the asset portfolio strategy in the asset portfolio strategy library, and perform version control management and classification management on the asset portfolio strategy; The distributed cloud computing engine module includes a resource management unit, a priority management unit, a unified operating environment deployment unit and a cloud service dynamic integration unit, wherein the resource management unit is used to use task queues and scheduling algorithms to perform time-sharing sharing of computing resources; the priority management unit is used to allocate computing resources according to the priority ranking of computing tasks; the unified operating environment deployment unit is used to use containerization technology to deploy a unified operating environment for computing tasks; the cloud service dynamic integration unit is used to configure additional computing resources through cloud service technology according to the needs of computing tasks during the peak period of resource demand; The additional computing resources are configured through cloud service technology according to the needs of the computing task during the peak period of resource demand, specifically including: calculating the number of computing resources required for the current task based on the existing computing resources and the computing resources required for the current task, and judging whether the existing computing resources meet the number of computing resources required for the current task, and if not, configuring additional computing resources through cloud service technology; if so, there is no need to configure additional computing resources through cloud service technology; The management module includes a personnel account management unit, a performance evaluation unit, a log recording and auditing unit, and a user feedback and support unit, wherein the personnel account management unit is used to manage user accounts and assign different roles and different access rights to different users according to different application usage environments; the performance evaluation unit is used to track and evaluate the work performance of team members according to the completion, quality and efficiency of each team member's work tasks; the log recording and auditing unit is used to view and record operation logs; the user feedback and support unit is used to obtain and respond to questions, suggestions or opinions raised by users, and optimize the distributed quantitative strategy development platform according to the corresponding questions, suggestions or opinions; Among them, the online code editor uses the CodeMirror front-end component, configures the corresponding language mode, uses the python flake8 library to check the code, and lists all code style issues and potential errors. At the same time, the output information of the python flake8 library also includes the file, line number and error description where the error occurred; The distributed quantitative strategy development platform also includes: introducing performance management software, integrating the performance management software into the distributed quantitative strategy development platform, the performance management software is configured with functional modules for setting and tracking performance goals, conducting regular evaluation and feedback, and generating performance reports; introducing a performance appraisal process, the performance appraisal process stipulates the setting and evaluation standards of performance goals, regular performance evaluation and feedback mechanisms, and the content of the performance reward and punishment system.

Images