Financial risk prediction method and system based on artificial intelligence

Abstract

The invention relates to the technical field of financial risk prediction, in particular to a financial risk prediction method and system based on artificial intelligence, and the method comprises the following steps: S1, data collection and preprocessing: collecting multi-dimensional data from a plurality of financial markets in real time; s2, constructing a cross-market financial product association network: constructing the cross-market financial product association network, and identifying the association among the financial markets; s3, risk conduction path identification: identifying a risk conduction path between financial markets; s4, risk analysis: predicting risk fluctuations in different financial markets through the financial risk prediction model; and S5, risk prediction result output: outputting prediction results of risks of different financial markets, wherein the prediction results comprise risk scores and risk fluctuation trends of the markets. According to the invention, the potential risk conduction path in the market can be effectively identified, so that a more comprehensive risk early warning and management tool is provided for market participants.

Description

Technical Field

[0001] The present invention relates to the technical field of financial risk prediction, and in particular to a financial risk prediction method and system based on artificial intelligence. Background Art

[0002] With the rapid development of the global financial market, financial products are becoming increasingly diversified, and the correlation and mutual influence between different markets are becoming more complex. The risk factors in the financial market are not limited to a single market, but are complex dynamic risks transmitted across markets. Therefore, how to conduct efficient and accurate financial risk prediction in a multi-market environment has become a key research topic in the field of financial technology. Traditional financial risk prediction methods are often based on analysis of data from a single market or a single dimension, which makes it difficult to capture the complex relationships across markets and products and cannot provide comprehensive risk management support.

[0003] Existing financial risk prediction technologies have many shortcomings when facing complex multi-dimensional and multi-market environments. First, traditional methods are often limited to simple regression models of historical trading data, which make it difficult to capture nonlinear fluctuations in the market and cross-market risk transmission mechanisms. Second, existing models often lack adaptive adjustment mechanisms when facing dynamic market changes and are unable to promptly reflect real-time changes in market sentiment and volatility. Finally, many financial risk prediction methods suffer from low computational efficiency when processing large-scale financial data, making it difficult to provide real-time risk warnings in a high-frequency market environment.

[0004] The purpose of this invention is to propose a financial risk prediction method and system based on artificial intelligence, which can not only accurately capture the risk transmission path between financial markets, but also dynamically adjust the model weights to adapt to market fluctuations and changes in sentiment, thereby providing more comprehensive and accurate support for risk management and decision-making in financial markets. Summary of the invention

[0005] The present invention provides a financial risk prediction method and system based on artificial intelligence.

[0006] A financial risk prediction method based on artificial intelligence, comprising the following steps:

[0007] S1, data collection and preprocessing: real-time collection of multi-dimensional data from multiple financial markets, including historical transaction data, macroeconomic indicators and market sentiment data, and preprocessing of the collected multi-dimensional data, including data cleaning, denoising and standardization;

[0008] S2, Construction of cross-market financial product association network: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products;

[0009] S3, risk transmission path identification: Based on the constructed cross-market financial product association network, the risk transmission path between financial markets is identified;

[0010] S4, Risk Analysis: Combined with the identified risk transmission paths, the risk fluctuations in different financial markets are predicted through financial risk prediction models;

[0011] S5, risk prediction result output: output the prediction results of different financial market risks, including the market risk score and risk fluctuation trend.

[0012] Optionally, the data collection and preprocessing in S1 includes:

[0013] S11, data collection: real-time collection of multi-dimensional data from stock, bond and futures financial markets, including historical transaction data (opening price, closing price, trading volume), macroeconomic indicators (GDP, inflation rate, interest rate) and market sentiment data (sentiment analysis results obtained from news and social media);

[0014] S12, data cleaning and denoising: clean the collected multi-dimensional data, use missing value interpolation to fill in the missing data, and use wavelet transform to remove noise signals in financial data;

[0015] S13, data standardization: standardize the multidimensional data after cleaning and denoising, and use the z-score standardization method to normalize the data to a distribution with a mean of 0 and a standard deviation of 1.

[0016] Optionally, the construction of the cross-market financial product association network in S2 includes:

[0017] S21, multi-dimensional data mapping and conversion: converting the collected multi-dimensional data into node attributes and edge weights;

[0018] S22, graph structure construction of association network: Based on node and edge weights, a graph neural network (GNN) is used to construct a cross-market financial product association network;

[0019] S23, Cross-market correlation analysis and optimization: Based on the graph structure of the constructed correlation network, analyze the correlation between financial markets, use the aggregation function in the graph neural network (GNN) to summarize and update the neighborhood information of the nodes, identify high-risk correlation paths, and dynamically adjust the edge weights.

[0020] Optionally, the multi-dimensional data mapping and conversion in S21 includes:

[0021] S211, node attribute generation: map the collected multi-dimensional data into the attributes of each financial product node. Each node represents a financial product. The node attributes include the transaction volume T of the product. v 、Price Fluctuation v and volatility V v ;

[0022] S212, edge weight calculation: The edge weights between nodes are calculated based on the transaction volume and price correlation between financial products. The edge weights reflect the strength of the association between two financial products.

[0023] Optionally, the construction of the graph structure of the association network in S22 includes:

[0024] S221, graph structure initialization: Based on the generated node attributes and edge weights, a graph structure G = (V, E) is constructed, where V is a node set representing different financial products, and E is an edge set representing the relationship between financial products. uv The weight of ∈E is determined by the edge weight, which represents the strength of the association between nodes u and v;

[0025] S222, feature aggregation of graph neural network layer: Use the aggregation function in the graph neural network (GNN) to update the features of each node, and set the feature vector of node v to At the kth layer, feature updating is achieved by aggregating the features of its neighboring nodes.

[0026] Optionally, the cross-market correlation analysis and optimization in S23 includes:

[0027] S231, node feature aggregation and update: Based on the constructed cross-market financial product association network, the neighborhood information of the node is aggregated and updated, that is, the feature vector of node v In the k+1th layer, by aggregating the features of its neighboring nodes renew;

[0028] S232, high-risk association path identification: By analyzing the updated feature vectors of the nodes, we can identify the high-risk association paths across markets. Based on the strength of the association between financial products and market volatility, the high edge weights w on the paths are uv And the node volatility V v The superposition reflects a potential high-risk association;

[0029] S233, Dynamic adjustment of edge weights: After identifying high-risk association paths, dynamically adjust edge weights w based on real-time market performance and new multi-dimensional data uv .

[0030] Optionally, the risk transmission path identification in S3 includes:

[0031] S31, Path Definition: Based on the constructed cross-market financial product association network, define the risk transmission path between financial products. The path consists of multiple connected nodes and edges. Each edge represents the strength of the association between different financial products. The node represents the financial product. The path length is determined by the degree of association between financial products and the transmission delay.

[0032] S32, Risk transmission path weight calculation: After identifying the potential transmission path between markets, calculate the overall risk transmission strength of the path. The transmission strength is determined by the weight w of each side on the path. uv and the market volatility V of the node v The combined impact of

[0033] S33, Identification of risk transmission direction: Determine the direction of risk transmission by combining historical multi-dimensional data and price fluctuations of the financial market. Through the correlation of edges and the market volatility information of nodes, identify the trend of risk transmission from high-volatility products to low-volatility products and form a transmission path.

[0034] Optionally, the financial risk prediction model in S4 adopts a Gaussian mixture model (GMM), and the Gaussian mixture model (GMM) includes:

[0035] S41, multi-market state modeling: In view of the polymorphism of different products and markets in the financial market (such as bull market, bear market, volatile market, etc.), multiple Gaussian distributions are introduced to simulate different market states;

[0036] S42, Dynamic Weight Adjustment: To adapt to the dynamic changes in the financial market, the mixed weight π of each Gaussian distribution is adjusted. k Make dynamic adjustments and dynamically update weights based on the historical volatility of the market;

[0037] S43, Dynamic update of covariance matrix of time series: The data of financial markets have the characteristics of time series. The dynamic covariance matrix update mechanism is introduced to capture the risk transmission and market correlation changes of financial markets at different time points by continuously updating the covariance matrix.

[0038] S44, introduction of market sentiment data: Market sentiment data captures the emotional fluctuations of market participants through social media and news. Market sentiment data is introduced as an additional input feature into the Gaussian mixture model, and the weight of market status is adjusted in combination with subjective emotional fluctuations;

[0039] S45, Risk score calculation: Based on the volatility, sentiment data and weights of different market states, calculate the risk score R for each market state k .

[0040] Optionally, the risk prediction result output in S5 includes:

[0041] S51, Risk score output: Calculate and output the risk score R for each market state based on the mixed weight, volatility and market sentiment data of different market states k , the risk score reflects the comprehensive risk level of the market in its current state;

[0042] S52, Risk volatility trend prediction: Based on the covariance matrix and sentiment fluctuations of historical multi-dimensional data, the volatility trend of the financial market is predicted through the autoregressive moving average model (ARMA). By analyzing market sentiment, volatility and historical trends, the future market volatility trend is output, including trend signals of rising or falling risks.

[0043] A financial risk prediction system based on artificial intelligence, used to implement the above-mentioned financial risk prediction method based on artificial intelligence, includes the following modules:

[0044] Data collection and preprocessing module: collects multi-dimensional data from multiple financial markets in real time, including historical transaction data, macroeconomic indicators, and market sentiment data, and cleans, denoises, and standardizes the collected data;

[0045] Cross-market financial product association network construction module: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products;

[0046] Risk transmission path identification module: Based on the cross-market financial product association network, it identifies the risk transmission path between different financial markets;

[0047] Risk analysis module: Combined with the identified risk transmission paths, the financial risk prediction model is used to analyze and predict the risk fluctuations in various financial markets;

[0048] Risk prediction result output module: outputs the risk score and fluctuation trend of each market.

[0049] Beneficial effects of the present invention:

[0050] The present invention integrates the real-time collection and processing of multi-dimensional data, realizes the accurate capture of the complex relationships between different financial markets, and builds a cross-market financial product association network by using graph neural networks. It combines the transaction volume, price fluctuations and volatility between financial products to enhance the ability to analyze cross-market risks. It can effectively identify potential risk transmission paths in the market, thereby providing market participants with more comprehensive risk warning and management tools.

[0051] The present invention, by introducing a Gaussian mixture model, combines multi-market state modeling, dynamic weight adjustment, updating of the time series covariance matrix and market sentiment data, can accurately adapt to dynamic market changes, comprehensively capture risk fluctuations and transmission paths under different market states, improve the accuracy and stability of financial market risk prediction, and provide a solid foundation for risk assessment and decision-making.

[0052] The present invention can effectively combine market volatility, sentiment data and historical trends through the output of risk prediction results, provide real-time market risk scoring and volatility trend prediction, provide users with forward-looking decision support under different market conditions, and greatly improve the response efficiency and accuracy of financial risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 A schematic diagram of a prediction method flow chart of an embodiment of the present invention;

[0055] Figure 2 Schematic diagram of system function modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0057] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0058] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0059] like Figure 1 As shown, a financial risk prediction method based on artificial intelligence includes the following steps:

[0060] S1, data collection and preprocessing: real-time collection of multi-dimensional data from multiple financial markets, including historical transaction data, macroeconomic indicators and market sentiment data, and preprocessing of the collected multi-dimensional data, including data cleaning, denoising and standardization;

[0061] S2, Construction of cross-market financial product association network: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products;

[0062] S3, risk transmission path identification: Based on the constructed cross-market financial product association network, the risk transmission path between financial markets is identified;

[0063] S4, Risk Analysis: Combined with the identified risk transmission paths, the risk fluctuations in different financial markets are predicted through financial risk prediction models;

[0064] S5, risk prediction result output: output the prediction results of different financial market risks, including the market risk score and risk fluctuation trend;

[0065] Through the above content, efficient prediction and management of complex risks among different financial markets are achieved, which can respond quickly to real-time market changes, improve the accuracy and systematicness of financial risk prediction, and provide a more comprehensive basis for financial decision-making.

[0066] Data collection and preprocessing in S1 include:

[0067] S11, data collection: real-time collection of multi-dimensional data from stock, bond and futures financial markets, including historical transaction data (opening price, closing price, trading volume), macroeconomic indicators (GDP, inflation rate, interest rate) and market sentiment data (sentiment analysis results obtained from news and social media);

[0068] S12, data cleaning and denoising: Clean the collected multi-dimensional data, use missing value interpolation to fill in the missing data, and use wavelet transform to remove the noise signal in the financial data, which is expressed as:

[0069]

[0070] Among them, x i is a missing value, x i-1 and x i+1 are adjacent data points;

[0071] D(t)=∑ k W k ψ k (t);

[0072] Where D(t) is the denoised signal, W k is the wavelet coefficient, ψ k (t) is the corresponding wavelet function;

[0073] S13, data standardization: standardize the multi-dimensional data after cleaning and denoising, and use the z-score standardization method to normalize the data to a distribution with a mean of 0 and a standard deviation of 1;

[0074] Through the above content, the quality of financial data can be effectively improved, noise interference can be reduced, stable and efficient input can be provided for subsequent risk prediction models, and the prediction accuracy and robustness of the model can be significantly improved.

[0075] The construction of the cross-market financial product association network in S2 includes:

[0076] S21, multi-dimensional data mapping and conversion: converting the collected multi-dimensional data into node attributes and edge weights;

[0077] S22, graph structure construction of association network: Based on node and edge weights, a graph neural network (GNN) is used to construct a cross-market financial product association network;

[0078] S23, cross-market correlation analysis and optimization: Based on the graph structure of the constructed correlation network, analyze the correlation between financial markets, use the aggregation function in the graph neural network (GNN) to summarize and update the neighborhood information of the nodes, identify high-risk correlation paths, and dynamically adjust the edge weights to enhance the robustness of the model when the market fluctuates violently and optimize the network's predictive ability;

[0079] Through the above content, we can effectively capture the complex dynamic relationships between financial markets, use graph neural networks to analyze the correlation between market products, improve the ability to identify cross-market risks, and enhance the robustness of the system in the event of drastic market fluctuations by dynamically adjusting network weights, significantly improving the accuracy and stability of financial risk prediction.

[0080] Multi-dimensional data mapping and conversion in S21 include:

[0081] S211, node attribute generation: map the collected multi-dimensional data into the attributes of each financial product node. Each node represents a financial product. The node attributes include the transaction volume T of the product. v 、Price Fluctuation v and volatility V v , expressed as:

[0082] A v ={T v ,P v ,V v}

[0083] Among them, A v Represents the attribute set of node v, including market data of multiple dimensions;

[0084] S212, edge weight calculation: The edge weights between nodes are calculated based on the transaction volume and price correlation between financial products. The edge weight reflects the strength of the association between two financial products and is expressed as:

[0085]

[0086] Among them, w uv represents the edge weight between nodes u and v, corr(P u ,P v ) is the correlation coefficient of the price fluctuations of the two products, T u and T v are the trading volumes of the two products respectively;

[0087] Through the above content, the multi-dimensional data of the financial market can be effectively structured, generating node and edge attributes that are helpful for subsequent graph neural network analysis, thereby improving the accuracy and efficiency of cross-market risk correlation analysis.

[0088] The graph structure construction of the association network in S22 includes:

[0089] S221, graph structure initialization: Based on the generated node attributes and edge weights, a graph structure G = (V, E) is constructed, where V is a node set representing different financial products, and E is an edge set representing the relationship between financial products. uvThe weight of ∈E is determined by the edge weight, which represents the strength of the association between nodes u and v;

[0090] S222, feature aggregation of graph neural network layer: Use the aggregation function in the graph neural network (GNN) to update the features of each node, and set the feature vector of node v to At the kth layer, feature update is achieved by aggregating the features of its neighboring nodes, expressed as:

[0091]

[0092] in, is the feature vector of node v at the k+1th layer, W (k) is the learnable weight matrix of the kth layer, σ is the activation function, is the feature vector of node v at layer k, is the set of neighbor nodes of node v, w uv For edge e uv The weight of is the feature vector of node u in layer k, where u is the neighbor node of v;

[0093] Through the above content, we can effectively capture the complex cross-market relationships in the financial market, and use multi-layer networks to gradually integrate the information of nodes and their neighboring nodes, making the correlation between financial products more accurate and comprehensive, and providing efficient structural support for the dynamic analysis of cross-market risks.

[0094] Cross-market correlation analysis and optimization in S23 include:

[0095] S231, node feature aggregation and update: Based on the constructed cross-market financial product association network, the neighborhood information of the node is aggregated and updated, that is, the feature vector of node v In the k+1th layer, by aggregating the features of its neighboring nodes renew;

[0096] S232, high-risk association path identification: By analyzing the updated feature vectors of the nodes, we can identify the high-risk association paths across markets. Based on the strength of the association between financial products and market volatility, the high edge weights w on the paths are uv And the node volatility V v The overlap reflects a potential high-risk association, expressed as:

[0097] R uv =w uv ×(V u +V v );

[0098] Among them, R uvScore the risk association between node u and node v, V u and V v is the market volatility of the corresponding node;

[0099] S233, Dynamic adjustment of edge weights: After identifying high-risk association paths, dynamically adjust edge weights w based on real-time market performance and new multi-dimensional data uv , which enables the model to more sensitively capture changes in market risk, expressed as:

[0100]

[0101] in, is the new edge weight between node u and node v, is the original edge weight between node u and node v, α is the smoothing factor, which controls the speed of edge weight update, corr(P u ,P v ) is the correlation between the prices of two nodes;

[0102] Through the above content, it is possible to accurately identify the high-risk correlation paths in the financial market. It can not only capture the complex correlations between markets, but also have the ability to adaptively adjust. As the market fluctuates, the risk assessment model is continuously optimized, which enhances the model's sensitivity to sudden risks and the accuracy of overall predictions.

[0103] The risk transmission path identification in S3 includes:

[0104] S31, Path Definition: Based on the constructed cross-market financial product association network, define the risk transmission path between financial products. The path consists of multiple connected nodes and edges. Each edge represents the strength of the association between different financial products. The node represents the financial product. The path length is determined by the degree of association between financial products and the transmission delay.

[0105] S32, Risk transmission path weight calculation: After identifying the potential transmission path between markets, calculate the overall risk transmission strength of the path. The transmission strength is determined by the weight w of each side on the path. uv and the market volatility V of the node v The comprehensive influence of is determined as:

[0106] R path =∑ (u,v)∈path w uv ×(V u +V v );

[0107] Among them, R path is the transmission risk score of the path, w uv is the edge weight, V u and Vv are the volatility of the nodes on the path respectively;

[0108] S33, risk transmission direction identification: Combine the historical multi-dimensional data and price fluctuations of the financial market to determine the direction of risk transmission. Through the correlation of edges and the market volatility information of nodes, identify the trend of risk transmission from high volatility products to low volatility products, and form a transmission path, which can be expressed as:

[0109]

[0110] Among them, D uv is the intensity of risk transmission from node u to node v, V u and V v denote the market volatility of node u and node v respectively, corr(P u ,P v ) is node u and node v The price correlation is calculated using the Pearson correlation coefficient. If D uv >1, it means that node u transmits risk to node v, that is, the risk is transmitted from the node with larger fluctuation to the node with smaller fluctuation. uv When <1, it means that node v transmits risk to node u;

[0111] Through the above content, not only can the risk transmission path between markets be accurately identified, but also the intensity and direction of risk flow can be quantified by calculating the transmission direction, helping users to more clearly understand the source and propagation method of risk, thereby providing more effective risk management and early warning in a changing market environment, and improving the foresight and accuracy of financial decision-making.

[0112] The financial risk prediction model in S4 adopts the Gaussian mixture model (GMM), which includes:

[0113] S41, multi-market state modeling: In view of the polymorphism of different products and different markets in the financial market (such as bull market, bear market, volatile market, etc.), multiple Gaussian distributions are introduced to simulate different states of the market, expressed as:

[0114]

[0115] Among them, p(x) represents the probability distribution of the observed data, K is the number of Gaussian distributions, corresponding to different states of the financial market (such as bull market, energy market, shock market, etc.), π k is the mixed weight of each Gaussian distribution, indicating the probability of the market being in a certain state, is a Gaussian distribution, μ k is the mean, ∑ k is the covariance matrix, which represents the characteristics of the market in this state;

[0116] S42, Dynamic Weight Adjustment: To adapt to the dynamic changes in the financial market, the mixed weight π of each Gaussian distribution is adjusted. k Dynamic adjustments are made, and the weights are dynamically updated based on the historical volatility of the market to better reflect the market status, expressed as:

[0117]

[0118] in, is the mixed weight of the kth market state at time t+1, is the mixed weight of the kth market state at time t, β is the adjustment coefficient, which controls the influence of volatility on weight update, M is the total number of volatility in different market states, v olatility k is the volatility of the kth market state;

[0119] S43, Dynamic update of covariance matrix of time series: The data of financial market has the characteristics of time series. The dynamic covariance matrix update mechanism is introduced. By continuously updating the covariance matrix, the risk transmission and market correlation changes of the financial market at different time points are captured, which is expressed as:

[0120]

[0121] in, is the covariance matrix of the kth market state at time t+1, is the covariance matrix of the k-th market state at time t after adjusting the historical volatility of the market, Cov(X t ,X t-1 ) is the covariance between the observations at time t and t-1, and α is a smoothing factor used to control the trade-off between current and historical covariance;

[0122] S44, introduction of market sentiment data: Market sentiment data captures the emotional fluctuations of market participants through sources such as social media and news. Market sentiment data is introduced as an additional input feature into the Gaussian mixture model, and the weight of the market state is adjusted in combination with subjective emotional fluctuations, expressed as:

[0123]

[0124] in, is the mixed weight of the kth market state at time t+1, is the mixed weight of the kth market state at time t after the dynamic covariance matrix is ​​updated, the weight at the current time point, γ is the sentiment adjustment coefficient, which controls the influence of market sentiment on weight adjustment, sentiment kis the sentiment data of the k-th market state. Sentiment data is usually obtained by quantifying the market sentiment fluctuations from data sources such as social media and news reports;

[0125] S45, Risk score calculation: Based on the volatility, sentiment data and weights of different market states, calculate the risk score R for each market state k , expressed as:

[0126] R k =π″′ k ·(volatiolity k +sentiment k );

[0127] Among them, R k is the risk score of the k-th market state, π″′ k is the mixed weight adjusted by sentiment data, volatility k is the volatility under market state k, sentiment k is the sentiment fluctuation under market state k;

[0128] Through the above content, combined with the market's multi-state modeling, dynamic weight adjustment, time series covariance matrix update and the introduction of market sentiment data, it is possible to more comprehensively capture the volatility and risks in the financial market, effectively adapt to the dynamic changes in the market, capture the risk transmission path under different market conditions, and provide more accurate risk scores by combining volatility and sentiment data, thereby enhancing the market's risk prediction capabilities and making risk assessment more flexible, comprehensive and accurate.

[0129] The risk prediction results output in S5 include:

[0130] S51, Risk score output: Calculate and output the risk score R for each market state based on the mixed weight, volatility and market sentiment data of different market states k ,The risk score reflects the comprehensive risk level of the market in its current state, and can help users understand the risk level of each market;

[0131] S52, Risk volatility trend prediction: Based on the covariance matrix and sentiment fluctuations of historical multi-dimensional data, the volatility trend of the financial market in the future is predicted through the autoregressive moving average model (ARMA). By analyzing market sentiment, volatility and historical trends, the future market volatility trend is output, including the trend signal of risk increase or decrease, which is expressed as:

[0132]

[0133] Among them, X tis the observed value at time t (such as market volatility or risk value), are the autoregressive coefficients, θ1,θ2,…,θ q is the moving average coefficient, ∈ t is the error term, representing random fluctuations;

[0134] Through the above content, risk scores and volatility trends of the financial market can be provided in a timely manner, helping users make more informed decisions under different market conditions. Combined with volatility and market sentiment data, the output risk score can accurately reflect the risk level of the market, and the prediction of risk volatility trends can provide early warning of market changes.

[0135] like Figure 2 As shown, a financial risk prediction system based on artificial intelligence is used to implement the above-mentioned financial risk prediction method based on artificial intelligence, including the following modules:

[0136] Data collection and preprocessing module: collects multi-dimensional data from multiple financial markets in real time, including historical transaction data, macroeconomic indicators, and market sentiment data, and cleans, denoises, and standardizes the collected data;

[0137] Cross-market financial product association network construction module: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products;

[0138] Risk transmission path identification module: Based on the cross-market financial product association network, it identifies the risk transmission path between different financial markets;

[0139] Risk analysis module: Combined with the identified risk transmission paths, the financial risk prediction model is used to analyze and predict the risk fluctuations in various financial markets;

[0140] Risk prediction result output module: outputs the risk score and fluctuation trend of each market.

[0141] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0142] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A financial risk prediction method based on artificial intelligence, characterized in that: The following steps are involved: S1, data collection and preprocessing: real-time collection of multi-dimensional data from multiple financial markets, including historical transaction data, macroeconomic indicators and market sentiment data, and preprocessing of the collected multi-dimensional data, including data cleaning, denoising and standardization; S2, Construction of cross-market financial product association network: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products; S3, risk transmission path identification: Based on the constructed cross-market financial product association network, the risk transmission path between financial markets is identified; S4, Risk Analysis: Combined with the identified risk transmission paths, the risk fluctuations in different financial markets are predicted through financial risk prediction models; S5, risk prediction result output: output the prediction results of different financial market risks, including the market risk score and risk fluctuation trend.

2. The method for predicting financial risks based on artificial intelligence according to claim 1, characterized in that: The data collection and preprocessing in S1 include: S11, data collection: real-time collection of multi-dimensional data from stock, bond and futures financial markets, including historical transaction data, macroeconomic indicators and market sentiment data; S12, data cleaning and denoising: clean the collected multi-dimensional data, use missing value interpolation to fill in the missing data, and use wavelet transform to remove noise signals in financial data; S13, data standardization: standardize the multidimensional data after cleaning and denoising, and use the z-score standardization method to normalize the data to a distribution with a mean of 0 and a standard deviation of 1.

3. The financial risk prediction method based on artificial intelligence according to claim 1, characterized in that: The construction of the cross-market financial product association network in S2 includes: S21, multi-dimensional data mapping and conversion: converting the collected multi-dimensional data into node attributes and edge weights; S22, graph structure construction of association network: Based on node and edge weights, a graph neural network is used to construct a cross-market financial product association network; S23, cross-market correlation analysis and optimization: Based on the graph structure of the constructed correlation network, analyze the correlation between financial markets, use the aggregation function in the graph neural network to summarize and update the neighborhood information of the nodes, identify high-risk correlation paths, and dynamically adjust the edge weights.

4. The method for predicting financial risks based on artificial intelligence according to claim 3, characterized in that: The multi-dimensional data mapping and conversion in S21 includes: S211, node attribute generation: map the collected multi-dimensional data into the attributes of each financial product node. Each node represents a financial product. The node attributes include the transaction volume T of the product. v 、Price Fluctuation v and volatility V v ; S212, edge weight calculation: The edge weights between nodes are calculated based on the transaction volume and price correlation between financial products. The edge weights reflect the strength of the association between two financial products.

5. The method for predicting financial risks based on artificial intelligence according to claim 4, characterized in that: The graph structure construction of the association network in S22 includes: S221, graph structure initialization: Based on the generated node attributes and edge weights, a graph structure G = (V, E) is constructed, where V is a node set representing different financial products, and E is an edge set representing the relationship between financial products. uv The weight of ∈E is determined by the edge weight, which represents the strength of the association between nodes u and v; S222, feature aggregation of graph neural network layer: Use the aggregation function in the graph neural network to update the features of each node, and set the feature vector of node v to At the kth layer, feature updating is achieved by aggregating the features of its neighboring nodes.

6. The method for predicting financial risks based on artificial intelligence according to claim 5, characterized in that: The cross-market correlation analysis and optimization in S23 includes: S231, node feature aggregation and update: Based on the constructed cross-market financial product association network, the neighborhood information of the node is aggregated and updated, that is, the feature vector of node v In the k+1th layer, by aggregating the features of its neighboring nodes renew; S232, high-risk association path identification: By analyzing the updated feature vectors of the nodes, we can identify the high-risk association paths across markets. Based on the strength of the association between financial products and market volatility, the high edge weights w on the paths are uv And the node volatility V v The superposition reflects a potential high-risk association; S233, Dynamic adjustment of edge weights: After identifying high-risk association paths, dynamically adjust edge weights w based on real-time market performance and new multi-dimensional data uv .

7. The method for predicting financial risks based on artificial intelligence according to claim 6, characterized in that: The risk transmission path identification in S3 includes: S31, Path Definition: Based on the constructed cross-market financial product association network, define the risk transmission path between financial products. The path consists of multiple connected nodes and edges. Each edge represents the strength of the association between different financial products. The node represents the financial product. The path length is determined by the degree of association between financial products and the transmission delay. S32, Risk transmission path weight calculation: After identifying the potential transmission path between markets, calculate the overall risk transmission strength of the path. The transmission strength is determined by the weight w of each side on the path. uv and the market volatility V of the node v The combined impact of S33, Identification of risk transmission direction: Determine the direction of risk transmission by combining historical multi-dimensional data and price fluctuations of the financial market. Through the correlation of edges and the market volatility information of nodes, identify the trend of risk transmission from high-volatility products to low-volatility products and form a transmission path.

8. The method for predicting financial risks based on artificial intelligence according to claim 1, characterized in that: The financial risk prediction model in S4 adopts a Gaussian mixture model, which includes: S41, multi-market state modeling: In view of the polymorphism of different products and markets in the financial market, multiple Gaussian distributions are introduced to simulate different market states; S42, Dynamic Weight Adjustment: To adapt to the dynamic changes in the financial market, the mixed weight π of each Gaussian distribution is adjusted. k Make dynamic adjustments and dynamically update weights based on the historical volatility of the market; S43, Dynamic update of covariance matrix of time series: The data of financial markets have the characteristics of time series. The dynamic covariance matrix update mechanism is introduced to capture the risk transmission and market correlation changes of financial markets at different time points by continuously updating the covariance matrix. S44, introduction of market sentiment data: Market sentiment data captures the emotional fluctuations of market participants through social media and news. Market sentiment data is introduced as an additional input feature into the Gaussian mixture model, and the weight of market status is adjusted in combination with subjective emotional fluctuations; S45, Risk score calculation: Based on the volatility, sentiment data and weights of different market states, calculate the risk score R for each market state k .

9. The method for predicting financial risks based on artificial intelligence according to claim 8, characterized in that: The risk prediction result output in S5 includes: S51, Risk score output: Calculate and output the risk score R for each market state based on the mixed weight, volatility and market sentiment data of different market states k , the risk score reflects the comprehensive risk level of the market in its current state; S52, Risk volatility trend prediction: Based on the covariance matrix and sentiment fluctuations of historical multi-dimensional data, the volatility trend of the financial market is predicted through the autoregressive moving average model. By analyzing market sentiment, volatility and historical trends, the future market volatility trend is output, including trend signals of rising or falling risks.

10. A financial risk prediction system based on artificial intelligence, used to implement a financial risk prediction method based on artificial intelligence as claimed in any one of claims 1 to 9, characterized in that: Includes the following modules: Data collection and preprocessing module: collects multi-dimensional data from multiple financial markets in real time, including historical transaction data, macroeconomic indicators, and market sentiment data, and cleans, denoises, and standardizes the collected data; Cross-market financial product association network construction module: Based on the collected multi-dimensional data, a cross-market financial product association network is constructed, and the correlation between financial markets is identified by analyzing the transaction volume, price fluctuations and volatility between different financial products; Risk transmission path identification module: Based on the cross-market financial product association network, it identifies the risk transmission path between different financial markets; Risk analysis module: Combined with the identified risk transmission paths, the financial risk prediction model is used to analyze and predict the risk fluctuations in various financial markets; Risk prediction result output module: outputs the risk score and fluctuation trend of each market.

Images