Method, device and equipment for identifying causal relationship between financial data and storage medium
By obtaining the real time series of financial variables and using mutual information and shadow manifold techniques to identify causal relationships between financial data, the problem of low accuracy in causal relationship identification in existing technologies has been solved, and highly accurate causal relationship judgment has been achieved in complex financial systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI UNIV
- Filing Date
- 2023-11-03
- Publication Date
- 2026-07-24
AI Technical Summary
Existing technologies are not very accurate in identifying causal relationships between time series data in financial systems, especially in complex and loosely coupled financial systems, where they are difficult to meet the requirements for separability.
By obtaining the real time series of the first and second financial variables, estimating and calculating their mutual information, comparing the magnitude of the mutual information, determining the dependent and outcome variables, calculating the weights using the shadow manifold and Euclidean distance, and constructing the optimal shadow manifold to improve estimation accuracy.
It can effectively identify causal relationships in complex nonlinear financial systems with high accuracy and strong applicability. It can identify the dependent variable and improve the accuracy of causal relationship identification.
Smart Images
Figure CN117453768B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of financial technology, specifically to methods, apparatus, devices, and storage media for identifying causal relationships between financial data. Background Technology
[0002] The complex scientific background causes the causal relationships between variables in financial systems to vary intermittently. For example, the causal relationships between variables may exhibit positive coupling, negative coupling, or no correlation at different time periods. Causality is mainly used for time series data, exploring whether time series X is meaningful in estimating another time series Y through regression and statistical tests, thereby constructing a causal relationship between X and Y. For example, when the US dollar index changes, does it lead to a significant change in oil prices? If so, a causal relationship can be considered between the US dollar and oil prices.
[0003] Granger causality represents a significant milestone in the study of causal relationships between variables. Its core concept is that if, after eliminating all potential causal factors from one variable, the estimability of another variable decreases, then the eliminated variable is called the "Granger cause," and the other is the "Granger outcome." However, a core assumption of this method is "separability," meaning that the cause and outcome can exist independently. In purely random and linear systems, Granger causality performs well because it satisfies the requirement of separability.
[0004] However, for deterministic systems, especially complex and loosely coupled financial systems, the accuracy of using Granger causality to determine the causal relationships between time series data in a financial system is not high because the system variable data are generally indivisible and do not meet the requirement of separability. Summary of the Invention
[0005] In view of this, the purpose of the present invention is to provide a method, apparatus, device and storage medium for identifying causal relationships between financial data, so as to solve the problem that the accuracy of identifying causal relationships between time series data in financial systems is not high in the prior art.
[0006] According to a first aspect of the present invention, a method for identifying causal relationships between financial data is provided, comprising: acquiring the real time series of a first financial variable and a second financial variable, wherein the real time series are the real values of the financial variables at different times;
[0007] The second financial variable is estimated based on the true time series of the first financial variable, and the first financial variable is estimated based on the true time series of the second financial variable, to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable. The estimated time series are the estimated values of the financial variable at different times.
[0008] The mutual information of the second financial variable is calculated based on the real time series and the estimated time series of the second financial variable, and the mutual information of the first financial variable is calculated based on the real time series and the estimated time series of the first financial variable.
[0009] Compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, then the first financial variable is determined to be the dependent variable of the second financial variable.
[0010] Preferably, estimating the second financial variable based on the true time series of the first financial variable includes:
[0011] Based on the real time series of the first financial variable and multiple preset spatial dimensions, construct multiple shadow manifolds corresponding to the first financial variable;
[0012] Select the optimal spatial dimension from the plurality of preset spatial dimensions, and use its corresponding shadow manifold as the optimal shadow manifold corresponding to the first financial variable;
[0013] The estimated time series of the second financial variable is calculated based on the optimal shadow manifold of the first financial variable.
[0014] Preferably, the step of calculating the estimated time series of the second financial variable based on the optimal shadow manifold of the first financial variable includes:
[0015] Identify all nodes in the optimal shadow manifold of the first financial variable;
[0016] Calculate the Euclidean distance between each node and all other nodes;
[0017] Select the N+1 nodes that are closest in Euclidean distance as the neighbor nodes of the node, where N is the optimal spatial dimension;
[0018] Based on the Euclidean distance between the node and its neighboring nodes, multiple weights corresponding to the node are calculated.
[0019] Based on the multiple weights of each node and the actual time series of the second financial variable corresponding to the neighboring nodes of that node, the estimated time series of the second financial variable is calculated.
[0020] Preferably, the step of constructing multiple shadow manifolds corresponding to the first financial variable based on the real time series of the first financial variable and multiple preset spatial dimensions includes:
[0021] The shadow manifold M corresponding to the first financial variable is calculated using the following formula. X :
[0022] x (t)=<X(t),X(t-τ),X(t-2τ),…,X(t-(E-1)τ)> ;
[0023] in x X(t) is the value of the shadow manifold corresponding to the first financial variable at time t, X(t) is the true value of the first financial variable at time t, τ is the preset lag time, and E is the preset spatial dimension.
[0024] Preferably, the step of calculating the mutual information of the second financial variable based on its historical time series and estimated time series includes:
[0025] Calculate the mutual information of the second financial variable using the following formula.
[0026]
[0027] in, yes The joint entropy of Y(t) and Y(t) Under the condition of Y(t) conditional entropy, In order to be in The conditional entropy of Y(t) under given conditions. Y(t) is the estimated time series of the second financial variable, and Y(t) is the actual time series of the second financial variable.
[0028] Preferably, the method further includes:
[0029] If the mutual information value of the first financial variable is less than the mutual information value of the second financial variable, the second financial variable is determined to be the dependent variable of the first financial variable.
[0030] According to a second aspect of the present invention, an apparatus for identifying causal relationships between financial data is provided, comprising:
[0031] The data acquisition module is used to acquire the real time series of the first financial variable and the real time series of the second financial variable, wherein the real time series are the real values of the financial variables at different times.
[0032] The data estimation module is used to estimate the second financial variable based on the real time series of the first financial variable, and to estimate the first financial variable based on the real time series of the second financial variable, so as to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable, wherein the estimated time series are the estimated values of the financial variable at different times;
[0033] The data calculation module is used to calculate the mutual information of the second financial variable based on the real time series and the estimated time series of the second financial variable, and to calculate the mutual information of the first financial variable based on the real time series and the estimated time series of the first financial variable.
[0034] The relationship identification module is used to compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, it is determined that the first financial variable is the dependent variable of the second financial variable.
[0035] According to a third aspect of the present invention, an apparatus for identifying causal relationships between financial data is provided, comprising:
[0036] Memory, on which executable programs are stored;
[0037] A processor for executing the executable program in the memory to implement the steps of any of the methods described above.
[0038] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a computer to perform the steps of any of the methods described above.
[0039] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:
[0040] By obtaining the true time series of the first financial variable and the true time series of the second financial variable, respectively, the second financial variable is estimated based on the true time series of the first financial variable, and the first financial variable is estimated based on the true time series of the second financial variable, resulting in estimated time series of the second financial variable and estimated time series of the first financial variable. The true time series represents the actual values of the financial variables at different times, and the estimated time series represents the estimated values of the financial variables at different times. Then, the mutual information of the second financial variable is calculated based on the true and estimated time series, and the mutual information of the first financial variable is calculated based on the true and estimated time series. The mutual information of the second financial variable is compared with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, the first financial variable is determined to be the dependent variable of the second financial variable. This method can detect causal relationships in complex nonlinear systems and can effectively determine the causal relationships between financial variables in complex financial systems, while also identifying the dependent variable. Because it uses a large amount of data for identification and judgment, its accuracy is high and its applicability is strong.
[0041] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description
[0042] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0043] Figure 1 This is a flowchart illustrating a method for identifying causal relationships between financial data according to an exemplary embodiment;
[0044] Figure 2 This is a schematic diagram illustrating a causal relationship between financial data according to yet another exemplary embodiment;
[0045] Figure 3 This is a block diagram illustrating an apparatus for identifying causal relationships between financial data according to an exemplary embodiment. Detailed Implementation
[0046] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.
[0047] This invention provides a method for identifying causal relationships between financial data, see [link to relevant documentation]. Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying causal relationships between financial data according to an exemplary embodiment. The method includes:
[0048] Step S11: Obtain the real time series of the first financial variable and the real time series of the second financial variable respectively. The real time series are the real values of the financial variables at different times.
[0049] Step S12: Estimate the second financial variable based on the true time series of the first financial variable, and estimate the first financial variable based on the true time series of the second financial variable, to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable. The estimated time series are the estimated values of the financial variable at different times.
[0050] Step S13: Calculate the mutual information of the second financial variable based on the real time series and the estimated time series of the second financial variable; calculate the mutual information of the first financial variable based on the real time series and the estimated time series of the first financial variable.
[0051] Step S14: Compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, determine that the first financial variable is the dependent variable of the second financial variable.
[0052] Specifically, the true time series of the first financial variable and the true time series of the second financial variable are obtained. The true time series are the true values of the financial variables at different times. That is, the true time series of the first financial variable is the true value of the first financial variable at different times, and the true time series of the second financial variable is the true value of the second financial variable at different times.
[0053] In one example, time series data of the closing prices of gold futures and US dollar index futures on the New York Stock Exchange from January 2022 to October 2023 are obtained, where the time can be daily. The returns of the closing prices of gold futures and US dollar index futures are calculated. The daily returns of the closing prices of gold futures are used as the real time series of the first financial variable, and the daily returns of the closing prices of US dollar index futures are used as the real time series of the second financial variable. The causal relationship between the returns of the closing prices of gold futures and US dollar index futures is then calculated.
[0054] Specifically, based on the true time series of the first financial variable, the second financial variable is estimated to obtain the estimated time series of the second financial variable. The estimated time series is the estimated value of the financial variable at different times, and the estimated time series of the second financial variable is the estimated value of the second financial variable at different times.
[0055] Based on the true time series of the second financial variable, the first financial variable is estimated to obtain the estimated time series of the first financial variable. The estimated time series of the first financial variable is the estimated value of the first financial variable at different times.
[0056] By estimating the estimated values of another financial variable at different times based on the true values of one financial variable at different times, and by analyzing the relationship between the true values and the estimated values, we can investigate whether there is a causal relationship between the two financial variables.
[0057] Continuing with the example above, based on the true value of the daily closing price return of gold futures, the daily closing price return of the US dollar index futures is estimated, resulting in an estimated value for the daily closing price return of the US dollar index futures, thus forming a time series of estimated returns for the closing price of the US dollar index futures.
[0058] Specifically, based on the true and estimated time series of the second financial variable, the mutual information between the true and estimated values of the second financial variable is calculated. This mutual information represents the linear or non-linear dependency between two financial variables. A higher mutual information value indicates a stronger correlation and dependency between the estimated and true values of the financial variables, making the estimation more reliable. Furthermore, if a causal relationship exists between the two financial variables, the mutual information value will converge. A higher mutual information value between the estimated and true values of the second financial variable indicates a stronger driving force of the second financial variable on the first financial variable; changes in the second financial variable are more likely to cause changes in the first financial variable, and the second financial variable is more likely to be the dependent variable of the first financial variable.
[0059] Similarly, based on the true and estimated time series of the first financial variable, the mutual information between the estimated and true values of the first financial variable is calculated. A higher mutual information value for the first financial variable indicates a stronger driving force of the first financial variable on the second financial variable, meaning that changes in the first financial variable are more likely to cause changes in the second financial variable, and the first financial variable is more likely to be the dependent variable of the second financial variable.
[0060] As the length of the time series increases, when there is a causal relationship between the first financial variable and the second financial variable, the mutual information between the first financial variable and the second financial variable will gradually converge to a specific value.
[0061] like Figure 2 As shown, Figure 2 This is a schematic diagram illustrating a causal relationship between financial data according to another exemplary embodiment. When the mutual information of the first financial variable and the mutual information of the second financial variable gradually converge to a specific value, after a certain time, as the time series increases, the magnitudes of the mutual information of the second financial variable and the mutual information of the first financial variable are compared, as shown by the convergence curve in the figure. When the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, it is considered that the driving force of the first financial variable on the second financial variable is stronger than the driving force of the second financial variable on the first financial variable, and the first financial variable is determined to be the dependent variable of the second financial variable.
[0062] It is understood that the technical solution provided in this embodiment obtains the true time series of the first financial variable and the true time series of the second financial variable respectively. Based on the true time series of the first financial variable, the second financial variable is estimated, and based on the true time series of the second financial variable, the first financial variable is estimated, resulting in the estimated time series of the second financial variable and the estimated time series of the first financial variable. The true time series represents the true value of the financial variable at different times, and the estimated time series represents the estimated value of the financial variable at different times. Then, based on the true time series and the estimated time series of the second financial variable, the mutual information of the second financial variable is calculated. Based on the true time series and the estimated time series of the first financial variable, the mutual information of the first financial variable is calculated. The mutual information of the second financial variable is compared with the mutual information of the first financial variable. If the value of the first financial variable is greater than the value of the mutual information of the second financial variable, the first financial variable is determined to be the dependent variable of the second financial variable. The above method can detect causal relationships in complex nonlinear systems and can effectively determine the causal relationships between financial variables in complex financial systems, while identifying the dependent variable. Because it uses a large amount of data for identification and judgment, its accuracy is high and its applicability is very strong.
[0063] Preferably, in step S12, estimating the second financial variable based on the true time series of the first financial variable includes:
[0064] Step S121: Based on the real time series of the first financial variable and multiple preset spatial dimensions, construct multiple shadow manifolds corresponding to the first financial variable;
[0065] Step S122: Select the optimal spatial dimension from the plurality of preset spatial dimensions, and use its corresponding shadow manifold as the optimal shadow manifold corresponding to the first financial variable;
[0066] Step S123: Calculate the estimated time series of the second financial variable based on the optimal shadow manifold of the first financial variable.
[0067] Specifically, when estimating the second financial variable based on the true time series of the first financial variable, the following steps are included:
[0068] Multiple spatial dimensions are preset, typically any one of 3, 4, 5, or 6. For any preset spatial dimension, a shadow manifold corresponding to the first financial variable is constructed based on the true time series of the first financial variable. The method for constructing the shadow manifold can be Tukens' theorem. Different preset spatial dimensions result in different shadow manifolds.
[0069] Because the spatial dimension of the shadow manifold is affected by factors such as system complexity and time series length, choosing an excessively large spatial dimension may lead to redundant information in the reconstructed space, increasing the complexity of the analysis. Conversely, choosing an excessively small spatial dimension may fail to capture all the feature information of the original space, affecting the estimated time series of the second financial variable and reducing the recognition accuracy. Selecting the optimal spatial dimension from multiple preset spatial dimensions and using its corresponding shadow manifold as the optimal shadow manifold for the first financial variable further increases the accuracy of the estimated time series of the second financial variable and improves the final causal relationship recognition accuracy.
[0070] Based on the optimal shadow manifold of the first financial variable selected above, the second financial variable is estimated, and the estimated time series of the second financial variable is calculated.
[0071] Preferably, in step S123, calculating the estimated time series of the second financial variable based on the optimal shadow manifold of the first financial variable includes:
[0072] Identify all nodes in the optimal shadow manifold of the first financial variable;
[0073] Calculate the Euclidean distance between each node and all other nodes;
[0074] Select the N+1 nodes that are closest in Euclidean distance as the neighbor nodes of the node, where N is the optimal spatial dimension;
[0075] Based on the Euclidean distance between the node and its neighboring nodes, multiple weights corresponding to the node are calculated.
[0076] Based on the multiple weights of each node and the actual time series of the second financial variable corresponding to the neighboring nodes of that node, the estimated time series of the second financial variable is calculated.
[0077] Specifically, when estimating the second financial variable, the optimal shadow manifold of the first financial variable corresponding to the optimal spatial dimension first determines all nodes in the optimal shadow manifold of the first financial variable, and each node represents the true value of the first financial variable at a given time.
[0078] For any given node, calculate the Euclidean distance between that node and all other nodes. The method for calculating the Euclidean distance is existing technology and will not be described in detail in this invention. For all other nodes, select the N+1 nodes with the closest Euclidean distance as the node's neighbor nodes, where N is the optimal spatial dimension.
[0079] For a given node and its multiple neighboring nodes, the weights between the node and each of its neighbors are calculated based on the Euclidean distance between the node and each of its neighbors. Each node corresponds to one weight with each of its neighbors, and thus, a node has weights from multiple neighboring nodes. Each neighboring node has a corresponding true value in the real time series of the second financial variable.
[0080] In a specific example, for the shadow manifold M X Its i-th neighbor node is denoted as x (t i Find its N+1 neighbor nodes in M. Y Mapping points on y (t1), y (t2),…, y (t E+1 ), which represents the true value of the second financial variable.
[0081] We perform a weighted summation based on the weight of each neighboring node and the real time series of the second financial variable corresponding to each neighboring node to obtain the estimated value of the corresponding second financial variable.
[0082] For the shadow manifold corresponding to the first financial variable, the estimated time series of the second financial variable is calculated. The estimated value of the second financial variable... Calculate using the following formula:
[0083]
[0084] Among them, w i Based on M X The weight of the i-th neighbor node.
[0085] w i The calculation formula is as follows:
[0086] w i =ui / Σu j j = 1…E+1
[0087] u i =exp{-d[ x (t), x (t i )] / d[ x (t), x (t1)]},
[0088] Where, d[ x (t), x (t i )]for x (t) and x (t i The Euclidean distance between ) and d[ x (t), x (t1)] is x (t) and x The Euclidean distance between (t1) and (t1).
[0089] Using the method described above, the estimated value of the second financial variable corresponding to each node is calculated, and the estimated time series of the second financial variable is obtained.
[0090] To determine the optimal spatial dimension, cross-validation is employed. For the multiple preset spatial dimensions, corresponding shadow manifolds are constructed. Following the same method, Euclidean distance is used to calculate the E+1 neighbor nodes of each node, as well as the Euclidean distance between them, where E is the preset spatial dimension. Weights are calculated based on the Euclidean distance, and the weighted average of the neighbor nodes is used as the corresponding estimate. Finally, the overall estimation error is calculated using the minimum mean square error method. For all preset spatial dimensions, the preset spatial dimension with the smallest overall estimation error is selected as the optimal spatial dimension, and the corresponding shadow manifold is the optimal shadow manifold.
[0091] In a specific example, the optimal spatial dimension is 4, so each node has 5 neighboring nodes.
[0092] Preferably, in step S121, constructing multiple shadow manifolds corresponding to the first financial variable based on the real time series of the first financial variable and multiple preset spatial dimensions includes:
[0093] The shadow manifold M corresponding to the first financial variable is calculated using the following formula. X :
[0094] x (t)=<X(t),X(t-τ),X(t-2τ),…,X(t-(E-1)τ)> ;
[0095] inx X(t) is the value of the shadow manifold corresponding to the first financial variable at time t, X(t) is the true value of the first financial variable at time t, τ is the preset lag time, and E is the preset spatial dimension.
[0096] Specifically, when constructing multiple shadow manifolds corresponding to the first financial variable based on the real time series of the first financial variable and multiple preset spatial dimensions, the process is as follows:
[0097] The shadow manifold M corresponding to the first financial variable is calculated using the following formula. X :
[0098] x (t)=<X(t),X(t-τ),X(t-2τ),…,X(t-(E-1)τ)> ;
[0099] in x X(t) is the value of the shadow manifold corresponding to the first financial variable at time t, X(t) is the true value of the first financial variable at time t, τ is the preset lag time, and E is the preset spatial dimension.
[0100] Here, the optimal shadow play manifold, determined by the optimal spatial dimension, can be determined. All of them... x (t) forms a shadow manifold M corresponding to the first financial variable. X .
[0101] Preferably, in step S13, calculating the mutual information of the second financial variable based on its historical time series and estimated time series includes:
[0102] Calculate the mutual information of the second financial variable using the following formula.
[0103] in, yes The joint entropy of Y(t) and Y(t) Under the condition of Y(t) conditional entropy, In order to be in The conditional entropy of Y(t) under given conditions. Y(t) is the estimated time series of the second financial variable, and Y(t) is the actual time series of the second financial variable.
[0104] Specifically, based on the historical and estimated time series of the second financial variable, the mutual information of the second financial variable is calculated, including:
[0105] Calculate the mutual information of the second financial variable using the following formula.
[0106]
[0107] in, yes The joint entropy of Y(t) and Y(t) Under the condition of Y(t) conditional entropy, In order to be in The conditional entropy of Y(t) under given conditions. Y(t) is the estimated time series of the second financial variable, and Y(t) is the actual time series of the second financial variable.
[0108] It should be noted that, using the same method described above, the shadow manifold of the second financial variable can be calculated, as can the estimated time series of the first financial variable and the mutual information of the first financial variable.
[0109] Preferably, the method further includes:
[0110] If the mutual information value of the first financial variable is less than the mutual information value of the second financial variable, the second financial variable is determined to be the dependent variable of the first financial variable.
[0111] Specifically, the mutual information values of the first financial variable and the second financial variable are compared. When the mutual information value of the first financial variable is less than the mutual information value of the second financial variable, the driving force of the second financial variable on the first financial variable is stronger than the driving force of the first financial variable on the second financial variable, and the second financial variable is determined to be the dependent variable of the first financial variable.
[0112] It should be noted that if the mutual information of the second financial variable and the mutual information of the first financial variable do not converge, then there is no causal relationship between the first financial variable and the second financial variable.
[0113] It is understood that the technical solution provided in this embodiment obtains the true time series of the first financial variable and the true time series of the second financial variable respectively. Based on the true time series of the first financial variable, the second financial variable is estimated, and based on the true time series of the second financial variable, the first financial variable is estimated, resulting in the estimated time series of the second financial variable and the estimated time series of the first financial variable. The true time series represents the true value of the financial variable at different times, and the estimated time series represents the estimated value of the financial variable at different times. Then, based on the true time series and the estimated time series of the second financial variable, the mutual information of the second financial variable is calculated. Based on the true time series and the estimated time series of the first financial variable, the mutual information of the first financial variable is calculated. The mutual information of the second financial variable is compared with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, the first financial variable is determined to be the dependent variable of the second financial variable. The above method can detect causal relationships in complex nonlinear systems and can effectively determine the causal relationships between financial variables in complex financial systems, while identifying the dependent variable. Because it uses a large amount of data for identification and judgment, its accuracy is high and its applicability is very strong.
[0114] This invention provides a device for identifying causal relationships between financial data, see [link to device]. Figure 3 , Figure 3 This is a block diagram illustrating an apparatus for identifying causal relationships between financial data according to an exemplary embodiment, comprising:
[0115] Data acquisition module 31 is used to acquire the real time series of the first financial variable and the real time series of the second financial variable respectively, wherein the real time series are the real values of the financial variables at different times;
[0116] The data estimation module 32 is used to estimate the second financial variable based on the real time series of the first financial variable, and to estimate the first financial variable based on the real time series of the second financial variable, so as to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable, wherein the estimated time series are the estimated values of the financial variable at different times.
[0117] The data calculation module 33 is used to calculate the mutual information of the second financial variable based on the real time series and the estimated time series of the second financial variable, and to calculate the mutual information of the first financial variable based on the real time series and the estimated time series of the first financial variable.
[0118] The relationship identification module 34 is used to compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, it is determined that the first financial variable is the dependent variable of the second financial variable.
[0119] Preferably, estimating the second financial variable based on the true time series of the first financial variable includes:
[0120] Based on the real time series of the first financial variable and multiple preset spatial dimensions, construct multiple shadow manifolds corresponding to the first financial variable;
[0121] Select the optimal spatial dimension from the plurality of preset spatial dimensions, and use its corresponding shadow manifold as the optimal shadow manifold corresponding to the first financial variable;
[0122] The estimated time series of the second financial variable is calculated based on the optimal shadow manifold of the first financial variable.
[0123] Preferably, the step of calculating the estimated time series of the second financial variable based on the optimal shadow manifold of the first financial variable includes:
[0124] Identify all nodes in the optimal shadow manifold of the first financial variable;
[0125] Calculate the Euclidean distance between each node and all other nodes;
[0126] Select the N+1 nodes that are closest in Euclidean distance as the neighbor nodes of the node, where N is the optimal spatial dimension;
[0127] Based on the Euclidean distance between the node and its neighboring nodes, multiple weights corresponding to the node are calculated.
[0128] Based on the multiple weights of each node and the actual time series of the second financial variable corresponding to the neighboring nodes of that node, the estimated time series of the second financial variable is calculated.
[0129] Preferably, the step of constructing multiple shadow manifolds corresponding to the first financial variable based on the real time series of the first financial variable and multiple preset spatial dimensions includes:
[0130] The shadow manifold M corresponding to the first financial variable is calculated using the following formula. X :
[0131] x (t)=<X(t),X(t-τ),X(t-2τ),…,X(t-(E-1)τ)> ;
[0132] in xX(t) is the value of the shadow manifold corresponding to the first financial variable at time t, X(t) is the true value of the first financial variable at time t, τ is the preset lag time, and E is the preset spatial dimension.
[0133] Preferably, the step of calculating the mutual information of the second financial variable based on its historical time series and estimated time series includes:
[0134] Calculate the mutual information of the second financial variable using the following formula.
[0135] in, yes The joint entropy of Y(t) and Y(t) Under the condition of Y(t) conditional entropy, In order to be in The conditional entropy of Y(t) under given conditions. Y(t) is the estimated time series of the second financial variable, and Y(t) is the actual time series of the second financial variable.
[0136] Preferably, the device further includes:
[0137] If the mutual information value of the first financial variable is less than the mutual information value of the second financial variable, the second financial variable is determined to be the dependent variable of the first financial variable.
[0138] It is understood that the technical solution provided in this embodiment, by obtaining the true time series of the first financial variable and the true time series of the second financial variable respectively, as mentioned in the above embodiment, estimating the second financial variable based on the true time series of the first financial variable, and estimating the first financial variable based on the true time series of the second financial variable, obtains the estimated time series of the second financial variable and the estimated time series of the first financial variable. The true time series represents the true value of the financial variable at different times, and the estimated time series represents the estimated value of the financial variable at different times. Then, based on the true time series and the estimated time series of the second financial variable, the mutual information of the second financial variable is calculated. Based on the true time series and the estimated time series of the first financial variable, the mutual information of the first financial variable is calculated. The mutual information of the second financial variable is compared with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, the first financial variable is determined to be the dependent variable of the second financial variable. This method can detect causal relationships in complex nonlinear systems and can effectively determine the causal relationships between financial variables in complex financial systems, while also identifying the dependent variable. Because it uses a large amount of data for identification and judgment, its accuracy is high and its applicability is very strong.
[0139] The present invention also provides a device for identifying causal relationships between financial data, comprising:
[0140] Memory, on which executable programs are stored;
[0141] A processor for executing the executable program in the memory to implement the steps of any of the methods described above.
[0142] Furthermore, the present invention also provides a computer-readable storage medium storing computer instructions for causing a computer to perform the steps of any of the methods described above. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.
[0143] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.
[0144] It should be noted that in the description of this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means at least two.
[0145] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.
[0146] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0147] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0148] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0149] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.
[0150] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0151] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for identifying causal relationships between financial data, characterized in that, include: Obtain the true time series of the first financial variable and the second financial variable respectively, wherein the true time series are the true values of the financial variables at different times; The second financial variable is estimated based on the true time series of the first financial variable, and the first financial variable is estimated based on the true time series of the second financial variable, to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable. The estimated time series are the estimated values of the financial variable at different times. The mutual information of the second financial variable is calculated based on the real time series and the estimated time series of the second financial variable, and the mutual information of the first financial variable is calculated based on the real time series and the estimated time series of the first financial variable. Compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, determine that the first financial variable is the dependent variable of the second financial variable. The estimation of the second financial variable based on the true time series of the first financial variable includes: Based on the real time series of the first financial variable and multiple preset spatial dimensions, construct multiple shadow manifolds corresponding to the first financial variable; Select the optimal spatial dimension from the plurality of preset spatial dimensions, and use its corresponding shadow manifold as the optimal shadow manifold corresponding to the first financial variable; Based on the optimal shadow manifold of the first financial variable, the estimated time series of the second financial variable is calculated; The step of constructing multiple shadow manifolds corresponding to the first financial variable based on the real time series of the first financial variable and multiple preset spatial dimensions includes: The shadow manifold corresponding to the first financial variable is calculated using the following formula. : ； in Let t be the value of the shadow manifold corresponding to the first financial variable at time t. Let be the true value of the first financial variable at time t. E represents the preset lag time and the preset spatial dimension.
2. The method according to claim 1, characterized in that, The step of calculating the estimated time series of the second financial variable based on the optimal shadow manifold of the first financial variable includes: Identify all nodes in the optimal shadow manifold of the first financial variable; Calculate the Euclidean distance between each node and all other nodes; Select the N+1 nodes that are closest in Euclidean distance as the neighbor nodes of the node, where N is the optimal spatial dimension; Based on the Euclidean distance between the node and its neighboring nodes, multiple weights corresponding to the node are calculated. Based on the multiple weights of each node and the actual time series of the second financial variable corresponding to the neighboring nodes of that node, the estimated time series of the second financial variable is calculated.
3. The method according to claim 1, characterized in that, The step of calculating the mutual information of the second financial variable based on its historical time series and estimated time series includes: Calculate the mutual information of the second financial variable using the following formula. : in, yes and joint entropy, In order to be in under conditions conditional entropy, In order to be in under conditions conditional entropy, It is the estimated time series of the second financial variable. It is the true time series of the second financial variable.
4. The method according to claim 1, characterized in that, The method further includes: If the mutual information value of the first financial variable is less than the mutual information value of the second financial variable, the second financial variable is determined to be the dependent variable of the first financial variable.
5. A device for identifying causal relationships between financial data, characterized in that, A method for identifying causal relationships between financial data as described in any one of claims 1 to 4, wherein the apparatus for identifying causal relationships between financial data comprises: The data acquisition module is used to acquire the real time series of the first financial variable and the real time series of the second financial variable, wherein the real time series are the real values of the financial variables at different times. The data estimation module is used to estimate the second financial variable based on the real time series of the first financial variable, and to estimate the first financial variable based on the real time series of the second financial variable, so as to obtain the estimated time series of the second financial variable and the estimated time series of the first financial variable, wherein the estimated time series are the estimated values of the financial variable at different times; The data calculation module is used to calculate the mutual information of the second financial variable based on the real time series and the estimated time series of the second financial variable, and to calculate the mutual information of the first financial variable based on the real time series and the estimated time series of the first financial variable. The relationship identification module is used to compare the mutual information of the second financial variable with the mutual information of the first financial variable. If the value of the mutual information of the first financial variable is greater than the value of the mutual information of the second financial variable, it is determined that the first financial variable is the dependent variable of the second financial variable.
6. A device for identifying causal relationships between financial data, characterized in that, include: Memory, on which executable programs are stored; A processor for executing the executable program in the memory to implement the steps of the method according to any one of claims 1 to 4.
7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the steps of the method according to any one of claims 1 to 4.