Multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction

Abstract

The invention discloses a multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction, and relates to the technical field of digital gene recombination, and the method comprises the following steps: S1, multi-dimensional factor quantification; s2, carrying out digital gene recombination; s3, strategy construction and back testing are carried out; and S4, strategy optimization and real disk transaction. According to the multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction, all the steps are mutually associated and supported, and a complete and efficient quantitative analysis and optimization decision process is jointly formed; through behavior deconstruction and multi-dimensional factor mining, investment opportunities and risks in the market can be comprehensively captured, processing operations such as factor quantification are matched, and a solid foundation is provided for subsequent factor synthesis and strategy construction; according to the method, factors are screened and subjected to weight configuration based on digital gene recombination, the optimal factor combination can be efficiently found, high flexibility is achieved, and dynamic adjustment and optimization can be conducted according to market changes.

Description

Technical Field

[0001] The present invention relates to the field of digital gene recombination technology, and in particular to a multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction. Background Art

[0002] Factor quantification is an important information reorganization and optimization method, which aims to improve the accuracy and efficiency of decision-making. In factor quantification, multiple factors are selected and quantified to construct a quantitative model to evaluate potential returns and risks. Quantitative investment refers to a trading method that issues buy and sell orders in a quantitative and computer-programmed manner to obtain stable returns. It uses modern statistical and mathematical methods and computer technology to find a variety of "high-probability" strategies and rules that can bring excess returns from massive historical data. On this basis, it is comprehensively summarized into a factor model program, and finally implemented in a disciplined manner according to the quantitative models constructed by these strategies, striving to achieve stable, sustainable, and above-average excess returns.

[0003] However, in traditional quantitative strategies, factor mining and selection often rely on limited historical data and empirical knowledge, resulting in limitations in the effectiveness and stability of factors. In addition, the construction and optimization process of quantitative strategies is relatively complex and requires a lot of computing resources and time. At the same time, the performance of strategies is easily affected by market changes and lacks sufficient dynamic adaptability, making it difficult to cope with changes in the market environment and the behavior of actors.

[0004] Therefore, there is an urgent need to improve this shortcoming. The present invention studies and improves the existing technology and its shortcomings, and provides a multi-dimensional factor quantification and digital gene recombination method based on behavioral deconstruction. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction, comprising the following steps:

[0007] S1. Multi-dimensional factor quantification:

[0008] S11. Factor selection and preliminary processing: Select multi-dimensional factors related to the behavior of actors from market data, including valuation factors, volatility factors, operating capacity factors, scale factors, technical factors, leverage factors, financial quality factors, and growth factors, etc., and perform quantitative processing on the selected factors after de-extreme value, standardization, and neutralization, and convert them into computable numerical indicators;

[0009] S12. Behavior deconstruction and factor analysis: Based on behavioral principles, deeply deconstruct the behavior of actors, analyze the impact of different behavioral factors on decision-making, and conduct correlation analysis and principal component analysis on the selected factors to identify the relationship between key factors and potential factors;

[0010] S13. Multi-dimensional factor synthesis: Synthesize highly correlated factors to form composite factors. Synthesis can reduce redundant information between factors and improve the explanatory power and predictive ability of factors.

[0011] S2. Digital gene recombination:

[0012] S21. Gene encoding: Encode the digital information to be optimized into a gene sequence, that is, convert the complex factors into binary or other forms of encoding;

[0013] S22, Genetic optimization: genetic operations and recombination are performed on the encoded gene sequence to find the optimal solution, that is, the optimal factor combination;

[0014] S23, gene decoding: after the optimization process is completed, the gene sequence is decoded into original digital information, that is, the encoded gene sequence is converted back to the original digital form for subsequent analysis and application;

[0015] S3. Strategy construction and backtesting:

[0016] Construct quantitative strategies based on the optimal factor portfolio and backtest the strategies to evaluate their historical performance and risk characteristics;

[0017] S4. Strategy optimization and real trading:

[0018] Optimize and adjust the strategy based on the backtesting results, apply the optimized strategy to real trading, and make dynamic adjustments based on market changes.

[0019] Furthermore, in step S11, de-extreme value, standardization and neutralization are used to eliminate the correlation and scale effect between factors and improve the stability and accuracy of the model, as follows:

[0020] Remove extreme values: Use the 3σ method, percentile method or MAD method to eliminate extreme values ​​in the data caused by data entry errors, abnormal events or other reasons; in this embodiment, the 3σ method is selected, and the specific operations are as follows: calculate the mean μ and standard deviation σ of the factor, and set the threshold parameters, adjust the factor values ​​that exceed the range [μ-3σ, μ+3σ], replace the factor values ​​that exceed the range with the boundary values ​​(i.e., μ-3σ or μ+3σ), or directly delete these values;

[0021] Standardization: Z-score standardization or Min-Max standardization is used to convert the data into a dimensionless form for comparison and regression analysis. In this embodiment, Z-score standardization is selected. The specific operation is as follows: the mean μ and standard deviation σ of the factor are calculated, and each factor value is converted into a standard score using the formula (Z = (x-μ) / σ), and the standardized data will have a distribution with a mean of 0 and a standard deviation of 1.

[0022] Neutralization: Based on regression analysis, eliminate the factor's preference for market value or industry to ensure the fairness of factor analysis. In this embodiment, the specific operations are as follows: collect the market value, industry classification and factor data of all stocks at each time point, and use a linear regression model to neutralize each factor with respect to market value and industry at each time section. Then, use market value and industry data as independent variables and the factor to be removed as the dependent variable to perform cross-sectional regression, and extract residuals from the regression results. These residuals are the factor values ​​after eliminating the influence of market value and industry.

[0023] Furthermore, in step S12, the actor's behavior factors include the actor's psychological expectations, risk preferences, and behavior habits, as follows:

[0024] Psychological expectations: actors’ predictions and expectations of future market performance, economic prospects, policy changes, etc. Psychological expectations will affect actors’ decision-making process, forming different market expectations, thus leading to market price fluctuations;

[0025] Risk preference: the attitude and tolerance of the actor towards risk, which can be divided into risk averse (preferring low risk and avoiding high risk), risk seekers (willing to take high risk in pursuit of high return) and risk neutral (neither avoiding risk nor actively seeking risk, the criterion for selection is the size of expected return). Risk preference will affect the actor's strategy and configuration, thus affecting the market supply and demand relationship and price fluctuations;

[0026] Behavioral habits: stable behavioral patterns formed by actors through long-term practice.

[0027] Furthermore, in step S12, in the actor behavior analysis, the mutual influence between different behavioral factors is studied through correlation analysis, and the high-dimensional space of the original data is reduced to a low-dimensional space through principal component analysis, while retaining the main information of the data, so as to discover the factors that play a dominant role in the actor behavior.

[0028] Furthermore, in step S22, the optimization process must include a selection operation, a crossover operation, and a mutation operation;

[0029] The selection operation: selecting excellent gene sequences as parents and eliminating poor gene sequences according to the fitness function (a standard for evaluating the quality of gene sequences);

[0030] The crossover operation simulates the process of gene recombination, performs a crossover operation on the parent gene sequence, exchanges gene fragments, and generates a daughter gene sequence;

[0031] The mutation operation: performs random mutation operation on the offspring gene sequence, randomly changes some elements in the gene sequence, introduces new gene information, so as to increase the exploration ability of the search space, and helps to find the global optimal solution.

[0032] Furthermore, the optimization process may include an iterative operation, wherein the iterative operation includes: repeating selection, crossover and mutation operations to form a new generation of gene sequences, evaluating the quality of the new generation of gene sequences according to a fitness function, and retaining excellent individuals.

[0033] Furthermore, in step S3, the construction process of the quantitative strategy is as follows:

[0034] Model construction: Based on the results of factor analysis, a quantitative model is constructed to predict the trend of stock prices. The quantitative model includes but is not limited to the mean-variance model, arbitrage strategy model, and capital asset pricing model (CAPM);

[0035] Strategy design: Design strategies based on the model prediction results, determine key parameters such as the strategy’s holding ratio, position adjustment frequency, stop loss and take profit conditions, etc. The strategies include multi-factor strategies, momentum strategies, hedging strategies, etc.;

[0036] Risk management: Controlling risk by setting stop-loss points, adjusting positions, etc., helps ensure the stability and sustainability of the strategy;

[0037] The stop loss point is a point set by the actor when buying. When this point is reached or exceeded, the trading system will automatically execute a sell operation to prevent further losses. The stop loss point setting methods include based on a fixed amount, based on a fixed ratio, based on technical analysis, and based on a psychological price.

[0038] The said adjustment of positions: the actor controls risk exposure by adjusting positions, thereby reducing risks, and the position management methods include: funnel-type position management method (the initial position is small, and the position is gradually increased as the price falls, and the proportion of each increase increases), rectangular position management method (the initial position is fixed, and subsequent increases follow a fixed proportion), and pyramid-type position management method (the initial position is large, and then the position is increased in proportion).

[0039] Furthermore, in step S3, the preparation work before backtesting includes: clarifying the backtesting target, collecting historical data related to the strategy, and writing corresponding backtesting code according to the strategy;

[0040] The backtest implementation steps are as follows:

[0041] Set backtesting parameters: determine the time horizon for backtesting, which should be long enough to cover different market cycles, and set transaction costs to more accurately assess the actual returns of the strategy;

[0042] Run backtest: Run the written backtest code on historical data, simulate the actual trading process, and record the detailed information of each transaction;

[0043] Result output: After the backtest is completed, the backtest results are output, including risk-return indicators such as rate of return, volatility, maximum drawdown, and performance comparison of strategies in different market environments.

[0044] Furthermore, in step S4, the strategy optimization and adjustment specifically includes:

[0045] Parameter adjustment: Based on the backtest results, identify the weaknesses and improvement directions of the strategy, and adjust the key parameters of the strategy, such as the length of the moving average, the setting of the stop loss point, the proportion of position management, etc., and use algorithms such as grid search and particle swarm optimization to optimize the parameters to find the best parameter combination and improve the performance of the strategy;

[0046] Model fusion: combining different strategies or models through weighted average, voting method, etc. to reduce the risk of a single strategy and improve the robustness and profitability of the strategy;

[0047] Risk management optimization: Re-evaluate the risk level of the strategy to ensure that the strategy can obtain reasonable excess returns while taking certain risks, and adjust risk management measures according to market changes, such as setting more reasonable stop-loss points and using hedging tools.

[0048] The present invention provides a multi-dimensional factor quantification and digital gene recombination method based on behavior deconstruction. Each step is interrelated and mutually supportive, and together constitutes a complete and efficient quantitative analysis and optimization decision-making process, which has the following beneficial effects:

[0049] 1. Through behavioral deconstruction and multi-dimensional factor mining, it is possible to fully capture investment opportunities and risks in the market. Combined with processing operations such as factor quantification, it improves the accuracy and comparability of data, providing a solid foundation for subsequent factor synthesis and strategy construction.

[0050] 2. Screening and weighting of factors based on digital gene recombination can efficiently find the optimal factor combination, and has high flexibility, and can be dynamically adjusted and optimized according to market changes.

[0051] 3. Through backtesting and optimization, it can be ensured that the strategy has good performance and risk characteristics based on historical data, and at the same time can adapt to different market environments and investment styles, providing actors with stable investment returns.

[0052] 4. Combining modern computer technology and artificial intelligence technology, it realizes the intelligence and automation of strategies, reduces the operational difficulty and time cost of actors, and improves investment efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the steps of the method for quantifying multi-dimensional factors and digital gene recombination based on behavior deconstruction of the present invention;

[0054] Figure 2 It is a schematic diagram of the multi-dimensional factor quantification process based on the behavior deconstruction multi-dimensional factor quantification and digital gene recombination method of the present invention;

[0055] Figure 3 It is a schematic diagram of the digital gene recombination process of the present invention based on the multi-dimensional factor quantification and digital gene recombination method of behavior deconstruction. DETAILED DESCRIPTION

[0056] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0057] like Figure 1-Figure 3 As shown, the method of multi-dimensional factor quantification and digital gene recombination based on behavioral deconstruction includes the following steps:

[0058] S1. Multi-dimensional factor quantification:

[0059] S11. Factor selection and preliminary processing: Select multi-dimensional factors related to the behavior of actors from market data, including valuation factors, volatility factors, operating capacity factors, scale factors, technical factors, leverage factors, financial quality factors, and growth factors, etc., and perform quantitative processing on the selected factors after de-extreme value, standardization, and neutralization, and convert them into computable numerical indicators;

[0060] In this embodiment, the 3σ method is selected for extreme value removal to eliminate extreme values ​​in the data. The specific operations are as follows: the mean μ and standard deviation σ of the factor are calculated, and the threshold parameters are set. The factor values ​​that exceed the range [μ-3σ, μ+3σ] are adjusted, and the factor values ​​that exceed the range are replaced with boundary values ​​(i.e., μ-3σ or μ+3σ), or these values ​​are directly deleted;

[0061] Standardization Select Z-score standardization to convert the data into dimensionless form. The specific operation is as follows: calculate the mean μ and standard deviation σ of the factor, and use the formula (Z = (x-μ) / σ) to convert each factor value into a standard score. The standardized data will have a distribution with a mean of 0 and a standard deviation of 1.

[0062] Neutralization is based on regression analysis, eliminating the factor's preference for market value or industry to ensure the fairness of factor analysis. The specific operation is: collect the market value, industry classification and factor data of all stocks at each time point, and use the linear regression model to neutralize each factor by market value and industry at each time section, and then use the market value and industry data as independent variables and the factor to be removed as the dependent variable to perform cross-sectional regression, and extract the residuals from the regression results. These residuals are the factor values ​​after eliminating the market value and industry effects. The residual sequence is orthogonal to the independent variable sequence (market value and industry data), so the residual can be considered as the pure value of the factor after eliminating the market value and industry effects.

[0063] S12. Behavior deconstruction and factor analysis: Based on the principle of behavioral science, we deeply deconstruct the behavior of the actors, analyze the impact of different behavioral factors on decision-making, and conduct correlation analysis and principal component analysis on the selected factors to identify the relationship between key factors and potential factors; specifically, we study the mutual influence between different behavioral factors through correlation analysis, and reduce the high-dimensional space of the original data to a low-dimensional space through principal component analysis, while retaining the main information of the data, so as to find the factors that play a leading role in the behavior of the actors;

[0064] In this embodiment, the actor behavior factors include the actor's psychological expectations, risk preferences, and behavioral habits, as follows:

[0065] Psychological expectations: actors’ predictions and expectations of future market performance, economic prospects, policy changes, etc. Psychological expectations will affect actors’ decision-making process, forming different market expectations, thus leading to market price fluctuations;

[0066] Risk preference: the attitude and tolerance of actors towards risk, which can be divided into risk avoiders (prefer low risk and avoid high risk), risk seekers (willing to take high risk in pursuit of high return) and risk neutrals (neither avoid risk nor actively seek risk, the selection criteria are the size of expected return). Risk preference will affect the strategies and configurations of actors, thus affecting the market supply and demand relationship and price fluctuations;

[0067] Behavioral habits: stable behavioral patterns formed by actors in long-term practice;

[0068] S13. Multi-dimensional factor synthesis: Synthesize highly correlated factors to form composite factors. Synthesis can reduce redundant information between factors and improve the explanatory power and predictive ability of factors.

[0069] S2. Digital gene recombination:

[0070] S21. Gene encoding: Encode the digital information to be optimized into a gene sequence, that is, convert the complex factors into binary or other forms of encoding;

[0071] S22, Genetic optimization: genetic operations and recombination are performed on the encoded gene sequence to find the optimal solution, that is, the optimal factor combination;

[0072] In this embodiment, the optimization process must include selection operations, crossover operations, and mutation operations;

[0073] Selection operation: select excellent gene sequences as parents according to the fitness function (the standard for evaluating the quality of gene sequences) and eliminate poor gene sequences;

[0074] Crossover operation: simulates the process of gene recombination, performs crossover operation on the parent gene sequence, exchanges gene fragments, and generates the offspring gene sequence;

[0075] Mutation operation: Random mutation operation is performed on the offspring gene sequence, which randomly changes some elements in the gene sequence and introduces new gene information to increase the exploration ability of the search space and help find the global optimal solution;

[0076] In addition, the optimization process may include iterative operations: repeated selection, crossover and mutation operations to form a new generation of gene sequences, and evaluate the quality of the new generation of gene sequences according to the fitness function, and retain excellent individuals;

[0077] S23, gene decoding: after the optimization process is completed, the gene sequence is decoded into original digital information, that is, the encoded gene sequence is converted back to the original digital form for subsequent analysis and application;

[0078] S3. Strategy construction and backtesting:

[0079] Construct quantitative strategies based on the optimal factor portfolio and backtest the strategies to evaluate their historical performance and risk characteristics;

[0080] In this embodiment, the construction process of the quantitative strategy is as follows:

[0081] Model construction: Based on the results of factor analysis, a quantitative model is constructed to predict the trend of stock prices. The quantitative model includes but is not limited to the mean-variance model, arbitrage strategy model, and capital asset pricing model (CAPM);

[0082] Strategy design: Design strategies based on the model prediction results, determine key parameters such as the strategy's holding ratio, position adjustment frequency, stop loss and take profit conditions, etc. Strategies include multi-factor strategies, momentum strategies, hedging strategies, etc.

[0083] Risk management: Controlling risk by setting stop-loss points, adjusting positions, etc., helps ensure the stability and sustainability of the strategy;

[0084] Set a stop loss point: a point set by the actor when buying. When this point is reached or exceeded, the trading system will automatically execute a sell operation to prevent further losses. The stop loss point can be set based on a fixed amount, a fixed ratio, technical analysis, or a psychological price.

[0085] Adjusting positions: The actors control risk exposure by adjusting positions, thereby reducing risks. Position management methods include: funnel position management method (the initial position is small, and the position is gradually increased as the price falls, and the proportion of each increase increases), rectangular position management method (the initial position is fixed, and subsequent increases follow a fixed proportion), and pyramid position management method (the initial position is large, and then the position is increased in proportion);

[0086] In this embodiment, the preparation work before backtesting includes: clarifying the backtesting target, collecting historical data related to the strategy, and writing corresponding backtesting code according to the strategy; and the backtesting implementation steps are as follows:

[0087] Set backtesting parameters: determine the time horizon for backtesting, which should be long enough to cover different market cycles, and set transaction costs to more accurately assess the actual returns of the strategy;

[0088] Run backtest: Run the written backtest code on historical data, simulate the actual trading process, and record the detailed information of each transaction;

[0089] Result output: After the backtest is completed, the backtest results are output, including risk-return indicators such as rate of return, volatility, maximum drawdown, and performance comparison of strategies in different market environments;

[0090] S4. Strategy optimization and real-time trading: Optimize and adjust the strategy according to the backtest results, apply the optimized strategy to real-time trading, and dynamically adjust it according to market changes; in this embodiment, strategy optimization and adjustment specifically include:

[0091] Parameter adjustment: Based on the backtest results, identify the weaknesses and improvement directions of the strategy, and adjust the key parameters of the strategy, such as the length of the moving average, the setting of the stop loss point, the proportion of position management, etc., and use algorithms such as grid search and particle swarm optimization to optimize the parameters to find the best parameter combination and improve the performance of the strategy;

[0092] Model fusion: combining different strategies or models through weighted average, voting method, etc. to reduce the risk of a single strategy and improve the robustness and profitability of the strategy;

[0093] Risk management optimization: Re-evaluate the risk level of the strategy to ensure that the strategy can obtain reasonable excess returns while taking certain risks, and adjust risk management measures according to market changes, such as setting more reasonable stop-loss points and using hedging tools.

[0094] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. Based on behavioral deconstruction, multi-dimensional factor quantification and digital gene recombination method, characterized by: The following steps are involved: S1. Multi-dimensional factor quantification: S11. Factor selection and preliminary processing: Select multi-dimensional factors related to the behavior of actors from market data, and after de-extreme value, standardization and neutralization of the selected factors, perform quantitative processing and convert them into computable numerical indicators; S12. Behavior deconstruction and factor analysis: Based on behavioral principles, deeply deconstruct the behavior of actors, analyze the impact of different behavioral factors on decision-making, and conduct correlation analysis and principal component analysis on the selected factors to identify the relationship between key factors and potential factors; S13. Multi-dimensional factor synthesis: synthesize highly correlated factors to form composite factors; S2. Digital gene recombination: Drawing on the idea of ​​gene recombination technology, further optimize and combine the synthesized composite factors; S3. Strategy construction and backtesting: Construct quantitative strategies based on the optimal factor combination, and backtest the strategies to evaluate their historical performance and risk characteristics; S4. Strategy optimization and real-time trading: Optimize and adjust the strategy based on the backtesting results, apply the optimized strategy to real-time trading, and make dynamic adjustments based on market changes.

2. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1 is characterized in that: In step S11, de-extreme value, standardization and neutralization are used to eliminate the correlation and scale effect between factors, as follows: Remove extreme values: Use the 3σ method, percentile method or MAD method to eliminate extreme values ​​in the data due to data entry errors, abnormal events or other reasons; Standardization: Use Z-score standardization or Min-Max standardization to convert the data into dimensionless form for comparison and regression analysis; Neutralization: Based on regression analysis, eliminate the factor's preference for market capitalization or industry to ensure the fairness of factor analysis.

3. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1 is characterized in that: In step S12, the actor's behavior factors include the actor's psychological expectations, risk preferences, and behavior habits, as follows: Psychological expectations: actors’ predictions and expectations of future market performance, economic prospects, and policy changes. Psychological expectations will affect actors’ decision-making process, forming different market expectations, thus leading to market price fluctuations. Risk preference: the attitude and tolerance of the actor towards risk, which can be divided into risk-averse, risk-seeking and risk-neutral. Risk preference will affect the actor's strategy and allocation, thereby affecting the market supply and demand relationship and price fluctuations; Behavioral habits: stable behavioral patterns formed by actors through long-term practice.

4. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 3 is characterized in that: In step S12, in the actor behavior analysis, the mutual influence between different behavioral factors is studied through correlation analysis, and the high-dimensional space of the original data is reduced to a low-dimensional space through principal component analysis, while retaining the main information of the data, so as to discover the factors that play a dominant role in the actor behavior.

5. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1 is characterized in that: The digital gene recombination specifically includes the following sub-steps: S21. Gene encoding: Encode the digital information to be optimized into a gene sequence, that is, convert the complex factors into binary or other forms of encoding; S22, Genetic optimization: genetic operations and recombination are performed on the encoded gene sequence to find the optimal solution, that is, the optimal factor combination; S23. Gene decoding: After the optimization process is completed, the gene sequence is decoded into the original digital information, that is, the encoded gene sequence is converted back to the original digital form.

6. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 5 is characterized in that: In step S22, the optimization process must include selection operation, crossover operation and mutation operation; The selection operation: selecting excellent gene sequences as parents according to the fitness function and eliminating poor gene sequences; The crossover operation simulates the process of gene recombination, performs a crossover operation on the parent gene sequence, exchanges gene fragments, and generates a daughter gene sequence; The mutation operation: performs random mutation operation on the offspring gene sequence, randomly changes some elements in the gene sequence, introduces new gene information, so as to increase the exploration ability of the search space, and helps to find the global optimal solution.

7. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 6 is characterized in that: The optimization process may include iterative operations, wherein the iterative operations include: repeated selection, crossover and mutation operations to form a new generation of gene sequences, and evaluate the quality of the new generation of gene sequences according to the fitness function, and retain excellent individuals.

8. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1 is characterized in that: In step S3, the construction process of the quantitative strategy is as follows: Model construction: Based on the results of factor analysis, a quantitative model is constructed to predict the trend of stock prices. The quantitative model includes but is not limited to the mean-variance model, arbitrage strategy model, and capital asset pricing model; Strategy design: Design strategies based on the model prediction results and determine the key parameters of the strategies, including multi-factor strategies, momentum strategies, and hedging strategies; Risk management: Control risks by setting stop-loss points and adjusting positions; The stop loss point is a point set by the actor when buying. When this point is reached or exceeded, the trading system will automatically execute a sell operation to prevent further losses. The stop loss point setting methods include based on a fixed amount, based on a fixed ratio, based on technical analysis, and based on a psychological price. The position adjustment described above: the actor controls risk exposure by adjusting positions, thereby reducing risks, and the position management methods include: funnel-type position management method, rectangular position management method, and pyramid-type position management method.

9. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1, characterized in that: In step S3, the preparation work before backtesting includes: clarifying the backtesting target, collecting historical data related to the strategy, and writing corresponding backtesting code according to the strategy; The backtest implementation steps are as follows: Set backtesting parameters: determine the time frame for backtesting, which should be long enough to cover different market cycles, and set transaction costs; Run backtest: Run the written backtest code on historical data, simulate the actual trading process, and record the detailed information of each transaction; Result output: After the backtest is completed, the backtest results are output, including risk-return indicators such as rate of return, volatility, and maximum drawdown, as well as a comparison of the performance of strategies in different market environments.

10. The method for multi-dimensional factor quantification and digital gene recombination based on behavior deconstruction according to claim 1, characterized in that: In step S4, the strategy optimization and adjustment specifically includes: Parameter adjustment: Based on the backtesting results, identify the weaknesses and improvement directions of the strategy, and adjust the key parameters in the strategy. Use grid search and particle swarm optimization algorithms to optimize the parameters to find the best parameter combination; Model fusion: combining different strategies or models through weighted averaging and voting; Risk management optimization: Re-evaluate the risk level of the strategy to ensure that the strategy can obtain reasonable excess returns while taking certain risks, and adjust risk management measures according to market changes.

Images