Credit monitoring assessment method and system based on market risk quantitative analysis

Abstract

The invention relates to the technical field of credit management, in particular to a credit monitoring and checking method and system based on market risk quantitative analysis. According to the credit monitoring and assessment method based on market risk quantitative analysis, the potential loss of financial assets in a user-defined time period is evaluated, then a credit risk rating model is constructed, and the credit risk grade of an enterprise is predicted; then constructing a dishonesty probability model, and predicting the dishonesty probability of the enterprise; and finally, step S4, generating a credit detection report based on the prediction results of the maximum loss amount VaR, the enterprise credit risk level and the enterprise dishonesty probability that the enterprise will face in the customized time period. The credit monitoring assessment method and system based on market risk quantitative analysis not only improve the scientificity and accuracy of credit monitoring, but also enhance the ability of a management department in market supervision and risk prevention.

Description

Technical Field

[0001] The present invention relates to the technical field of credit management, and particularly to a credit monitoring and assessment method and system based on market risk quantitative analysis. Background Art

[0002] Traditional credit monitoring can only collect enterprise credit data and conduct statistical analysis in dimensions such as industry, dishonesty, and time to form a report.

[0003] The dependence on historical data in the traditional method, the lag in market changes, the emotional cognitive bias, and the lack of real-time monitoring ability lead to a relatively subjective decision-making process, making it difficult to ensure the scientificity and accuracy of credit monitoring, and thus unable to enhance the ability of the management department in market supervision and risk prevention.

[0004] In view of the above problems, the present invention proposes a credit monitoring and assessment method and system based on market risk quantitative analysis. Summary of the Invention

[0005] In order to make up for the defects of the prior art, the present invention provides a simple and efficient credit monitoring and assessment method and system based on market risk quantitative analysis.

[0006] The present invention is realized by the following technical solutions:

[0007] A credit monitoring and assessment method based on market risk quantitative analysis, characterized in that it includes the following steps:

[0008] Step S1, evaluate the potential loss of financial assets within a custom time period;

[0009] Adopt the historical simulation method of Value at Risk (VaR) as a risk management tool, and use historical enterprise data to simulate and evaluate the maximum loss amount VaR that a financial asset or investment portfolio will face within a custom time period. Without making assumptions about the distribution of returns, it can more accurately reflect the actual situation of the enterprise;

[0010] In the step S1, the processing process of the historical simulation method of Value at Risk (VaR) is as follows:

[0011] Step S1.1, calculate the return rate r of the asset at a certain moment t t , and the calculation formula is as follows:

[0012]

[0013] where Pt is the market value at time t, and Pt-1 is the market value at time t-1;

[0014] Step S1.2: Use the quicksort method to sort the yields from low to high to obtain a sample with a total of N;

[0015] Step S1.3: Determine the confidence level α, and determine the position of the maximum loss amount VaR in the sample according to the total number of samples N. The calculation formula is as follows:

[0016] VaR position = N × (1 - α)

[0017] The confidence level α is set to 95% or 99%;

[0018] Step S1.4: Among the sorted yields, select the yield at the position of the maximum loss amount VaR as the estimated maximum loss amount VaR. The formula is as follows:

[0019] VaR = -r (N×(1-α))

[0020] In the formula, the negative sign represents the loss caused by the risk, and r (1) , r (2) , r (3) ,..., r (N) are all the sorted yields in the sample;

[0021] Step S2: Build a credit risk rating model to predict the credit risk level of enterprises;

[0022] Step S2.1: Sort out the detection indicators according to the urban credit detection rules, and configure the corresponding enterprise data for the detection indicators;

[0023] Step S2.2: Use a machine learning statistical model to create a decision tree model, and customize the hyperparameters of the decision tree according to the actual situation, including the maximum depth of the tree and the minimum number of sample splits, to prevent overfitting;

[0024] In step S2, the decision tree model is implemented using the scikit-learn library in Python.

[0025] Step S2.3: Use the sample set composed of historical data of enterprise credit risk levels to train the decision tree model, and evaluate the performance of the trained decision tree model to determine its effectiveness;

[0026] During the evaluation, calculate the accuracy, recall rate, F1-score, and ROC curve of the decision tree model to comprehensively evaluate the performance of the decision tree model.

[0027] Step S2.3: Based on the decision tree model after passing the evaluation and the enterprise data crawled in real time, make predictions to generate the credit risk level of the enterprise;

[0028] Step S3: Build a default probability model to predict the default probability of an enterprise;

[0029] Step S3.1: Use the logistic regression algorithm to build a default probability model for evaluating the default probability P of an enterprise within a custom time period. The formula for the logistic regression algorithm is as follows:

[0030]

[0031] where Y is the default event, 1 indicates default, 0 indicates non - default, X1, X2,..., X m are the characteristic variables composed of credit data, and β0, β1, β2, β m are the regression coefficients corresponding to various types of credit data;

[0032] In step S3, use the scikit - learn library in Python for modeling to build a default probability model.

[0033] Step S3.2: Use the training set constructed from the historical data of default enterprises to fit the default probability model;

[0034] Step S3.3: Evaluate the performance of the fitted default probability model to determine its effectiveness. The evaluation metrics include accuracy, recall, precision, F1 - score, ROC curve, and AUC value;

[0035] Step S3.4: Based on the qualified default probability model after evaluation and the real - time crawled enterprise data, predict the default probability of the enterprise.

[0036] Step S4: Generate a credit detection report based on the maximum loss amount VaR that the enterprise will face within a custom time period, the enterprise credit risk level, and the prediction results of the enterprise default probability.

[0037] A credit monitoring and assessment system based on market risk quantitative analysis for implementing the above method, including a risk management tool, a credit risk rating module, and a default probability prediction module;

[0038] The risk management tool is responsible for using the historical simulation method of Value at Risk (VaR) and using historical enterprise data to simulate and evaluate the maximum loss amount VaR that a financial asset or investment portfolio will face within a custom time period;

[0039] The credit risk rating module is responsible for building a credit risk rating model, training the model using a sample set composed of historical data of enterprise credit risk levels, evaluating the performance of the trained model, and making predictions based on the qualified model and real - time crawled enterprise data to generate the enterprise credit risk level;

[0040] The default probability prediction module is responsible for constructing a default probability model, fitting the default probability model using a training set constructed from historical data of defaulting enterprises, evaluating the performance of the fitted default probability model, and predicting the default probability of enterprises based on the qualified default probability model and real-time crawled enterprise data.

[0041] A credit monitoring and assessment device based on quantitative analysis of market risks, characterized by: including a memory and a processor; the memory is used to store a computer program, and the processor is used to implement the above method steps when executing the computer program.

[0042] A readable storage medium, characterized by: a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the above method steps are implemented.

[0043] The beneficial effects of the present invention are: the credit monitoring and assessment method and system based on quantitative analysis of market risks not only improve the scientificity and accuracy of credit monitoring, but also enhance the capabilities of management departments in market supervision and risk prevention. Brief Description of the Drawings

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0045] Appendix Figure 1 It is a schematic diagram of the credit monitoring and assessment method based on quantitative analysis of market risks of the present invention. Detailed Embodiments

[0046] In order to enable those skilled in the art of the present technology to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in combination with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0047] The credit monitoring and assessment method based on quantitative analysis of market risks includes the following steps:

[0048] Step S1: Evaluate the potential losses of financial assets within a custom time period;

[0049] Adopt the historical simulation method of Value at Risk (VaR) as a risk management tool, and use historical enterprise data to simulate and evaluate the maximum loss amount VaR that a financial asset or investment portfolio will face within a custom time period. Without making assumptions about the distribution of returns, it can more accurately reflect the actual situation of the enterprise;

[0050] In the step S1, the processing process of the historical simulation method of Value at Risk (VaR) is as follows:

[0051] Step S1.1: Calculate the return rate r of the asset at a certain moment t t , and the calculation formula is as follows:

[0052]

[0053] Among them, Pt is the market value at time t, and Pt-1 is the market value at time t-1;

[0054] Step S1.2: Use the quicksort method to sort the return rates from low to high to obtain a total of N samples;

[0055] Step S1.3: Determine the confidence level α, and determine the position of the maximum loss amount VaR in the sample according to the total number of samples N. The calculation formula is as follows:

[0056] VaR position = N × (1 - α)

[0057] The confidence level α is set to 95% or 99%;

[0058] Step S1.4: In the sorted return rates, select the return rate at the position of the maximum loss amount VaR as the estimated maximum loss amount VaR. The formula is as follows:

[0059] VaR = -r (N×(1-α))

[0060] In the formula, the negative sign represents the loss caused by the risk, and r (1) , r (2) , r (3) ,..., r (N) are all the sorted return rates in the sample;

[0061] Step S2: Build a credit risk rating model to predict the credit risk level of the enterprise;

[0062] Step S2.1: Sort out the detection indicators according to the urban credit detection rules, and configure the corresponding enterprise data for the detection indicators;

[0063] Step S2.2: Create a decision tree model using a machine learning statistical model, and customize the hyperparameters of the decision tree according to the actual situation, including the maximum depth of the tree and the minimum number of samples for splitting, to prevent overfitting;

[0064] In step S2, the scikit-learn library of Python is used to implement the decision tree model.

[0065] from sklearn.tree import DecisionTreeClassifier

[0066] from sklearn.model_selection import train_test_split

[0067] # Assume X is the feature set and y is the label (credit rating result)

[0068] X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.2,random_state=42)

[0069] # Create a decision tree model

[0070] model=DecisionTreeClassifier(max_depth=5)# Adjust parameters as needed

[0071] model.fit(X_train,y_train)

[0072] Step S2.3: Use the sample set composed of historical data of enterprise credit risk levels to train the decision tree model, and evaluate the performance of the trained decision tree model to determine its effectiveness;

[0073] During evaluation, calculate the accuracy, recall, F1-score, and ROC curve of the decision tree model to comprehensively evaluate the performance of the decision tree model.

[0074] from sklearn.metrics import accuracy_score,classification_report

[0075] y_pred=model.predict(X_test)

[0076] print("Accuracy:",accuracy_score(y_test,y_pred))

[0077] print(classification_report(y_test, y_pred))

[0078] Step S2.3: Based on the qualified decision tree model and the real-time crawled enterprise data, make predictions to generate the credit risk level of the enterprise;

[0079] Step S3: Build a default probability model to predict the default probability of the enterprise;

[0080] Step S3.1: Use the logistic regression algorithm to build a default probability model for evaluating the default probability P of the enterprise within a custom time period; the logistic regression algorithm formula is as follows:

[0081]

[0082] Where Y is the default event, 1 represents default, 0 represents non-default, X1, X2,..., X m are the characteristic variables composed of credit data, and β0, β1, β2, β m are the regression coefficients corresponding to various types of credit data;

[0083] In the said Step S3, use the scikit-learn library in Python for modeling to build a default probability model.

[0084] Step S3.2: Use the training set constructed from the historical data of defaulting enterprises to fit the default probability model;

[0085] import pandas as pd

[0086] from sklearn.model_selection import train_test_split

[0087] from sklearn.linear_model import LogisticRegression

[0088] from sklearn.preprocessing import StandardScaler

[0089] # Assume df is a data frame containing features and target variables

[0090] X = df.drop('target', axis = 1) # Features

[0091] y = df['target'] # Target variable

[0092] # Data Partitioning

[0093] X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 42)

[0094] # Standardize Features

[0095] scaler = StandardScaler()

[0096] X_train_scaled = scaler.fit_transform(X_train)

[0097] X_test_scaled = scaler.transform(X_test)

[0098] # Create and Fit Logistic Regression Model

[0099] model = LogisticRegression()

[0100] model.fit(X_train_scaled, y_train)

[0101] Step S3.3: Evaluate the performance of the fitted credit default probability model to determine its effectiveness. The evaluation metrics include accuracy, recall, precision, F1-score, ROC curve, and AUC value;

[0102] from sklearn.metrics import accuracy_score, classification_report, roc_auc_score

[0103] # Prediction

[0104] y_pred = model.predict(X_test_scaled)

[0105] y_pred_prob = model.predict_proba(X_test_scaled)[:, 1] # Predicted probability

[0106] # Evaluation Metrics

[0107] print("Accuracy:", accuracy_score(y_test, y_pred))

[0108] print(classification_report(y_test,y_pred))

[0109] print("AUC:",roc_auc_score(y_test,y_pred_prob))

[0110] Step S3.4: Based on the qualified credit default probability model and the enterprise data crawled in real time, predict the enterprise credit default probability.

[0111] Through the Internet data crawling technology, crawl data such as financial news, industry reports, announcements, and analyst ratings on news websites, financial media, and company official websites as enterprise data, and predict the enterprise credit risk level and the enterprise credit default probability.

[0112] Step S4: Generate a credit detection report based on the predicted results of the maximum loss amount VaR that the enterprise will face within a custom time period, the enterprise credit risk level, and the enterprise credit default probability.

[0113] The credit monitoring and assessment system based on market risk quantitative analysis is used to implement the above method, including a risk management tool, a credit risk rating module, and a credit default probability prediction module;

[0114] The risk management tool is responsible for using the historical simulation method of Value at Risk (VaR) and using historical enterprise data to simulate and evaluate the maximum loss amount VaR that a certain financial asset or investment portfolio will face within a custom time period;

[0115] The credit risk rating module is responsible for constructing a credit risk rating model, training the model using a sample set composed of historical data of enterprise credit risk levels, evaluating the performance of the trained model, and making predictions based on the qualified model and the enterprise data crawled in real time to generate the enterprise credit risk level;

[0116] The credit default probability prediction module is responsible for constructing a credit default probability model, fitting the credit default probability model using a training set constructed with historical data of defaulting enterprises, evaluating the performance of the fitted credit default probability model, and predicting the enterprise credit default probability based on the qualified credit default probability model and the enterprise data crawled in real time.

[0117] The credit monitoring and assessment device based on market risk quantitative analysis includes a memory and a processor; the memory is used to store a computer program, and the processor is used to implement the above method steps when executing the computer program.

[0118] The readable storage medium has a computer program stored thereon, and when the computer program is executed by a processor, the method steps described above are implemented.

[0119] The credit monitoring and assessment method and system based on market risk quantitative analysis can significantly improve the management department's ability to identify and manage credit risks. Through real-time monitoring and data analysis, the management department can timely identify potential default risks of enterprises and the financial market, enhance the early warning ability, and provide a scientific basis for formulating management strategies. This data-driven monitoring mechanism enables the management department to formulate differentiated regulatory management strategies according to the credit risk levels of enterprises, flexibly adjust regulatory measures, and improve regulatory adaptability.

[0120] The management department can also use market risk quantitative analysis tools to identify and monitor systemic risks, and take intervention measures in advance to prevent a small-scale credit crisis from spreading to the entire financial system. At the same time, this analysis method provides a scientific basis for formulating macroprudential management strategies to ensure the stability and security of the financial system.

[0121] In terms of credit system construction, quantitative analysis urges the management department to promote cross-departmental and cross-industry credit information sharing platforms to improve the overall social credit awareness and credit level. The management department can also use these data and models to promote the standardized development of the credit rating market and enhance the credibility of credit ratings. In addition, the quantitative analysis method can help the management department quantitatively evaluate the implementation effect of credit management strategies, so as to understand their actual impact on market stability and enterprise credit, and provide feedback basis for optimizing subsequent management strategies. Through long-term credit risk monitoring, the management department can track the long-term effect after the implementation of management strategies, ensure the realization of management strategy goals and make timely adjustments.

[0122] In summary, the application of the credit monitoring and assessment method and system based on market risk quantitative analysis not only improves the management department's credit risk management ability, but also helps to promote economic stability and development, and is of great significance to credit system construction and management strategy effect evaluation.

[0123] The above-described embodiments are only one of the specific implementation manners of the present invention, and the common changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A credit monitoring and assessment method based on market risk quantitative analysis, characterized in that: It includes the following steps: Step S1: Evaluate the potential loss of financial assets within a custom time period; Adopt the historical simulation method of value-at-risk as a risk management tool, and use historical enterprise data to simulate and evaluate the maximum loss amount VaR that a certain financial asset or investment portfolio will face within a custom time period; Step S2: Build a credit risk rating model to predict the credit risk level of an enterprise; Step S2.1: Sort out the detection indicators according to the urban credit detection rules, and configure the corresponding enterprise data for the detection indicators; Step S2.2: Use a machine learning statistical model to create a decision tree model, and customize the hyperparameters of the decision tree according to the actual situation, including the maximum depth of the tree and the minimum sample split number, to prevent overfitting; Step S2.3: Use the sample set composed of historical data of enterprise credit risk levels to train the decision tree model, and evaluate the performance of the trained decision tree model to determine its effectiveness; During the evaluation, calculate the accuracy rate, recall rate, F1-score and ROC curve of the decision tree model to comprehensively evaluate the performance of the decision tree model. Step S2.3: Based on the decision tree model after passing the evaluation and the enterprise data crawled in real time, make a prediction to generate the credit risk level of the enterprise; Step S3: Build a default probability model to predict the default probability of an enterprise; Step S3.1: Adopt a logistic regression algorithm to build a default probability model for evaluating the default probability P of an enterprise within a custom time period; Step S3.2: Use the training set constructed from the historical data of default enterprises to fit the default probability model; Step S3.3: Evaluate the performance of the fitted default probability model to determine its effectiveness. The evaluation indicators include accuracy rate, recall rate, precision rate, F1-score, ROC curve and AUC value; Step S3.4: Based on the default probability model after passing the evaluation and the enterprise data crawled in real time, predict the default probability of the enterprise; Step S4: Generate a credit detection report based on the predicted results of the maximum loss amount VaR that the enterprise will face within a custom time period, the enterprise credit risk level, and the enterprise default probability.

2. The credit monitoring and assessment method based on market risk quantitative analysis according to claim 1, wherein: In the above-mentioned step S1, the processing process of the historical simulation method of value-at-risk is as follows: Step S1.1: Calculate the return rate r of the asset at a certain moment t t , and the calculation formula is as follows: Among them, Pt is the market value at time t, and Pt-1 is the market value at time t-1; Step S1.2: Sort the yield rates from low to high to obtain a total of N samples; Step S1.3: Determine the confidence level α, and determine the position of the maximum loss amount VaR in the samples according to the total number of samples N. The calculation formula is as follows: VaR position = N × (1 - α) The confidence level α is set to 95% or 99%; Step S1.4: Among the sorted yield rates, select the yield rate at the position of the maximum loss amount VaR as the estimated maximum loss amount VaR. The formula is as follows: VaR = -r (N×(1-α)) In the formula, the negative sign represents the loss caused by the risk, and r (1) , r (2) , r (3) ,..., r (N) are all the sorted yields in the sample.

3. The credit monitoring and assessment method based on market risk quantitative analysis according to claim 1, characterized in that: In the above-mentioned step S2, the scikit-learn library of Python is used to implement the decision tree model.

4. The credit monitoring and assessment method based on market risk quantitative analysis according to claim 1, characterized in that: In the above-mentioned step S3, the scikit-learn library in Python is used for modeling to build a default probability model.

5. The credit monitoring and assessment method based on market risk quantitative analysis according to claim 1, characterized in that: The formula of the logistic regression algorithm is as follows: Among them, Y is a default event, where 1 indicates default and 0 indicates no default. X1, X2,..., X m are characteristic variables composed of credit data, and β0, β1, β2, β m are the regression coefficients corresponding to various types of credit data.

6. A credit monitoring and assessment system based on market risk quantitative analysis, characterized in that: For implementing the method according to any one of claims 1 to 5, including a risk management tool, a credit risk rating module, and a default probability prediction module; The risk management tool is responsible for using the value-at-risk historical simulation method and utilizing historical enterprise data to simulate and evaluate the maximum loss amount VaR that a certain financial asset or portfolio will face within a custom time period; The credit risk rating module is responsible for constructing a credit risk rating model, training the model using a sample set composed of historical data of enterprise credit risk levels, evaluating the performance of the trained model, and making predictions based on the qualified model and real-time crawled enterprise data to generate the credit risk level of the enterprise; The default probability prediction module is responsible for constructing a default probability model, fitting the default probability model using a training set constructed from historical data of defaulting enterprises, evaluating the performance of the fitted default probability model, and predicting the default probability of the enterprise based on the qualified default probability model and real-time crawled enterprise data.

7. A credit monitoring and assessment device based on market risk quantitative analysis, characterized in that: Including a memory and a processor; the memory is used to store a computer program, and the processor is used to implement the method according to any one of claims 1 to 5 when executing the computer program.

8. A readable storage medium, characterized in that: A computer program is stored on the readable storage medium, and when the computer program is executed by the processor, it implements the method according to any one of claims 1 to 5.

Images