Quantitative transaction data analysis method and system based on carbon value

Abstract

The invention discloses a quantitative transaction data analysis method and system based on a carbon value, and relates to the field of data analysis, and the method comprises the steps: obtaining a carbon market transaction data set, a macroeconomic index and a carbon transaction policy index; performing statistical analysis on the price return sequence of each carbon market based on the carbon market transaction data set, and estimating a corresponding marginal distribution function through a kernel density estimation method; based on a Bayesian method, fusing the macroeconomic index and the carbon transaction policy index to determine an estimation Copula parameter corresponding to each marginal distribution function; modeling a time-varying characteristic of each estimated Copula parameter based on a GARCH model to determine a corresponding time-varying Copula parameter, and further calculating a CoVaR value; and processing the target carbon transaction time sequence feature and the CoVaR value based on a price prediction model to determine a carbon quota predicted price of the target carbon market. Therefore, the mutual influence between different carbon markets is comprehensively considered, the carbon price prediction precision is improved, and the stability of a quantitative transaction strategy is enhanced.

Description

Technical Field

[0001] This application relates to the technical field of data analysis, and in particular, to a method and system for quantitative trading data analysis based on carbon values. Background Art

[0002] With the increasing severity of global climate change problems, the carbon trading market has become one of the core tools. Since the implementation of the carbon emission trading system, many countries and regions have begun to establish carbon markets to encourage enterprises to reduce greenhouse gas emissions and promote the realization of carbon emission reduction goals. The carbon trading market sets carbon emission quotas and allows enterprises and countries to trade carbon emission rights, thereby incentivizing emission reduction behaviors.

[0003] However, the carbon market has high volatility and uncertainty, which are jointly affected by factors such as policy changes, macroeconomic fluctuations, and global climate change. The price of the carbon market is not only affected by factors such as international energy prices, economic growth rates, and environmental regulations, but is also strongly driven by the behaviors of market participants and policy adjustments. For example, carbon prices may fluctuate violently due to sudden political decisions or climate events, making it more difficult to predict the price trend of the carbon market.

[0004] Secondly, most existing carbon market prediction methods rely on historical data or price models of a single market, and often ignore the non-linear correlations between markets, especially the risk spillover effects across markets. For example, price fluctuations in the EU carbon market may affect other regional carbon markets through external factors such as the international energy market and global economic recessions, thereby triggering price fluctuations between markets.

[0005] In response to the above problems, the industry has not yet proposed a better technical solution. Summary of the Invention

[0006] This application provides a method, system, storage medium, computer program product, and electronic device for quantitative trading data analysis based on carbon values, which are used to at least solve the problems in the current related technologies that the carbon market price prediction technology ignores the non-linear correlations between markets, cannot better cope with market changes, resulting in carbon market price prediction deviations and low accuracy of carbon market quantitative trading operations.

[0007] In a first aspect, an embodiment of the present application provides a method for analyzing quantitative trading data based on carbon values, including: obtaining a carbon market trading data set, macroeconomic indicators, and carbon trading policy indicators; the carbon market trading data set includes target carbon market trading data and multiple associated carbon market trading data; based on the carbon market trading data set, statistically analyze the price return sequences of each carbon market, and estimate the corresponding marginal distribution function by the kernel density estimation method; for each of the carbon markets, fuse the macroeconomic indicators and the carbon trading policy indicators based on the Bayesian method to determine the estimated Copula parameters corresponding to each of the marginal distribution functions; the estimated Copula parameters are used to measure the linear dependence between different marginal distribution functions under macroeconomic indicators and / or carbon trading policy indicators; model the time-varying characteristics of each of the estimated Copula parameters based on the GARCH model to determine the corresponding time-varying Copula parameters, and calculate the CoVaR value according to each of the time-varying Copula parameters; the CoVaR value is used to measure the risk spillover effect of the target carbon market when extreme events occur in each of the associated carbon markets; extract the target carbon trading time series characteristics corresponding to the target carbon market trading data, and process the target carbon trading time series characteristics and the CoVaR value based on a price prediction model to determine the carbon quota prediction price of the target carbon market; wherein, the carbon quota prediction price is transmitted to a quantitative trading decision engine to assist in determining the corresponding quantitative trading strategy.

[0008] Second aspect, an embodiment of the present application provides a carbon-value-based quantitative trading data analysis system, including: an information acquisition unit, configured to acquire a carbon market trading data set, macroeconomic indicators, and carbon trading policy indicators; the carbon market trading data set includes target carbon market trading data and multiple associated carbon market trading data; a distribution analysis unit, configured to perform statistical analysis on the price return sequences of each carbon market based on the carbon market trading data set, and estimate the corresponding marginal distribution function by using a kernel density estimation method; a Copula calculation unit, configured to, for each of the carbon markets, fuse the macroeconomic indicators and the carbon trading policy indicators based on a Bayesian method to determine the estimated Copula parameters corresponding to each of the marginal distribution functions; the estimated Copula parameters are used to measure the linear dependence between different marginal distribution functions under the macroeconomic indicators and / or carbon trading policy indicators; a CoVaR calculation unit, configured to model the time-varying characteristics of each of the estimated Copula parameters based on a GARCH model to determine the corresponding time-varying Copula parameters, and calculate a CoVaR value according to each of the time-varying Copula parameters; the CoVaR value is used to measure the risk spillover effect of the target carbon market when extreme events occur in each of the associated carbon markets; a price prediction unit, configured to extract the target carbon trading time series features corresponding to the target carbon market trading data, and process the target carbon trading time series features and the CoVaR value based on a price prediction model to determine the predicted carbon quota price of the target carbon market; wherein, the predicted carbon quota price is transmitted to a quantitative trading decision engine to assist in determining the corresponding quantitative trading strategy.

[0009] Third aspect, an electronic device is provided, which includes: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute the steps of the carbon-value-based quantitative trading data analysis method according to any embodiment of the present application.

[0010] Fourth aspect, an embodiment of the present application provides a storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, the steps of the carbon-value-based quantitative trading data analysis method according to any embodiment of the present application are implemented.

[0011] Fifth aspect, an embodiment of the present application provides a computer program product, including a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the carbon-value-based quantitative trading data analysis method according to any embodiment of the present application are implemented.

[0012] By means of the carbon-value-based quantitative trading data analysis method and system provided by the present application, the following technical effects can be at least achieved:

[0013] (1) By comprehensively integrating multiple data sources (such as carbon market trading data, macroeconomic indicators, and carbon trading policy indicators) and adopting a Copula model based on the Bayesian method to accurately model the non-linear correlation between different carbon markets, it is possible to more comprehensively consider the mutual influence between each market. Under the influence of external factors such as global climate change and policy adjustments, it is possible to more accurately capture the price fluctuation characteristics between different markets. Thus, it can effectively improve the accuracy of carbon market price prediction, reduce the prediction bias caused by traditional prediction methods ignoring cross-market dependence, improve the accuracy rate of quantitative trading operations in the carbon market, and thereby guarantee the rate of return of quantitative trading.

[0014] (2) By conducting statistical analysis on the return series of carbon market prices and using the Copula model to capture the dependence relationship between different markets, it is possible to quantify the risk spillover effect between different carbon markets under the background of macroeconomic changes and carbon trading policy changes. Adopting the GARCH model to model the time-varying characteristics of the estimated Copula parameters can capture the dynamic changes of carbon market price fluctuations and their risk spillover effects over time. Through time-varying modeling, the scheme can timely adjust the prediction results and risk assessment as the market environment changes. In particular, when there are macroeconomic fluctuations or policy adjustments, it can provide timely and accurate CoVaR value assessment, reflecting the potential risks of the target carbon market when specific extreme events occur.

[0015] (3) By analyzing the time series characteristics of the trading data of the target carbon market and combining with the CoVaR value, accurately grasping the trend of carbon emission prices enables the quantitative trading decision-making engine to more precisely identify market trends and risks and optimize trading strategies. In a complex carbon trading market environment, it can provide stronger risk control and return optimization capabilities for the quantitative trading system, thereby enhancing the scientific nature and stability of trading decisions.

[0016] Through this technical solution, considering the transnational and regional nature of the carbon market, incorporating the trading data of different carbon markets and the external economic environment into the analysis framework, and combining advanced statistical analysis methods and dynamic time-varying modeling techniques, it improves the accuracy of multi-dimensional collaborative price prediction and risk management capabilities in the carbon market, enhances the stability of quantitative trading strategies in the face of volatility and uncertainty, and provides valuable decision-making support for participants in the carbon emission trading market. Description of the Drawings

[0017] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 Fig. shows a flowchart of an example of a carbon-value-based quantitative trading data analysis method according to an embodiment of the present application;

[0019] Figure 2 Fig. shows according to Figure 1 an example of an operation flowchart of step S120 in

[0020] Figure 3 Fig. shows an operation flowchart of an example of determining an estimated Copula parameter corresponding to a marginal distribution function according to an embodiment of the present application;

[0021] Figure 4 Fig. shows according to Figure 1 an example of an operation flowchart of step S150 in

[0022] Figure 5 Fig. shows a schematic structural connection diagram of an example of a price prediction model according to an embodiment of the present application;

[0023] Figure 6 Fig. shows a schematic block diagram of an example of a carbon-value-based quantitative trading data analysis system according to an embodiment of the present application;

[0024] Figure 7 Fig. is a schematic structural diagram of an embodiment of an electronic device of the present application. Detailed Embodiments

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0026] In the technical solutions of the present application, for the processing of the collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved, etc., they all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0027] It should be noted that quantitative trading is an investment method based on mathematical models or statistical analysis algorithms, which automates trading decisions and executions through computer programs. The basic idea of quantitative trading is to utilize a large amount of historical and real-time data to design algorithmic models that can effectively predict market trends, thereby automating buy and sell decisions. Quantitative trading has been widely applied in financial markets. Especially in traditional financial markets such as stocks and futures, quantitative trading systems can react in real time based on data such as price fluctuations, market depth, and trading volume.

[0028] Although quantitative trading has achieved great success in financial markets, its application in the carbon trading market is still relatively preliminary. The carbon trading market is significantly different from traditional financial markets. Its main characteristics include that market prices are affected by multiple factors such as policies, macroeconomic fluctuations, and international situations, making it more difficult to control the prices of carbon emission allowances.

[0029] In view of this, Figure 1 The flowchart of an example of the carbon-value-based quantitative trading data analysis method according to an embodiment of the present application is shown.

[0030] Regarding the execution subject of the method according to the embodiment of the present application, it can be any controller or processor with computing or processing capabilities. Specifically, it can be implemented by a carbon market quantitative trading platform, which is used to automatically implement or assist decision-makers in implementing trading actions for carbon shares. Through the fusion analysis of multi-dimensional data, risks in the carbon market are systematically identified and evaluated from multiple perspectives, including not only price fluctuations but also external factors such as the correlation between markets and the impact of policy changes, and cross-market risk spillover analysis is carried out, improving the accuracy of carbon market price prediction, enabling better guidance of enterprises' behaviors in the carbon trading market, and achieving more efficient carbon emission reduction goals and carbon quota allocation.

[0031] In some examples, it can be integrated and configured in an electronic device or terminal in a software, hardware, or software-hardware combination manner, and the types of terminals or electronic devices can be diverse, such as mobile phones, tablets, or desktop computers, etc.

[0032] As Figure 1 As shown, in step S110, a carbon market trading data set, macroeconomic indicators, and carbon trading policy indicators are obtained.

[0033] Here, the carbon market trading data set includes target carbon market trading data (e.g., the Chinese carbon market) and multiple associated carbon market trading data (e.g., the EU carbon market, the US carbon market, etc.), which can be called through the authorized API interfaces of each carbon market. The data content of the carbon market trading data can be diversified. For example, the trading price of carbon emission allowances, trading volume, historical price fluctuations, etc. It should be noted that the obtained associated carbon market data is not limited to the international market and can also include trading data of cross-regional markets (e.g., carbon markets in different industries or different provinces) to capture the interactions between markets.

[0034] Through macroeconomic indicators, it is helpful to identify the macroeconomic situation of the target carbon market and is used to analyze the fluctuations of market prices. These can be obtained through platforms such as search engines or the official websites of statistical bureaus. These economic data play an important role in the price fluctuations of the carbon market. The types of macroeconomic indicators can also be diversified. For example, real-time prices of energy commodities (crude oil, natural gas, etc.), GDP growth rate, CPI (Consumer Price Index), PPI (Producer Price Index), interest rates, etc.

[0035] The carbon trading policy indicators can reflect the carbon market policy orientation of the country or region where the enterprise is located. Specifically, the carbon trading policy indicators can include carbon emission policies, carbon tax policies, carbon emission reduction target information, carbon pricing policies, carbon quota allocation rules, emission limit adjustments, etc. These policy information can affect the supply-demand balance and price fluctuations of carbon emission trading in the market to varying degrees. In terms of data source channels, it can be obtained through channels such as reports issued by the government, climate change policy research institutions, or international carbon market regulatory agencies.

[0036] Thus, multi-dimensional data input is provided, covering carbon market trading data, macroeconomic indicators, and policy environment. The integration of multi-source data provides a multi-angle perspective for subsequent analysis and modeling, ensuring a comprehensive analysis foundation.

[0037] In step S120, based on the carbon market trading data set, statistical analysis is performed on the price return sequences of each carbon market, and the corresponding marginal distribution function is estimated by the kernel density estimation method.

[0038] Specifically, first, perform return series analysis on the historical trading data of each carbon market, calculate price return series (such as daily return, weekly return), etc., and evaluate the characteristics of the data using statistical methods such as mean, variance, skewness, kurtosis, etc. Then, perform kernel density estimation (KDE) on the price returns of each carbon market. By selecting an appropriate kernel function (such as the Gaussian kernel function), the probability distribution of the carbon market price can be estimated, thereby obtaining a non-parametric marginal distribution function and realizing the distribution estimation of the price return series of the carbon market.

[0039] It should be noted that the core idea of kernel density estimation is to estimate the probability density function of the population through the weighted average of sample points. Through kernel density estimation, the errors brought by traditional distribution assumptions (such as normal distribution) can be effectively avoided, providing a more flexible market price distribution model. The marginal distribution function can more accurately describe the actual distribution characteristics of carbon market price returns, avoiding the biases that may be caused by using a fixed distribution model.

[0040] In step S130, for each carbon market, fuse macroeconomic indicators and carbon trading policy indicators based on the Bayesian method to determine the estimated Copula parameters corresponding to each marginal distribution function.

[0041] Specifically, use the Bayesian inference method to fuse macroeconomic indicators (such as GDP growth rate, real-time price of energy commodities, etc.) and carbon trading policy indicators (such as carbon emission policy, carbon tax policy, etc.) into the model of the marginal distribution function. Through the Bayesian framework, a prior distribution can be established, and the posterior distribution can be continuously updated in combination with actual data, and finally the optimal Copula parameters of the price return series of each carbon market can be obtained. In this way, by constructing a Bayesian network and establishing the parameter dependence structure of the model, the macroeconomic and policy indicators can be reasonably reflected in the dependence relationship of carbon market prices.

[0042] Here, the estimated Copula parameters are used to measure the linear dependence between different marginal distribution functions under macroeconomic indicators and / or carbon trading policy indicators. Calculate the dependence between the marginal distribution functions of each carbon market through the Bayesian method, that is, estimate the Copula parameters, which can accurately capture the non-linear dependence relationship between different carbon markets, and Gaussian Copula function, t-Copula function, etc. can be selected to match the characteristics of carbon market data.

[0043] By introducing the Bayesian method, the model can be dynamically adjusted according to new economic or policy data, and the Copula model is used to measure the non-linear correlation and dependence between markets, improving the adaptability to market changes.

[0044] In step S140, the time-varying characteristics of each estimated Copula parameter are modeled based on the GARCH model to determine the corresponding time-varying Copula parameter, and the CoVaR value is calculated according to each time-varying Copula parameter.

[0045] Here, the GARCH (Generalized Autoregressive Conditional Heteroskedasticity) model is used to perform time-varying modeling on the Copula parameter. The CoVaR value (or, conditional VaR value) is used to measure the risk spillover effect of the target carbon market when extreme events occur in each associated carbon market.

[0046] It should be noted that the GARCH model can handle the heteroskedasticity commonly found in financial markets, that is, the price volatility changes over time. Specifically, when extreme events occur in other associated markets, the target market may be affected by risks. By using the GARCH model to model the Copula parameters (such as correlation and dependence parameters) of each carbon market, time-varying Copula parameters that change over time are obtained. Then, the CoVaR value is calculated by integrating the time-varying Copula parameters of each associated carbon market, which reflects the risk spillover effect of each associated carbon market on the target carbon market and is an effective measure of systemic risk.

[0047] Thus, by introducing the GARCH model, it is possible to track the changes in market volatility in real time, capture the potential volatility clustering effect in the carbon market, and provide dynamic support for risk prediction. In addition, through the CoVaR value, the risk transmission effect of extreme events (such as sudden political decisions, climate events, energy price fluctuations, etc.) in different associated markets on the target market can be accurately measured, revealing the spillover effect of the associated market on the price of the target market.

[0048] In step S150, the target carbon trading time series characteristics corresponding to the target carbon market trading data are extracted, and the target carbon trading time series characteristics and the CoVaR value are processed based on the price prediction model to determine the predicted carbon quota price of the target carbon market.

[0049] In some embodiments, time series analysis is performed on the trading data of the target carbon market to extract characteristics such as trends, seasonality, and volatility, and technical analysis methods (such as moving average, ARIMA model) are used to analyze the historical trends and short-term trends of the market. Then, based on the price prediction model, machine learning methods (such as support vector machines, deep neural networks, etc.) are used to combine the time series characteristics of the target carbon market with the CoVaR value to predict the carbon quota price at a future time (such as tomorrow or next week).

[0050] Furthermore, the predicted carbon quota price is transmitted to the quantitative trading decision-making engine to assist in determining the corresponding quantitative trading strategy. Specifically, the quantitative trading engine can generate an optimal quantitative trading strategy by configuring a series of algorithms and models, such as price momentum-based strategies, arbitrage strategies, and hedging strategies, etc., so that the quantitative trading strategy can be adjusted according to the accurately predicted price trend of the carbon market, thereby effectively improving the trading decision-making efficiency and profit potential of the carbon market.

[0051] Through the embodiments of the present application, by using step-by-step statistical analysis and modeling methods, the carbon market trading data, macroeconomic indicators, and policy impacts are precisely integrated. Through the analysis of multi-dimensional data, the modeling of non-linear correlations, the time-varying risk assessment, and the application of prediction models, the prediction ability and risk management ability of the carbon market are comprehensively improved, providing a solid theoretical support and technical guarantee for the accurate quantitative trading decision-making of the carbon market.

[0052] Figure 2 shows an operation flow chart according to Figure 1 an example of step S120 in

[0053] As Figure 2 shown, in step S210, the logarithmic return of the carbon quota price in the carbon market at each time point is calculated to determine the corresponding price return sequence of the carbon market.

[0054] Specifically, by using logarithmic return for price volatility measurement, the influence of price level is eliminated, and it has the property of additivity, which can better analyze the carbon market price.

[0055]

[0056] In the formula, P i,t and P i,t-1 respectively represent the carbon quota trading prices of carbon market i at time t and time t - 1; r i,t is the price return of carbon market i at time t, representing the change rate relative to the price of the previous moment.

[0057] Thus, the price change rate of the carbon market within any time interval can be calculated to reflect the market volatility. In addition, after obtaining the return sequence, statistical analysis can also be carried out to better understand the market volatility pattern and its change characteristics. For example, statistical indicators such as the mean, variance, and skewness of the return sequence are calculated to identify the long-term trend and short-term volatility risk of the market.

[0058] In step S220, for the price return sequence of each carbon market, the corresponding marginal distribution function is obtained using the kernel density estimation method.

[0059] It should be noted that traditional assumed distributions (such as the normal distribution) may not be applicable to the carbon market because the price return series in the carbon market usually has the characteristics of a sharp peak and thick tails. Therefore, the kernel density estimation method is used to non-parametrically estimate the marginal distribution of each carbon market price return.

[0060] The kernel density estimation method does not rely on any prior assumptions and can directly capture the non-normal distribution characteristics such as sharp peaks and thick tails of price returns based on the actual transaction data of the carbon market. This is particularly important for a market like the carbon market that is driven by policies and external factors.

[0061]

[0062] In the formula, f i (r) is the marginal distribution function of carbon market i, which represents the density at the price return r; h is the bandwidth parameter used to control the width of the kernel function; T represents the time length of the price return series, and K(·) represents the Gaussian kernel function.

[0063] Kernel density estimation obtains a smooth estimate of the data by weighting the contributions of multiple data points. In the context of the carbon market, this method can be used to effectively estimate the marginal distribution function f i (r) of each carbon market price return. This distribution function reflects the overall characteristics of the market price return, including its central tendency and volatility characteristics. Thereby, it provides the probability density information of the price return, which can help quantify the range of price return changes (such as in which intervals most returns are concentrated), thus providing a basis for risk management.

[0064] In addition, the marginal distribution obtained by kernel density estimation can accurately capture the price return relationship between the target carbon market and the associated carbon market. It is the key basis for constructing the joint distribution among multiple carbon markets, can accurately reflect the price volatility characteristics of a single market, and thus improve the modeling accuracy of the joint distribution.

[0065] Figure 3 The operation flowchart showing an example of determining the estimated Copula parameter corresponding to the marginal distribution function according to an embodiment of the present application is shown.

[0066] As Figure 3 shown, in step S310, by the Bayesian method, external factors are incorporated into the estimation of the Gaussian Copula correlation parameter to determine the posterior distribution.

[0067] Specifically, the Gaussian Copula function is used to describe the linear correlation between multivariate random variables, and its joint distribution function is defined as:

[0068] C(u 1 ,u 2 ; ρ t ) = Φρ (Φ -1 (u 1 ), Φ -1 (u 2 )) , Equation (3)

[0069] u 1 = f i (r i,t ), u 2 = f j (r j,t ) , Equation (4)

[0070] Wherein, u 1 and u 2 respectively represent the marginal distribution function values of the target carbon market i and the associated carbon market j at time t, C(u 1 , u 2 ; ρ t ) represents the Gaussian Copula joint distribution function, describing the joint distribution between the target market and the associated market; ρ t is the dynamic correlation parameter of the Gaussian Copula, representing the linear correlation between the target carbon market i and the associated carbon market j at time t; Φ ρ is the cumulative distribution function of the two-dimensional normal distribution, and Φ -1 is the inverse cumulative distribution function of the standard normal distribution.

[0071] By estimating the Gaussian Copula joint distribution function C(u 1 , u 2 ; ρ t ) and the dynamic correlation parameter ρ t , a dynamic correlation model can be constructed between the target carbon market and multiple associated carbon markets. Specifically, the correlation parameter ρ t between the target market i and the associated market j at different times t dynamically characterizes the strength of the linkage of price fluctuations between the markets. In addition, when an extreme event (such as a carbon tax policy adjustment) occurs in the associated carbon market, the dynamic correlation parameter ρ t can reflect the impact intensity of this event on the target carbon market in real time.

[0072]

[0073] Wherein, z t represents the external influencing factors at time t, including macroeconomic indicators and carbon trading policy indicators; r i,t and r j,t respectively represent the price returns of the target carbon market i and the associated carbon market j at time t; p(ρ t |r i,t , r j,t , z t) is the posterior distribution, representing the probability distribution of ρ after observing the price return sequence r i,t , r j,t and the external factor z t ; p(r t , r i,t | ρ j,t ) represents the joint distribution probability of the price return sequences r t and r i,t given ρ j,t ; p(z t | ρ t ) represents the conditional probability distribution of the external factor z t given ρ t ; p(ρ t ) represents the prior Gaussian distribution, which is the initial assumption about ρ t ; p(r t , r i,t , r j,t , z t ) is the normalization constant to ensure that the integral of the posterior distribution is 1.

[0074] Through the estimation of the posterior distribution p(ρ t | r i,t , r j,t , z t ), and the adaptive adjustment of the dynamic correlation parameter ρ t , the risk transfer effect between markets can be quantified more accurately. Using the value of ρ t estimated by the posterior distribution, when an extreme event (such as a price slump) occurs in a given market, the price risk spillover effect of the target market can be calculated through the Copula function. Through the dynamic Gaussian Copula, the complex risk transfer network between multiple markets can be effectively captured, providing a quantitative analysis tool for carbon market risk management.

[0075] In step S320, the influence of external factors on the Gaussian Copula parameters is represented by linear weighting to determine the conditional distribution of external factors.

[0076]

[0077] In the formula, λ is the weight coefficient of z t on the correlation parameter ρ t , which is used to quantify the influence intensity of external factors on market correlation; is the variance of the external factor, representing the uncertainty of the influence of external factors on correlation.

[0078] Specifically, the Bayesian method is used to estimate the posterior distribution, combining external factors and historical market data to dynamically adjust the correlation parameters of the Gaussian Copula. The Bayesian method can adapt to changes in the external environment through the integration of prior distribution and data. In addition, the form of the conditional distribution is jointly determined by the linear influence of external factors on the correlation parameters and random noise (the variance of external factors ), ensuring that the model can balance the roles of external information and market data.

[0079] By introducing external factors (such as energy prices, GDP growth rates, carbon quota policy adjustments) as part of the conditional probability p(z t |ρ t ), the correlation parameter ρ t of the Gaussian Copula is dynamically corrected. Specifically, the conditional distribution p(z t |ρ t ) of the external factor z t can clearly reflect the driving intensity of each external factor on the change of market correlation. For example, when the energy price z t rises, the weight coefficient λ enhances the correlation between the target market and the associated market, reflecting the economic linkage between energy commodity prices and carbon market prices, and can quickly adapt to changes in the macroeconomic and policy environment.

[0080] In step S330, the log-likelihood function of the Gaussian Copula is constructed.

[0081]

[0082] In the formula, represents the log-likelihood function of the Gaussian Copula, which is used for the estimation of the dynamic correlation parameter ρ t ; is a correction factor to ensure the stability of the log-likelihood.

[0083] Here, constructing the log-likelihood function of the Gaussian Copula not only considers the marginal distribution values of the target market and the associated market, but also integrates the dynamic correlation intensity between markets, can quantify the consistency between the dynamic correlation parameter and the joint distribution, and avoids the overflow problem in numerical calculations through the correction factor , ensuring the computational efficiency and robustness of the model. The log-likelihood function combines the complex dynamic correlation relationships between multiple markets and can effectively optimize the dynamic correlation parameter.

[0084] In step S340, according to the Bayesian posterior distribution, the conditional distribution weight of the external factor is introduced to maximize the weighted log-likelihood function to calculate the estimated Copula parameter corresponding to the marginal distribution function.

[0085]

[0086] In the formula, represents the finally determined estimated Copula parameter at time t; represents the maximization operator used to optimize ρ t ; lnp(z t |ρ t ) represents the weighted correction term of the conditional distribution of external factors with respect to the log-likelihood.

[0087] Specifically, the logarithmic form of the conditional distribution of external factors is introduced into the objective function to quantify the adjustment effect of external factors on dynamic correlation. The maximization process of the objective function comprehensively considers the multi-faceted impacts of historical data, joint distribution, and external factors on dynamic correlation. After optimization, the dynamic correlation parameters at different times are obtained, which can reflect the correlation strength between the target market and multiple associated markets at different times, enabling the correlation parameters to reflect the fluctuations and changes in the market in real time. Thus, through the dynamic estimation of the Gaussian Copula correlation parameter , real-time data support is provided for the decision-making engine of the quantitative trading strategy. Based on the integrated analysis of the dynamic correlation parameters and external factors, the trading strategy can quickly respond to changes in market linkage relationships.

[0088] Regarding the implementation details of determining the time-varying Copula parameter based on the GARCH model in step S140, in some examples, the estimation of the Copula parameter is decomposed as follows:

[0089]

[0090] In the formula, μ is the mean term, which is the long-term mean correlation between the target market and the associated market, representing the stable part of the static correlation; ∈ t is the residual term, representing the short-term volatility characteristics at time t.

[0091] In this way, the dynamic characteristics of the time-varying Copula parameter can be decomposed into a long-term trend and a volatile residual.

[0092] Model the conditional volatility of the residual term based on the GARCH model:

[0093]

[0094] In the formula, h t represents the conditional volatility value at time t; μ, ω, α, β, and γ are the model parameters of the GARCH model, and ω is the constant term used to represent the benchmark volatility level; is the short-term volatility shock, representing the impact of the squared residual at the previous time point; βh t-1It represents the long-term memory effect, indicating the continuous influence of the conditional volatility at the previous time point; γz t is the external correction term, used to introduce the external influencing factor z t The impact on volatility.

[0095] Generate time-varying Copula parameters:

[0096]

[0097] In the formula, represents the time-varying Copula parameter at time t; η t ~N(0,1) is the standard normal random variable, used to introduce random perturbations to simulate the random fluctuations of the market.

[0098] Here, the GARCH model can capture the non-stationarity and volatility clustering effect of market prices through the dynamic combination of short-term shocks long-term memory (βh t-1 ) and external correction term (γz t ). Specifically, the carbon market price and trading volume usually exhibit the characteristics of leptokurtosis and heavy tails, directly affecting the volatility and price correlation of the market. By accurately modeling the volatility, traders can better understand the market dynamics and consider the volatility factor in trading strategies to avoid potential losses caused by abnormal market fluctuations. Furthermore, the dynamically generated time-varying Copula parameter accurately reflects the dynamic changes of market correlation at each time point.

[0099] In addition, the model parameters of the GARCH model are automatically determined by maximizing the parameter estimation of the log-likelihood function based on the historical data set, and each historical data in the historical data set contains the estimated Copula parameter and conditional volatility value corresponding to the historical time point.

[0100]

[0101] In the formula, n is the sample size of the historical data set, that is, the number of historical time points; is the estimated Copula parameter at the historical time point t u , is the conditional volatility value at the historical time point t u ; the parameter group Θ = {ω, α, β, γ, μ}, argmax Θ represents the maximization operator, used to optimize Θ; represents the optimal parameter group, used to set the model parameters of the GARCH model.

[0102] It should be noted that by taking the long-term mean μ and other GARCH model parameters (ω, α, β, γ) as part of the optimization of the log-likelihood function, the log-likelihood function combines historical data for parameter optimization, enabling the model parameters to automatically adapt to the dynamic environment of the carbon market, eliminating the subjectivity of traditional manual setting of model parameters, making parameter adjustment fully automated, and ensuring the scientificity and reliability of the optimization results. In addition, μ dynamically determined based on historical data optimization can provide an accurate reference for the long-term mean in the real-time generation of time-varying Copula parameters.

[0103] Therefore, the GARCH model parameters are adaptively optimized by maximizing the log-likelihood function to ensure that the estimated parameter group is the optimal fitting result of the historical data set and can better describe the conditional volatility and dynamic correlation of the carbon market. In addition, according to the historical data sets of different carbon markets, the GARCH model parameters can also be dynamically adjusted to adapt to the volatility characteristics and dynamic correlation of different carbon markets.

[0104] Regarding the implementation details of calculating the CoVaR value according to each time-varying Copula parameter in step S140, it should be noted that CoVaR is a risk measurement index used to evaluate the conditional risk of the target market when extreme events (such as price plummets) occur in the associated market. It represents the value change of the target market under the risk conditions of the associated market. In the carbon market, CoVaR can effectively reflect the risk spillover effect between different markets, thus helping traders quantify the potential impact of multi-market linkage on the investment portfolio.

[0105] In some examples, based on the Gaussian Copula function, the joint distribution of the target market and the associated market is characterized by time-varying Copula parameters.

[0106] The Gaussian Copula function is used to characterize the dynamic non-linear correlation between the target market and the associated market through time-varying Copula parameters to determine the corresponding joint distribution function. Here, a multi-dimensional Copula function is adopted to describe the dynamic non-linear dependence relationship between the target carbon market and multiple associated carbon markets.

[0107]

[0108] In the formula, is the multi-dimensional Copula joint distribution function, which describes the joint distribution relationship between the target market X and M associated markets Y 1 , Y 2 , …, Y M ; is the time-varying Copula parameter vector, which is used to describe the dynamic correlation between markets; u X = F X (x) and is the marginal distribution value of the target market X and the v-th associated market obtained through kernel density estimation; represents a multi-dimensional normal distribution with a mean of 0 and a covariance matrix of Σ t cumulative distribution function; Φ -1 is the inverse cumulative distribution function of the standard normal distribution, used to map the marginal distribution value u X , to the standard normal distribution space; in the covariance matrix Σ t ; represents the time-varying Copula parameter between the target market and the v-th associated market, ρ vs,t represents the time-varying Copula parameter between the v-th associated market and the s-th associated market.

[0109] Then, under the condition of the given associated market, the conditional distribution function of the target market is determined in the following way.

[0110] The conditional distribution is derived from the joint distribution, aiming to deduce the distribution characteristics of the target market under the condition of the given associated market. Specifically, through the multi-dimensional Copula joint distribution, the conditional distribution of the target market under the extreme events of multiple associated markets is calculated.

[0111] The conditional distribution formula is:

[0112]

[0113] It can be further expanded through the formula of Gaussian Copula as:

[0114]

[0115] In the formula, is the conditional distribution function, representing the conditional distribution of the target market X when extreme events occur in the associated markets Y 1 , Y 2 , …, Y M ; represents the time-varying correlation between the target market X and the v-th associated market, represents the joint contribution of the conditional effects of all associated markets to the conditional distribution of the target market; is a correction term to ensure the standardization of the conditional distribution function and avoid the cumulative correlation exceeding 1.

[0116] Here, the conditional distribution function is calculated through the partial derivative of Gaussian Copula, reflecting the dynamic impact of the associated market on the target market under extreme events. Through the derivation of the conditional distribution function, the risk transmission path of the extreme events of the associated market to the target market is clarified.

[0117] After that, the CoVaR value of the target market is calculated through the quantiles of the conditional distribution function.

[0118] It should be noted that the CoVaR value is defined by the quantiles of the conditional distribution function of the target market, that is, under the condition of extreme events in the given associated market, the loss level of the target market is obtained.

[0119] Specifically, the marginal distribution function is obtained through kernel density estimation, and the quantiles are calculated. Through Φ -1 (q), the VaR value is mapped to the standard normal distribution space. The dynamic distribution of the target market is corrected through the conditional distribution function to generate the conditional quantiles of the target market.

[0120]

[0121] In the formula, represents the CoVaR value, which is used to characterize the risk level of the target market X under extreme events in multiple associated markets Y 1 , Y 2 , …, Y M ; represents the inverse function of the distribution function of the target market; p is the confidence level of the target market, and the corresponding quantile is used to calculate the risk of the target market under extreme events; q v is the extreme event quantile of the v-th associated market, represents the comprehensive impact of extreme events in all associated markets on the risk of the target market; represents the correction term to ensure the standardization of the conditional distribution.

[0122] It should be noted that the traditional univariate VaR method only analyzes the risk of the target market itself and does not consider the linkage of multiple markets. In the embodiments of the present application, the CoVaR value can characterize the complex risk characteristics under the linkage of multiple markets by considering the conditional distribution of the associated market.

[0123] In addition, in traditional risk analysis, extreme events usually assume the independence between markets. In the embodiments of the present application, the CoVaR value can analyze the linkage effect between non-independent markets through Gaussian Copula dynamic adjustment, and is particularly suitable for risk assessment under extreme conditions.

[0124] Through the embodiments of the present application, the CoVaR value is used to characterize the target market X in multiple associated markets Y 1 , Y 2 , …, Y nUnder extreme events (such as its return being lower than the VaR value), the potential risk level can depict the complex non-linear dependence relationship and risk transmission mechanism among related markets by virtue of the dynamic adjustment ability of the Gaussian Copula model. Therefore, the risk spillover effects of the target market and multiple related markets are comprehensively considered, rather than the influence of a single related market, providing a more comprehensive risk quantification result for decision-makers.

[0125] Figure 4 Shows an operation flowchart of an example according to Figure 1 Step S150 in.

[0126] As Figure 4 shown, in step S410, the target trading price time series data and the target trading volume time series data corresponding to the target carbon market in the target carbon market trading data are parsed, and the EMD decomposition is performed on the target trading price time series data and the target trading volume time series data to respectively determine the corresponding first IMF component group and the second IMF component group.

[0127] It should be noted that the EMD decomposition (Empirical Mode Decomposition) as a non-linear and non-stationary signal processing method can decompose the trading price time series data and the trading volume time series data of the target carbon market into multiple IMF (Intrinsic Mode Function) components, respectively revealing the market fluctuation characteristics at different time scales. By effectively extracting and processing these IMF components, the high-frequency noise and low-frequency trend components in the price data can be eliminated, thereby improving the accuracy of the quantitative prediction of the carbon quota price.

[0128] Specifically, the first IMF component group represents the market fluctuation characteristics at different time scales revealed by the trading price time series data, and the second IMF component group represents the market fluctuation characteristics at different time scales revealed by the trading volume time series data. Exemplarily, the high-frequency components (such as short-term price fluctuations) in the first IMF component can reflect the short-term fluctuation characteristics and price change sensitivity of the market; the low-frequency trend components (such as long-term trend changes) in the first IMF component can reflect the long-term price trend and trend law of the market.

[0129] In step S420, based on the first IMF component group and the second IMF component group, the target carbon trading time series features are constructed.

[0130] Specifically, the features of the first IMF component group and the second IMF component group are fused to generate complete target carbon trading time series features. For example, weights can be assigned to the features of different IMF components, and the weights can be adjusted according to the prediction requirements of a specific time scale (such as short-term or long-term), and then the fused features are normalized so that the feature data of different time scales are in the same dimension, avoiding affecting the model performance due to data scale differences.

[0131] In step S430, the target carbon trading time series features and the CoVaR value are input into the price prediction model to determine the predicted price of carbon quotas in the target carbon market.

[0132] Through the price prediction model, a comprehensive analysis is performed on the multi-time scale features of the trading price, the volatility and trend features of the trading volume, and the CoVaR value, improving the accuracy of the predicted price of carbon quotas. Specifically, using the target carbon trading time series features, the short-term features of price fluctuations and the long-term trends are comprehensively considered, enabling the prediction model to more accurately reflect the complex dynamics of the market. In addition, taking the CoVaR value as one of the model input features can incorporate the systemic risk of the target market into the price prediction, improving the prediction accuracy of the model in extreme cases.

[0133] It should be noted that the types of price prediction models can be diverse, such as ARIMA models, deep learning models, etc., which are not restricted here for the time being. Exemplarily, the price prediction model can adopt a Transformer model, which can efficiently capture long-term dependencies in time series through the self-attention mechanism and achieve excellent performance when there is a large amount of historical data.

[0134] In some examples of the embodiments of the present application, the price prediction model adopts an LSTM-CNN hybrid model.

[0135] Figure 5 Shows a schematic structural connection diagram of an example of the price prediction model according to the embodiments of the present application.

[0136] As Figure 5 shown, the price prediction model 500 includes an LSTM module 510, a CNN module 520, and a fusion output layer 530. The LSTM module 510 and the CNN module 520 are connected in parallel to the fusion output layer 530. The LSTM module 510 and the CNN module 520 are respectively used to generate a first predicted price and a second predicted price, and the fusion output layer 530 is used to determine the predicted price of carbon quotas in the target carbon market according to the first predicted price and the second predicted price.

[0137] Capture long-term dependencies in time series data through the LSTM module, such as the long-term trends and periodic fluctuations in the carbon market, thereby providing important support for long-term changes in price prediction. Extract local volatility features through the CNN module, which is suitable for processing relatively fine-grained short-term fluctuations, such as the rapid rise and fall of the market within a specific period. Based on the parallel structure of the modules, effective prediction can be carried out at different time scales of short-term and long-term, greatly improving the prediction accuracy and generalization ability.

[0138] Specifically, the fusion output layer 530 is used to process the CoVaR value through the sigmoid function to calculate the fusion weight, and fuse the first predicted price and the second predicted price according to the fusion weight to determine the carbon quota predicted price of the target carbon market:

[0139]

[0140]

[0141] In the formula, represents the dynamic weight of the LSTM module, CoVaR represents the CoVaR value, represents the dynamic weight of the CNN module; ζ LSTM and θ LSTM are determined through sample learning optimization, where ζ LSTM represents the sensitivity of the Sigmoid function to changes in the CoVaR value, θ LSTM represents the CoVaR threshold; P fusion represents the final predicted result after fusion, that is, the carbon quota predicted price of the target carbon market; P LSTM represents the first predicted price generated by the LSTM module, P CNN represents the second predicted price generated by the CNN module.

[0142] Here, the Sigmoid function is used to perform a non-linear mapping on the CoVaR value to generate the dynamic weight of the LSTM module, and the parameters ζ LSTM and θ LSTMUsed to control the sensitivity and balance point of weight changes. Through the design of the fusion output layer that dynamically adjusts the weights using the Sigmoid function, the model can dynamically allocate the prediction contributions of the LSTM module and the CNN module according to the CoVaR value. Exemplarily, in high-risk scenarios (with a large CoVaR value), the weights tilt towards the LSTM module, enabling the model to pay more attention to long-term trends and historical dependencies, which helps generate more robust prediction results during sharp market fluctuations. In low-risk scenarios (with a small CoVaR value), the weights tilt towards the CNN module, and the model will pay more attention to short-term local features, thereby capturing rapid fluctuations within a short period and improving the timeliness of predictions. Thus, through the dynamic weight mechanism based on the CoVaR value, the prediction output contributions of the LSTM and CNN are adjusted, avoiding the overfitting problem that may be caused by using fixed weights in all market situations, and thereby improving the accuracy of the predicted price of carbon quotas in the target carbon market.

[0143] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of actions combined. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application. In the above embodiments, each embodiment is described with its own emphasis. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0144] Figure 6 The structural block diagram of an example of a carbon value-based quantitative trading data analysis system according to an embodiment of the present application is shown.

[0145] As Figure 6 shown, the carbon value-based quantitative trading data analysis system 600 includes an information acquisition unit 610, a distribution analysis unit 620, a Copula calculation unit 630, a CoVaR calculation unit 640, and a price prediction unit 650.

[0146] The information acquisition unit 610 is used to acquire carbon market trading data groups, macroeconomic indicators, and carbon trading policy indicators; the carbon market trading data group includes target carbon market trading data and multiple associated carbon market trading data.

[0147] The distribution analysis unit 620 is used to perform statistical analysis on the price return sequences of each carbon market based on the carbon market trading data group, and estimate the corresponding marginal distribution function by the kernel density estimation method.

[0148] The Copula calculation unit 630 is used to fuse the macroeconomic indicators and the carbon trading policy indicators for each of the carbon markets based on the Bayesian method to determine the estimated Copula parameters corresponding to each of the marginal distribution functions; the estimated Copula parameters are used to measure the linear dependence between different marginal distribution functions under the macroeconomic indicators and / or carbon trading policy indicators.

[0149] The CoVaR calculation unit 640 is used to model the time-varying characteristics of each of the estimated Copula parameters based on the GARCH model to determine the corresponding time-varying Copula parameters, and calculate the CoVaR value according to each of the time-varying Copula parameters; the CoVaR value is used to measure the risk spillover effect of the target carbon market when extreme events occur in each of the associated carbon markets.

[0150] The price prediction unit 650 is used to extract the target carbon trading time series features corresponding to the target carbon market trading data, and process the target carbon trading time series features and the CoVaR value based on a price prediction model to determine the predicted price of the carbon quota in the target carbon market; wherein, the predicted price of the carbon quota is transmitted to the quantitative trading decision-making engine to assist in determining the corresponding quantitative trading strategy.

[0151] In some embodiments, the embodiments of the present application provide a non-volatile computer-readable storage medium, in which one or more programs including execution instructions are stored, and the execution instructions can be read and executed by an electronic device (including but not limited to a computer, a server, or a network device, etc.) to be used to execute the steps of any one of the above-mentioned carbon value-based quantitative trading data analysis methods of the present application.

[0152] In some embodiments, the embodiments of the present application further provide a computer program product, the computer program product includes a computer program stored on a non-volatile computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer, the computer is made to execute the steps of any one of the above-mentioned carbon value-based quantitative trading data analysis methods.

[0153] In some embodiments, the embodiments of the present application further provide an electronic device, which includes: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steps of the carbon value-based quantitative trading data analysis method.

[0154] Figure 7 is a schematic hardware structure diagram of an electronic device for executing the carbon value-based quantitative trading data analysis method provided by another embodiment of the present application, asFigure 7 As shown, the device includes:

[0155] One or more processors 710 and a memory 720, Figure 7 Taking one processor 710 as an example.

[0156] The device for executing the carbon-value-based quantitative trading data analysis method may further include: an input device 730 and an output device 740.

[0157] The processor 710, the memory 720, the input device 730, and the output device 740 may be connected via a bus or other means, Figure 7 Taking connection via a bus as an example.

[0158] The memory 720, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the carbon-value-based quantitative trading data analysis method in the embodiments of the present application. The processor 710 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions, and modules stored in the memory 720, that is, implements the carbon-value-based quantitative trading data analysis method in the above method embodiments.

[0159] The memory 720 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the electronic device, etc. In addition, the memory 720 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 720 may optionally include a memory remotely set relative to the processor 710, and these remote memories can be connected to the electronic device through a network. Examples of the above network include but are not limited to the Internet, enterprise intranets, local area networks, mobile communication networks, and their combinations.

[0160] The input device 730 can receive input digital or character information, and generate signals related to the user settings and function control of the electronic device. The output device 740 may include a display device such as a display screen.

[0161] The one or more modules are stored in the memory 720, and when executed by the one or more processors 710, execute the carbon-value-based quantitative trading data analysis method in any of the above method embodiments.

[0162] The above products can execute the methods provided in the embodiments of the present application, and have the corresponding functional modules and beneficial effects for executing the methods. For technical details not described in detail in this embodiment, reference may be made to the methods provided in the embodiments of the present application.

[0163] The electronic devices in the embodiments of the present application exist in various forms, including but not limited to:

[0164] (1) Mobile communication devices: These devices are characterized by having mobile communication functions and mainly aim to provide voice and data communication. Such terminals include: smart phones, multimedia phones, functional phones, and low-end phones, etc.

[0165] (2) Ultra-mobile personal computer devices: These devices belong to the category of personal computers, have computing and processing functions, and generally also have the characteristic of mobile Internet access. Such terminals include: PDA, MID, and UMPC devices, etc.

[0166] (3) Portable entertainment devices: These devices can display and play multimedia content. Such devices include: audio and video players, handheld game consoles, e-books, and intelligent toys and portable vehicle navigation devices.

[0167] (4) Other airborne electronic devices with data interaction functions, such as in-vehicle device installed on a vehicle.

[0168] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0169] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solutions or the part that contributes to the related technologies can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A quantitative trading data analysis method based on carbon value, comprising: Acquire a carbon market transaction data group, macroeconomic indicators and carbon trading policy indicators; the carbon market transaction data group includes target carbon market transaction data and multiple related carbon market transaction data; Based on the carbon market transaction data set, statistically analyzing the price return series of each carbon market, and estimating the corresponding marginal distribution function by kernel density estimation method; For each of the carbon markets, the macroeconomic indicators and the carbon trading policy indicators are integrated based on the Bayesian method to determine the estimated Copula parameters corresponding to each of the marginal distribution functions; the estimated Copula parameters are used to measure the linear dependence between different marginal distribution functions under the macroeconomic indicators and / or carbon trading policy indicators; Modeling the time-varying characteristics of each of the estimated Copula parameters based on the GARCH model to determine the corresponding time-varying Copula parameters, and calculating the CoVaR value based on each of the time-varying Copula parameters; the CoVaR value is used to measure the risk spillover effect of the target carbon market when an extreme event occurs in each of the given associated carbon markets; The target carbon transaction time series characteristics corresponding to the target carbon market transaction data are extracted, and the target carbon transaction time series characteristics and the CoVaR value are processed based on a price prediction model to determine the carbon quota prediction price of the target carbon market; wherein the carbon quota prediction price is passed to a quantitative trading decision engine to assist in determining a corresponding quantitative trading strategy.

2. The method according to claim 1, wherein: The carbon market trading data includes the trading price and trading volume of carbon quotas in the carbon market; the macroeconomic indicators include the real-time prices of energy commodities, GDP growth rate, consumer price index and producer price index; the carbon trading policy indicators include carbon emission policies, carbon tax policies and carbon reduction target information.

3. The method according to claim 1, wherein: The method of performing statistical analysis on the price return series of each carbon market based on the carbon market transaction data set and estimating the corresponding marginal distribution function by a kernel density estimation method includes: Calculate the logarithmic return of the carbon quota price in the carbon market at each time point to determine the corresponding carbon market price return series: Where P i,t and P i,t-1 Respectively represent the carbon quota transaction price of carbon market i at time t and time t-1; r i,t is the price return of carbon market i at time t, representing the rate of change relative to the price at the previous moment; For each carbon market price return series, the kernel density estimation method is used to obtain its corresponding marginal distribution function: In the formula, f i (r) is the marginal distribution function of carbon market i, which represents the density at price return r; h is the bandwidth parameter used to control the width of the kernel function; T represents the time length of the price return series, and K(·) represents the Gaussian kernel function.

4. The method according to claim 3, wherein: Determining the estimated Copula parameter corresponding to the marginal distribution function includes: The Gaussian Copula function is used to describe the linear correlation between multivariate random variables, and its joint distribution function is defined as: C(u1,u2;ρ t )=Φ ρ (F -1 (u1),Φ -1 (u2)), u1=f i (r i,t ),u2=f j (r j,t ), In the formula, u1 and u2 represent the marginal distribution function values ​​of the target carbon market i and the associated carbon market j at time t, respectively, C(u1,u2;ρ t ) represents the Gaussian Copula joint distribution function, which describes the joint distribution between the target market and the associated market; ρ t is the dynamic correlation parameter of Gaussian Copula, which represents the linear correlation between target carbon market i and associated carbon market j at time t; Φ ρ is the cumulative distribution function of the two-dimensional normal distribution, Φ -1 is the inverse cumulative distribution function of the standard normal distribution; By using the Bayesian method, external factors are incorporated into the estimation of the Gaussian Copula correlation parameters to determine the posterior distribution: In the formula, z t represents the external influencing factors at time t, including macroeconomic indicators and carbon trading policy indicators; r i,t and r j,t denote the price returns of target carbon market i and associated carbon market j at time t respectively; p(ρ t |r i,t ,r j,t ,z t ) is the posterior distribution, which means that when the price return sequence r is observed i,t 、r j,t and external factors t After that, ρ t The probability distribution of p(r i,t ,r j,t |ρ t ) represents the price return sequence r i,t and r j,t At a given ρ t The joint distribution probability under t |ρ t ) represents external factors z t At a given ρ t The conditional probability distribution under t ) represents the prior Gaussian distribution, which is the distribution of ρ t The initial assumption is that p(r i,t ,r j,t ,z t ) is a normalization constant, ensuring that the integral of the posterior distribution is 1; The influence of external factors on Gaussian Copula parameters is expressed by linear weighting to determine the conditional distribution of external factors: In the formula, λ is z t For the correlation parameter ρ t The weight coefficient is used to quantify the impact of external factors on market relevance; is the variance of external factors, which indicates the uncertainty of the impact of external factors on the correlation; Construct the log-likelihood function of the Gaussian Copula: In the formula, Represents the log-likelihood function of the Gaussian Copula for the dynamic correlation parameter ρ t estimates; is the correction factor to ensure the stability of the log-likelihood; According to the Bayesian posterior distribution, the conditional distribution weights of external factors are introduced to maximize the weighted log-likelihood function to calculate the estimated Copula parameters corresponding to the marginal distribution function: In the formula, represents the final estimated Copula parameters at time t; represents the maximization operator, used to optimize ρ t ; lnp(z t |ρ t ) represents the weighted correction term of the conditional distribution of external factors with respect to the log-likelihood.

5. The method according to claim 4, wherein: Modeling the time-varying characteristics of each estimated Copula parameter based on the GARCH model to determine the corresponding time-varying Copula parameter includes: Decompose the estimated Copula parameters: Where μ is the mean term, which is the long-term mean correlation between the target market and the associated market, and represents the stable part of the static correlation; ∈ t is the residual term, which represents the short-term volatility characteristics at time t; Modeling the conditional volatility of the residual term based on the GARCH model: In the formula, h t represents the conditional volatility value at time t; μ, ω, α, β and γ are the model parameters of the GARCH model, and ω is a constant term used to represent the benchmark volatility level; is a short-term volatility shock, indicating the impact of the square of the residual at the previous time point; βh t-1 is the long-term memory effect, which indicates the continued influence of the volatility of the conditions at the previous time point; γz t is an external correction term, used to introduce external influencing factors z t Impact on volatility; Generate time-varying Copula parameters: In the formula, represents the time-varying Copula parameter at time t; η t ~N(0,1) is a standard normal random variable, which is used to introduce random disturbances to simulate random fluctuations in the market; Wherein, the model parameters of the GARCH model are automatically determined by maximizing the parameter estimation of the log-likelihood function according to the historical data set, and each historical data in the historical data set contains the estimated Copula parameters and conditional volatility values ​​at the corresponding historical time point; Where n is the sample size of the historical data set, that is, the number of historical time points; is the historical time point t u The estimated Copula parameters of is the historical time point t u Conditional fluctuation value; parameter group Θ = {ω, α, β, γ, μ}, argmax Θ represents the maximization operator, used to optimize Θ; Represents the optimal parameter group, which is used to set the model parameters of the GARCH model.

6. The method according to claim 5, wherein: The calculating of the CoVaR value according to each of the time-varying Copula parameters comprises: Based on the Gaussian Copula function, the joint distribution of the target market and the associated market is characterized by the time-varying Copula parameters: In the formula, is a multidimensional Copula joint distribution function, describing the target market X and M associated markets Y1, Y2, …, Y M The joint distribution relationship between is the time-varying Copula parameter vector, used to describe the dynamic correlation between markets; u X =F X (x) and is the marginal distribution value of the target market X and the vth associated market obtained by kernel density estimation; Indicates that the mean is 0 and the covariance matrix is ​​Σ t The cumulative distribution function of the multidimensional normal distribution; Φ -1 is the inverse cumulative distribution function of the standard normal distribution, which is used to convert the marginal distribution value u X , Mapped to the standard normal distribution space; in the covariance matrix Σ t middle, represents the time-varying Copula parameter between the target market and the vth related market, ρ vs,t represents the time-varying Copula parameters of the vth related market and the sth related market; Given the associated market conditions, the conditional distribution function of the target market is determined by: The conditional distribution formula is: The Gaussian Copula formula can be further expanded as follows: In the formula, is the conditional distribution function, which means that in the associated markets Y1, Y2, …, Y M The conditional distribution of target market X when an extreme event occurs; represents the time-varying correlation between the target market X and the vth associated market, represents the joint contribution of the conditional effects of all linked markets to the conditional distribution of the target market; To correct the term, ensure the normalization of the conditional distribution function to avoid the accumulation of correlations exceeding 1; The CoVaR value of the target market is calculated by the quantile of the conditional distribution function: In the formula, Represents the CoVaR value, which is used to characterize the target market X in multiple related markets Y1, Y2, …, Y M The level of risk under conditions of extreme events; represents the inverse function of the target market’s distribution function; p is the target market’s confidence level, and the corresponding quantile is used to calculate the target market’s risk under extreme event conditions; q v is the extreme event quantile of the vth related market, Represents the combined impact of extreme events in all related markets on the target market risk; represents a correction term to ensure the normalization of the conditional distribution.

7. The method according to any one of claims 1 to 6, wherein: The extracting the target carbon transaction time series characteristics corresponding to the target carbon market transaction data, and processing the target carbon transaction time series characteristics and the CoVaR value based on a price prediction model to determine the predicted price of the carbon quota of the target carbon market includes: Parsing the target transaction price time series data and the target transaction volume time series data corresponding to the target carbon market in the target carbon market transaction data, and performing EMD decomposition on the target transaction price time series data and the target transaction volume time series data to respectively determine the corresponding first IMF component group and second IMF component group; constructing a target carbon transaction time series feature based on the first IMF component group and the second IMF component group; Inputting the target carbon transaction time series characteristics and the CoVaR value into a price prediction model to determine the predicted price of carbon quotas in the target carbon market; the price prediction model comprises an LSTM module, a CNN module and a fusion output layer, and the LSTM module and the CNN module are connected in parallel to the fusion output layer; Among them, the LSTM module and the CNN module are used to generate a first predicted price and a second predicted price respectively, and the fusion output layer is used to determine the carbon quota predicted price of the target carbon market according to the first predicted price and the second predicted price.

8. The method according to claim 7, wherein: The fusion output layer is used to process the CoVaR value through a sigmoid function to calculate a fusion weight, and fuse the first predicted price and the second predicted price according to the fusion weight to determine the predicted price of carbon quotas in the target carbon market: In the formula, represents the dynamic weight of the LSTM module, CoVaR represents the CoVaR value, represents the dynamic weight of the CNN module; ζ LSTM and θ LSTM is determined by sample learning optimization, where ζ LSTM represents the sensitivity of the Sigmoid function to changes in the CoVaR value, θ LSTM represents the CoVaR threshold; P fusion represents the final prediction result after fusion, that is, the predicted price of carbon quota in the target carbon market; P LSTM represents the first predicted price generated by the LSTM module, P CNN Represents the second predicted price generated by the CNN module.

9. A quantitative trading data analysis system based on carbon value, comprising: An information acquisition unit, used to acquire a carbon market transaction data group, macroeconomic indicators and carbon trading policy indicators; the carbon market transaction data group includes target carbon market transaction data and multiple related carbon market transaction data; A distribution analysis unit, used to perform statistical analysis on the price return series of each carbon market based on the carbon market transaction data group, and estimate the corresponding marginal distribution function by a kernel density estimation method; A Copula calculation unit, for each of the carbon markets, fusing the macroeconomic indicators and the carbon trading policy indicators based on a Bayesian method to determine an estimated Copula parameter corresponding to each of the marginal distribution functions; the estimated Copula parameter is used to measure the linear dependence between different marginal distribution functions under macroeconomic indicators and / or carbon trading policy indicators; A CoVaR calculation unit, used for modeling the time-varying characteristics of each of the estimated Copula parameters based on a GARCH model to determine the corresponding time-varying Copula parameters, and calculating the CoVaR value according to each of the time-varying Copula parameters; the CoVaR value is used to measure the risk spillover effect of the target carbon market when an extreme event occurs in each of the given associated carbon markets; A price prediction unit is used to extract the target carbon transaction time series characteristics corresponding to the target carbon market transaction data, and process the target carbon transaction time series characteristics and the CoVaR value based on a price prediction model to determine the carbon quota prediction price of the target carbon market; wherein the carbon quota prediction price is transmitted to the quantitative trading decision engine to assist in determining the corresponding quantitative trading strategy.

10. A computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the steps of the carbon value-based quantitative trading data analysis method as described in any one of claims 1 to 8.

Images