Risk quantification driven elastic budget model and intelligent decision support method and system

Abstract

The invention discloses a risk quantification-driven elastic budget model and intelligent decision support method and system, and the method comprises the steps: obtaining initial data related to a budget, the initial data comprising internal data of a target unit and external market data; preprocessing the initial data to generate budget decision data; key risk factors in the budget decision data are identified, and risk factor probability distribution is determined; establishing a joint risk factor probability model comprising joint risk factor probability distribution of a plurality of joint key risk factors; generating joint probability distribution through a joint risk factor probability model, and determining a key risk factor value of the key risk factor based on the joint probability distribution; determining a fluctuation interval of the key risk factor based on the key risk factor value through Monte Carlo, and simulating and generating financial index distribution data; and inputting the financial index distribution data into a budget constraint model, and determining an elastic interval of the budget index by the budget constraint model based on a preset objective function and a fluctuation interval.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise financial management and risk control, and more particularly, to a risk quantification driven flexible budget model and an intelligent decision support method and system. BACKGROUND

[0002] Currently, under the background of globalization and digital transformation, enterprise budget management is facing the following challenges:

[0003] Traditional budget model is rigid: the traditional budget based on fixed assumptions (such as single growth rate, constant cost rate) cannot cope with market uncertainty (such as copper price quarterly fluctuation exceeding 20%, exchange rate annual fluctuation range reaching 15%), resulting in budget execution deviation rate generally exceeding 30%;

[0004] Risk quantification tools are insufficient: existing flexible budget only considers limited scenarios (such as optimistic / pessimistic / benchmark), does not build a continuous probability distribution model, and it is difficult to quantify the impact of extreme events;

[0005] Decision support capability is weak: lack of automated risk hedging recommendation mechanism, budget adjustment relies on manual experience judgment, resulting in resource allocation lagging behind market changes.

[0006] The prior art, based on scenario analysis, involves generating a flexible budget by presetting multiple scenarios (such as GDP growth of 3% / 5% / 7%). However, the prior art has the following problems:

[0007] Discrete scenario limitations: only a limited number of discrete scenarios are considered, and a continuous probability distribution (such as normal distribution of copper price fluctuations) is not built, which cannot cover extreme risk events;

[0008] Risk factor simplification: only macroeconomic indicators (such as GDP, CPI) are considered, without considering industry-specific risks (such as the impact of power grid renovation policy on cable demand);

[0009] Lack of dynamic optimization: scenarios are preset and fixed, and the probability distribution is not updated according to real-time data, resulting in budget flexibility lagging behind market changes.

[0010] Therefore, a technology is needed to realize a risk quantification driven flexible budget model and intelligent decision support technology. SUMMARY

[0011] The technical solution of the present application provides a risk quantification driven flexible budget model and an intelligent decision support method and system to solve the problem of how to determine the optimal budget flexibility interval by building a joint risk factor probability model.

[0012] In order to solve the above problems, the present application provides a risk quantification driven flexible budget model and an intelligent decision support method, which comprises:

[0013] Obtaining initial data related to budget, the initial data including internal data of target unit and external market data; preprocessing the initial data to generate budget decision data;

[0014] Identifying key risk factors in the budget decision data, determining risk factor probability distribution of the key risk factors; establishing a joint risk factor probability model including joint risk factor probability distribution of multiple joint key risk factors;

[0015] Generating joint probability distribution through the joint risk factor probability model, determining key risk factor values of the key risk factors based on the joint probability distribution;

[0016] Determining fluctuation intervals of the key risk factors based on the key risk factor values through Monte Carlo, and simulating to generate financial indicator distribution data;

[0017] Inputting the financial indicator distribution data into a budget constraint model, the budget constraint model determining flexible intervals of budget indicators based on preset target functions and the fluctuation intervals.

[0018] Preferably, the method comprises:

[0019] Monitoring real-time data of the key risk factors, and calculating conditional value at risk of the key risk factors based on the real-time data;

[0020] When the conditional value at risk of the key risk factors exceeds a preset risk threshold, providing risk hedging decisions through a trained risk hedging model.

[0021] Preferably, the algorithm of the risk hedging model comprises a random forest algorithm or a reinforcement learning algorithm;

[0022] Performing risk exposure assessment on the key risk factors, determining influence of changes of the key risk factors on the financial indicator distribution data, and obtaining risk exposure assessment results;

[0023] Taking the risk exposure assessment results, risk hedging tool library parameters and budget constraint conditions as inputs of the risk hedging model, training the risk hedging model until hedging cost and benefit of the risk hedging decisions reach preset target values.

[0024] Preferably, the risk factor probability distribution comprises normal distribution probability density, lognormal distribution probability density and t distribution;

[0025] The parameters in the Gaussian Copula function are determined by the maximum likelihood estimation method, the joint variation relationship between the joint risk factors is determined by the Gaussian Copula function with the determined parameters, and the joint risk factor probability model is established.

[0026] Preferably, the determination of the fluctuation interval of the key risk factor based on the key risk factor value by the Monte Carlo method and the simulation generation of the financial indicator distribution data comprise:

[0027] The financial indicator distribution data is generated by preset maximum iteration number loop iteration:

[0028] Based on the risk factor value obtained in the joint probability distribution, the financial indicator distribution data is calculated by the risk factor value, and the financial indicator distribution data is recorded;

[0029] The financial indicator distribution data is generated by continuous loop iteration until the maximum iteration number is reached.

[0030] Preferably, it further comprises statistical analysis of the financial indicator distribution data to determine the statistical value of the financial indicator distribution data;

[0031] Based on the statistical value of the financial indicator distribution data, the data characteristics of the financial indicator distribution data are determined;

[0032] Based on the data characteristics, the financial performance of the financial indicator distribution data is determined.

[0033] Preferably, the target unit internal data includes: historical purchase price, sales data and inventory turnover rate;

[0034] The external market data includes: copper price, central bank exchange rate mid-point, industry policy text;

[0035] The target unit internal data and the external market data are preprocessed, including: missing value filling, abnormal value correction and time series decomposition.

[0036] Preferably, the joint risk factor probability model including the joint key risk factor probability distribution is established, comprising:

[0037] The distribution of each key risk factor is determined by the K-S test method;

[0038] Based on the determined distribution of each key risk factor, the joint risk factor probability model is established by the Copula function, and the parameters in the joint risk factor probability model are determined by the maximum likelihood estimation method.

[0039] Based on another aspect of the present application, the present application provides a risk quantification driven flexible budget model and an intelligent decision support system, the system comprising:

[0040] An initial unit for obtaining initial data related to budget, the initial data comprising internal data of a target unit and external market data; preprocessing the initial data to generate budget decision data;

[0041] An establishment unit for identifying key risk factors in the budget decision data, determining risk factor probability distribution of the key risk factors, and establishing a joint risk factor probability model comprising joint risk factor probability distribution of multiple joint key risk factors;

[0042] A determination unit for generating joint probability distribution through the joint risk factor probability model, and determining key risk factor values of the key risk factors based on the joint probability distribution;

[0043] A generation unit for determining fluctuation intervals of the key risk factors based on the key risk factor values through Monte Carlo, and simulating to generate financial indicator distribution data;

[0044] A result unit for inputting the financial indicator distribution data into a budget constraint model, the budget constraint model determining flexible intervals of budget indicators based on preset target functions and the fluctuation intervals.

[0045] Preferably, the system comprises a hedging unit for:

[0046] Monitoring real-time data of the key risk factors, and calculating conditional value at risk of the key risk factors based on the real-time data;

[0047] When the conditional value at risk of the key risk factors exceeds a preset risk threshold, providing a risk hedging decision through a trained risk hedging model.

[0048] Preferably, the algorithm of the risk hedging model comprises a random forest algorithm or a reinforcement learning algorithm;

[0049] The hedging unit is used for risk exposure assessment of the key risk factors, determining influence of changes of the key risk factors on the financial indicator distribution data, and obtaining risk exposure assessment results;

[0050] Taking the risk exposure assessment results, risk hedging tool library parameters, and budget constraint conditions as inputs of the risk hedging model, the risk hedging model is trained until hedging cost and benefit of the risk hedging decision reach a preset target value.

[0051] Preferably, the risk factor probability distribution comprises: a normal distribution probability density, a lognormal distribution probability density, and a t-distribution.

[0052] The establishment unit is further configured to determine parameters in a Gaussian Copula function by a maximum likelihood estimation method, determine a joint variation relationship between joint risk factors by the Gaussian Copula function with the parameters, and establish a joint risk factor probability model.

[0053] Preferably, the generation unit is configured to determine a fluctuation interval of the key risk factor based on the key risk factor value by a Monte Carlo method, and simulate and generate financial indicator distribution data, and is further configured to:

[0054] The financial indicator distribution data is generated by a preset maximum iteration number of loop iterations:

[0055] Based on the risk factor value obtained from the joint probability distribution, the financial indicator distribution data is calculated by the risk factor value, and the financial indicator distribution data is recorded;

[0056] The loop iteration for generating the financial indicator distribution data is continued until the maximum iteration number is reached.

[0057] Preferably, the generation unit is further configured to statistically analyze the financial indicator distribution data to determine a statistical value of the financial indicator distribution data.

[0058] Based on the statistical value of the financial indicator distribution data, a data feature of the financial indicator distribution data is determined.

[0059] Based on the data feature, a financial performance of the financial indicator distribution data is determined.

[0060] The target unit internal data comprises: historical purchase prices, sales data, and inventory turnover rates.

[0061] The external market data comprises: copper prices, central bank exchange rate midpoints, and industry policy texts.

[0062] The target unit internal data and the external market data are preprocessed, including: missing value filling, abnormal value correction, and time series decomposition.

[0063] Preferably, the establishment unit is configured to establish a joint risk factor probability model comprising a joint risk factor probability distribution of multiple joint key risk factors, and is further configured to:

[0064] The distribution of each key risk factor is determined by a K-S test method.

[0065] Based on the determined distribution of each key risk factor, a joint risk factor probability model is established by a Copula function, and parameters in the joint risk factor probability model are determined by a maximum likelihood estimation method.

[0066] The technical scheme of the present application provides a risk quantification driven flexible budget model and an intelligent decision support method and system, wherein the method comprises: obtaining initial data related to budget, the initial data including internal data of a target unit and external market data; preprocessing the initial data to generate budget decision data; identifying key risk factors in the budget decision data and determining risk factor probability distribution of the key risk factors; establishing a joint risk factor probability model including joint risk factor probability distribution of multiple joint key risk factors; generating joint probability distribution through the joint risk factor probability model, determining key risk factor values of the key risk factors based on the joint probability distribution; determining fluctuation intervals of the key risk factors based on the key risk factor values through Monte Carlo, and simulating to generate financial indicator distribution data; inputting the financial indicator distribution data into a budget constraint model, and determining a flexible interval of a budget indicator based on a preset target function and the fluctuation intervals. The technical scheme of the present application constructs a joint probability distribution model containing internal operation indicators (such as procurement price fluctuation rate) and external market variables (exchange rate, commodity price), and breaks through the limitation of traditional discrete scenario analysis; the technical scheme of the present application generates financial indicator distribution under thousands of operating scenarios through Monte Carlo simulation, and automatically identifies an optimal budget flexible interval (such as a cost allocation scheme when income fluctuation is ± 15%). BRIEF DESCRIPTION OF DRAWINGS

[0067] The exemplary embodiments of the present application can be more completely understood in reference to the following drawings:

[0068] Figure 1 A flow chart of a risk quantification driven flexible budget model and an intelligent decision support method according to a preferred embodiment of the present application;

[0069] Figure 2 A flow chart of a risk quantification driven flexible budget model and an intelligent decision support method according to a preferred embodiment of the present application; and

[0070] Figure 3 A structure diagram of a risk quantification driven flexible budget model and an intelligent decision support system according to a preferred embodiment of the present application. DETAILED DESCRIPTION

[0071] Reference will now be made to the drawings to describe the exemplary embodiments of the present application in greater detail. The present application may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and fully convey the scope of the application to those skilled in the art. Like numbers refer to like elements throughout the description of the figures. In the drawings:

[0072] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0073] Figure 1 A risk quantification driven flexible budget model and intelligent decision support method according to a preferred embodiment of the present application.

[0074] The present application provides multi-dimensional risk factor probabilistic modeling: constructing joint probability distribution models containing both internal operating indicators (e.g. procurement price volatility) and external market variables (exchange rates, commodity prices), breaking through the limitations of traditional discrete scenario analysis;

[0075] The present application provides a dynamic flexible budget generation engine: generating the distribution of financial indicators under thousands of operating scenarios through Monte Carlo simulation, automatically identifying the optimal budget flexibility interval (e.g. cost allocation scheme when income fluctuates ± 15%);

[0076] The present application provides an intelligent risk hedging decision system: based on the risk value (VaR) and conditional risk value (CVaR) indicators, generating customized risk hedging suggestions (e.g. procurement option contracts, exchange rate forward locking), realizing the dynamic matching of budget flexibility and risk preference.

[0077] The present application provides an intelligent risk hedging decision system: based on the risk value (VaR) and conditional risk value (CVaR) indicators, generating customized risk hedging suggestions (e.g. procurement option contracts, exchange rate forward locking), realizing the dynamic matching of budget flexibility and risk preference.

[0078] The present application belongs to the technical field of enterprise financial management and risk control, and specifically relates to a probabilistic budget framework integrating Monte Carlo simulation and scenario planning, which is suitable for budgeting and resource allocation decisions in response to market fluctuations (exchange rates, commodity prices, policy changes, etc.) in manufacturing, retail, finance and other industries. By constructing a probabilistic distribution model of multi-dimensional risk factors, automatic identification of budget flexibility interval and dynamic risk hedging strategy generation are realized.

[0079] like Figure 1 As shown, this invention provides a risk quantification-driven elastic budgeting model and intelligent decision support method, the method comprising:

[0080] Step 101: Obtain initial budget-related data, including internal data of the target unit and external market data; preprocess the initial data to generate budget decision data;

[0081] Preferably, the internal data of the target unit includes: historical purchase prices, sales data, and inventory turnover rate;

[0082] External market data includes: copper prices, central bank exchange rate midpoint, and industry policy documents;

[0083] Preprocessing is performed on the internal data of the target unit and the external market data, including: missing value imputation, outlier correction, and time series decomposition.

[0084] This invention involves data acquisition and preprocessing:

[0085] Internal data: Historical purchase prices, sales data, and inventory turnover rate from the ERP system;

[0086] External data: LME copper price API, central bank exchange rate midpoint, industry policy texts (keywords extracted via NLP);

[0087] Preprocessing modules: missing value imputation, outlier correction, time series decomposition (trend / seasonal / cyclical components).

[0088] Step 102: Identify key risk factors in budget decision data and determine the risk factor probability distribution of key risk factors; establish a joint risk factor probability model that includes the joint risk factor probability distribution of multiple joint key risk factors.

[0089] Preferably, the probability distribution of the risk factors includes: normal probability density, log-normal probability density, and t-distribution;

[0090] The parameters in the Gaussian Copula function are determined by the maximum likelihood estimation method. The joint variation relationship between the joint risk factors is determined by the Gaussian Copula function with determined parameters, and a joint risk factor probability model is established.

[0091] Step 103: Generate a joint probability distribution using a joint risk factor probability model, and determine the key risk factor values ​​for key risk factors based on the joint probability distribution;

[0092] Step 104: Determine the fluctuation range of key risk factors based on key risk factor values ​​using Monte Carlo simulation, and generate financial indicator distribution data.

[0093] Preferably, the fluctuation range of key risk factors is determined based on key risk factor values ​​using Monte Carlo simulation, and financial indicator distribution data is generated through simulation, including:

[0094] Financial indicator distribution data is generated through iterative iterations using a preset maximum number of iterations.

[0095] Risk factor values ​​are obtained from the joint probability distribution, and financial indicator distribution data are calculated and recorded using the risk factor values.

[0096] Continue iterating to generate financial indicator distribution data until the maximum number of iterations is reached.

[0097] Preferably, the method further includes statistical analysis of the financial indicator distribution data to determine the statistical values ​​of the financial indicator distribution data;

[0098] Determine the data characteristics of financial indicator distribution data based on statistical values ​​of financial indicator distribution data;

[0099] Based on data characteristics, assess the financial performance of financial indicator distribution data.

[0100] Preferably, a joint risk factor probability model is established, comprising the probability distribution of multiple joint key risk factors, including:

[0101] The distribution of each key risk factor was determined using the KS test.

[0102] Based on the distribution of each key risk factor, a joint risk factor probability model is established using the Copula function, and the parameters in the joint risk factor probability model are determined using the maximum likelihood estimation method.

[0103] This invention establishes a probability model for risk factors:

[0104] Marginal distribution modeling: Fit the optimal probability distribution (normal distribution, log-normal distribution, t-distribution, etc.) for each risk factor (such as copper price, exchange rate), and verify the goodness of fit through the KS test;

[0105] Copula function construction: capturing the non-linear correlation between risk factors (such as the linkage between copper prices and the US dollar exchange rate).

[0106] This invention generates a joint probability distribution;

[0107] Scenario generation engine: Based on joint distribution, it generates 10,000+ business scenarios through Monte Carlo simulation, covering a 99.9% confidence interval.

[0108] Suppose we construct a probabilistic model involving two risk factors: copper price and exchange rate. We collect daily data on copper price and USD / CNY exchange rate over the past 10 years. First, we attempt to fit the copper price data using a normal distribution, calculating the mean and standard deviation. The distribution is described by the normal probability density function. A KS test comparing the actual data with the cumulative distribution function of the normal distribution reveals a large deviation, especially at the tails. We then take the logarithm of the copper price data and attempt a log-normal distribution, recalculating the mean and standard deviation. The KS test shows a significantly improved goodness of fit, confirming the log-normal distribution as the marginal distribution for copper prices. For the exchange rate data, we first fit it using a normal distribution. After calculating the relevant parameters, a KS test shows a poor fit. We then attempt a t-distribution, adjusting the parameters. The KS test shows the t-distribution fits the exchange rate data well, confirming the t-distribution as the marginal distribution for exchange rates. When constructing the Copula function, considering the non-linear relationship between copper price and exchange rate in international trade, we choose a Gaussian Copula function. Based on the collected historical data, we use maximum likelihood estimation to accurately determine the relevant parameters in the function, constructing a joint probability distribution that reflects the joint changes in both. Finally, scenario generation is performed. Based on the joint probability distribution, Monte Carlo simulation is used, with the number of simulations set at 15,000. In each simulation, a combination of copper price and exchange rate values ​​is extracted from the joint distribution using a random number generator to represent a market scenario. During the simulation, the risk factor values ​​extracted each time and the corresponding financial indicator calculation results are recorded to generate a large number of business scenarios covering various possible situations, thus completing the construction of the risk factor probability model.

[0109] The risk factor probability model generates numerous scenarios (e.g., 15,000 simulations) covering different combinations of risk factors (such as copper prices and exchange rates) through Monte Carlo simulations. This provides specific data inputs for the budget constraint model under each scenario. For example, when calculating the expected net present value E[NPV], the model calculates based on the cash inflows and outflows caused by risk factors in different scenarios. When calculating the conditional value at risk CVaRα(NPV), it assesses risk based on the net present value distribution corresponding to each scenario. Simultaneously, the risk factor probability model outputs the range of risk factor fluctuations (e.g., copper price 95...). The fluctuation range of ±10% to ±20% under certain scenarios can be directly used in the budget constraint model to determine the elasticity range of budget indicators (such as setting a fluctuation range of less than 15% for the procurement cost budget deviation rate), thus assisting in optimizing the adaptability of the budget. In addition, the impact weight of risk factors on each business segment analyzed by the risk factor probability model (such as the impact weight of copper price fluctuation on the profit of a certain business being 40%) provides a basis for the budget constraint model in resource allocation, enabling the model to prioritize resources towards high-risk areas (such as reserving option hedging funds to cope with copper price risks), thereby achieving a balance between risk and return.

[0110] Step 105: Input the financial indicator distribution data into the budget constraint model. The budget constraint model determines the elasticity range of the budget indicators based on the preset objective function and fluctuation range.

[0111] Preferably, the method includes:

[0112] Monitor real-time data of key risk factors and calculate the conditional value of risk of key risk factors based on the real-time data;

[0113] When the conditional risk value of a key risk factor exceeds a preset risk threshold, a risk hedging decision is provided through the trained risk hedging model.

[0114] Preferably, the algorithm for the risk hedging model includes: random forest algorithm or reinforcement learning algorithm;

[0115] Conduct risk exposure assessments on key risk factors, determine the impact of changes in key risk factors on the distribution data of financial indicators, and obtain the risk exposure assessment results.

[0116] The risk exposure assessment results, risk hedging tool library parameters, and budget constraints are used as inputs to train the risk hedging model until the hedging costs and benefits of the risk hedging decision reach the preset target values.

[0117] Elastic budget optimization layer: max(E[NPV]-λ·CVaRa(NPV))

[0118] Where E[NPV] is the expected net present value.

[0119] CVaR α (NPV) is the conditional value of risk at a confidence level of 'a', and the risk aversion coefficient (set by management).

[0120] Elastic range identification: Determine the optimal fluctuation range (e.g., ±10% to ±20%) for key budget indicators (such as procurement costs and sales expenses) through simulation results, ensuring that the budget deviation rate is <15% in 95% of scenarios;

[0121] Resource allocation optimization: Dynamically adjust resource allocation based on risk contribution (e.g., the impact of copper price fluctuations on profits has a weight of up to 40%), and prioritize budget flexibility for high-risk areas.

[0122] This invention constructs a budget constraint model by first defining the objective function (e.g., maximizing the expected net present value E[NPV] or minimizing the conditional value at risk CVaRα(NPV)). Constraints are then set based on business logic (e.g., revenue not falling below cost, cash flow balance, etc.). Scenario data generated by the risk factor probability model (e.g., cash inflows and outflows under different combinations of copper prices and exchange rates) are substituted into the model. The optimal budget scheme is then solved using algorithms such as linear programming and nonlinear programming. When identifying the elasticity range, the model is based on the risk factor fluctuation range output by the risk factor probability model (e.g., copper prices fluctuating ±10% to ±20% at a 95% confidence level). The impact of copper price fluctuations on various budget indicators (such as procurement costs and sales revenue) is analyzed. Sensitivity analysis is used to determine the elasticity range of key indicators (such as setting the deviation rate of procurement cost budget to no more than 15% under the influence of copper price fluctuations). Resource allocation optimization is based on the risk factor impact weight of each business segment calculated by the risk factor probability model (such as copper price fluctuations having a 40% impact weight on the profit of business A and a 20% impact weight on business B). In the budget constraint model, resources (such as funds and manpower) are preferentially allocated to high-risk impact segments (such as reserving more funds for business A to cope with copper price fluctuations), while taking into account overall risk diversification.

[0123] The flexible budget optimization of this invention provides a decision boundary for the risk hedging decision-making layer by identifying the flexible range of budget indicators (such as the fluctuation range of costs and revenues). The risk hedging decision-making layer formulates specific hedging strategies based on this range (such as initiating option hedging when the copper price fluctuation exceeds the upper limit of the flexible range). The two form a closed loop: the flexible budget optimization layer determines the tolerable risk range, the risk hedging decision-making layer selects the most cost-effective hedging tool (such as futures or options) within this range to perform hedging, and the hedging results are fed back to the flexible budget optimization layer to adjust the budget flexible range for the next period, thereby achieving dynamic collaborative optimization.

[0124] This invention enables risk hedging decisions:

[0125] Risk exposure assessment: Calculate the VaR and CVaR values ​​of each risk factor and identify key risk points (such as a 20% increase in copper prices leading to a 30% decrease in profits);

[0126] Hedging tool library: Includes tools such as purchase options, currency forward contracts, and futures hedging, and evaluates the cost-benefit ratio of different tools;

[0127] Intelligent suggestion generation: Based on the risk quantification results, recommend the optimal hedging combination (such as purchasing copper price call options + signing a fixed exchange rate contract) and automatically calculate the hedging cost and benefit break-even point.

[0128] When assessing risk exposure, this invention uses data on the fluctuation range of copper prices and exchange rates at different confidence levels and combined scenarios generated by a risk factor probability model. It calculates the magnitude of changes in financial indicators such as net profit and cash flow for each business unit under different scenarios. Delta sensitivity analysis quantifies the impact of a 1% change in copper prices on the profit of a specific business. The expected risk exposure is calculated by weighting the scenarios with probability of occurrence. Value at Risk (VaR) and Conditional Value at Risk (CVaR) are used to assess potential losses under extreme scenarios. Building a hedging tool library requires classifying and organizing the characteristics and applicable scenarios of various financial instruments. Futures contracts can lock in copper price costs, suitable for linear hedging needs, with low cost but insufficient flexibility. Option contracts, after paying a premium, grant the right to exercise the option, providing price volatility protection, suitable for non-linear risk exposure. Swap contracts can be used for long-term exchange rate fluctuation hedging, reducing cash flow uncertainty through fixed and floating exchange rate swaps. Simultaneously, a tool parameter database is established to record information such as copper futures prices, option strike prices, and volatility for different expiration dates. It also labels the transaction costs, liquidity levels, and historical hedging efficiency data for each tool. For example, the average efficiency of a certain option in hedging copper price risk over the past year is 85%. The intelligent suggestion generation module uses machine learning algorithms such as random forests and reinforcement learning to train the model to find the optimal hedging combination. This is achieved by taking the risk exposure assessment results (e.g., a certain business's copper price risk exposure is ±2 million yuan, and exchange rate risk exposure is ±1.5 million yuan), hedging tool library parameters (3-month copper futures hedging cost 0.5%, call option premium rate of 60,000 yuan / ton) and budget constraints (total hedging cost not exceeding 10% of expected profit) as input. When copper price fluctuations exceed the upper limit of the 95% confidence interval, the model automatically recommends buying 100 lots of put options with a strike price of 65,000 yuan / ton to hedge 60% of the exposure, while simultaneously selling 50 lots of futures contracts to hedge 30% of the exposure. It also generates operation instructions and expected effect reports, showing that the risk exposure is expected to be reduced to ±800,000 yuan, with costs accounting for 8% of profits, thus achieving automated intelligent decision-making from risk assessment to hedging strategies.

[0129] The following provides illustrative examples of embodiments of the present invention (e.g.) Figure 2 As shown):

[0130] Risk factor identification and data preparation:

[0131] Principal component analysis (PCA) is used to screen key risk factors affecting the budget (such as copper prices, exchange rates, and employee spending trends) and determine their historical volatility characteristics.

[0132] Applying Box-Cox transformation to non-normally distributed risk factors (such as commodity prices) can improve the model's fitting accuracy.

[0133] Risk factor identification and data preparation are mainly implemented in the risk exposure assessment module, because accurately assessing risk exposure requires first identifying which risk factors are involved and their potential impact, and then preparing relevant data for quantitative analysis.

[0134] This invention employs principal component analysis to reduce the dimensionality of an initial set of risk factors, such as copper prices, exchange rates, employee consumption trends, and market demand fluctuations. It calculates the factor loading matrix, identifies key factors that contribute more than 80% to the cumulative variance of the budget impact, and extracts the top three principal components corresponding to copper prices, exchange rates, and employee consumption trends. Based on quarterly data from the past five years, it analyzes the volatility characteristics of each factor, including the distribution of mean, standard deviation, skewness, kurtosis, and extreme values, clarifying that the historical maximum single-day drop in copper prices is 12% and the annual fluctuation range of exchange rates is ±15%. For data preparation, for non-normally distributed risk factors such as copper prices, the normality hypothesis was rejected using the KS test. A Box-Cox transformation was then employed, and the optimal transformation parameter λ = 0.3 was determined through maximum likelihood estimation. This reduced the skewness of the transformed data from 1.8 to 0.5 and the kurtosis from 4.2 to 3.1, approximating a normal distribution. Simultaneously, all risk factor data were standardized to a mean of 0 and a standard deviation of 1. The training and test sets were then divided in a 7:3 ratio. The training set was used for model parameter estimation, such as the volatility parameter in the risk factor probability model, while the test set was used to validate the model's prediction error. After the transformation, the RMSE (RMS of the copper price prediction) decreased from 0.8 million yuan / ton to 0.5 million yuan / ton. Finally, a structured dataset containing fields such as time series, factor values, and transformed values ​​was formed.

[0135] Monte Carlo simulation and results analysis:

[0136] Each simulation iteration includes the following steps:

[0137] Python

[0138] run

[0139] for i in range(10000):

[0140] #Sampling risk factor values ​​from the joint distribution

[0141] risk_factors=copula_generator.sample()

[0142] # Input the operating model to calculate financial indicators

[0143] financial_metrics=operation_model(risk_factors)

[0144] #Record Results

[0145] results.append(financial_metrics)

[0146] This invention performs statistical analysis on the simulation results, generates probability distribution curves for indicators such as profit and cash flow, and calculates statistical quantities such as mean, standard deviation, and quantiles.

[0147] This invention is implemented using Python code, employing a for loop for 10,000 iterations. In each iteration, risk factor values ​​are first extracted from the joint distribution generated by the risk factor probability model (e.g., a joint distribution constructed using the Copula function) using the `copula_generator.sample()` function, resulting in a set of specific numerical combinations of risk factors such as copper prices and exchange rates. Then, these risk factor values ​​are input into the operation_model, an operation model constructed based on the company's operational logic, to calculate corresponding financial indicators, such as net profit and cash flow. Finally, the results of each calculated financial indicator are recorded in the `results` list, completing one simulation iteration. After the simulation, statistical analysis is performed on the large number of financial indicator results recorded in the `results` list. The mean of these indicators is calculated to obtain the average level of the financial indicators; the standard deviation is calculated to measure the volatility of the financial indicators. By generating probability distribution curves for indicators such as profit and cash flow, the probability of the indicators under different values ​​is visually displayed. Simultaneously, quantiles are calculated, such as the 25th, 50th, and 75th quantiles, to further understand the characteristics of the indicator distribution and judge the performance of financial indicators at different levels.

[0148] The risk factor identification and data preparation steps identified key risk factors and their historical volatility characteristics, and preprocessed the data. This work provided the foundation for Monte Carlo simulations, where the risk factor values ​​extracted during the simulation were based on the identified and processed risk factors and their distributions, ensuring that the simulation accurately reflects the actual risk situation.

[0149] The probability distribution, mean, and standard deviation of financial indicators obtained from Monte Carlo simulations and results analysis serve as crucial data for generating flexible budgeting and hedging strategies. Regarding flexible budgeting, these results determine the fluctuation range of budget indicators (such as procurement costs and sales expenses), setting reasonable flexibility limits. For hedging strategies, risk indicators (such as the degree to which copper price fluctuations cause profit fluctuations) determine whether to trigger hedging recommendations. When the risk exceeds a preset threshold (such as the CVaR of copper price fluctuations exceeding a set value), simulation results are used to further calculate parameters such as the strike price and contract size of the optimal hedging instrument, thus formulating an effective hedging strategy.

[0150] Flexible budgeting and hedging strategy generation:

[0151] Based on the risk aversion coefficient (e.g., λ = 0.5 indicates moderate risk aversion), find the Pareto optimal solution on the risk-return plane;

[0152] When the CVaR of copper price fluctuations exceeds a preset threshold, a purchase option recommendation is automatically triggered, and the optimal strike price and contract size are calculated.

[0153] A risk aversion coefficient is determined, and a large amount of scenario data simulated by the risk factor probability model is input into the budget constraint model. Budget indicators such as procurement costs and sales expenses are continuously adjusted within the risk-return plane. An optimization algorithm is used to find the Pareto optimal solution, maximizing returns while keeping risks under control. Simultaneously, based on the volatility range of risk factors obtained from Monte Carlo simulations, their impact on financial indicators is analyzed. For example, the impact of copper price fluctuations on procurement costs is analyzed, and the range of fluctuations in the procurement cost budget where the deviation rate is less than a certain percentage is calculated to determine the elasticity range. For hedging strategy generation, changes in risk factors are continuously monitored, and the conditional value at risk (CVaR) of key risk factors is calculated. When the CVaR of key risk factors such as copper prices exceeds a pre-set value, the hedging mechanism is triggered. From a hedging tool library containing instruments such as procurement options, currency forward contracts, and futures hedging, the costs and benefits of different instruments under the current risk situation are evaluated. For example, the premium and potential return of buying copper price call options are evaluated. Finally, based on the risk exposure assessment results and the characteristics of the hedging tools themselves, the optimal parameters are precisely calculated, such as the strike price of buying copper call options being a certain percentage above the current copper price, and the contract size covering a certain percentage of the company's expected copper purchase volume, to complete the hedging strategy formulation.

[0154] This invention significantly improves the accuracy of budgeting:

[0155] The budget deviation rate was reduced from over 30% using traditional methods to less than 15%. After implementation, a manufacturing company saw its budget deviation rate due to copper price fluctuations decrease from 42% to 12%.

[0156] The coverage of extreme risk events (such as daily exchange rate fluctuations exceeding 3%) has increased from less than 50% to 99%, significantly enhancing the budget's ability to withstand risks.

[0157] This invention enhances risk quantification and decision support capabilities:

[0158] By making risks visible and quantifiable, management can intuitively see the budget flexibility range at different confidence levels (such as revenue fluctuation of ±18% within a 95% confidence range);

[0159] The risk hedging recommendations enabled the company to save approximately 8% on procurement costs during the copper price upswing cycle in 2023, and avoid exchange rate losses exceeding 5 million yuan by locking in exchange rates in advance.

[0160] This invention optimizes resource allocation efficiency:

[0161] The resource allocation mechanism based on risk contribution increases budget flexibility by 30% in high-risk areas (such as raw material procurement) and reduces resource consumption in low-risk areas (such as administrative expenses) by 15%.

[0162] Automated decision-making processes reduce budget adjustment response time from the traditional monthly to real-time, significantly enhancing agility in responding to market changes.

[0163] This invention breaks through the rigidity of traditional budgeting by using probabilistic modeling and intelligent decision support, and constructs a flexible budget management system driven by risk quantification. It achieves deep integration of budget preparation, risk control and resource allocation, and has significant technological innovation and industry application value.

[0164] Figure 3 This is a structural diagram of a risk quantification-driven elastic budget model and intelligent decision support system according to a preferred embodiment of the present invention.

[0165] like Figure 3 As shown, this invention provides a risk quantification-driven elastic budgeting model and intelligent decision support system, the system comprising:

[0166] Initial unit 301 is used to acquire initial data related to the budget, including internal data of the target unit and external market data; the initial data is preprocessed to generate budget decision data.

[0167] Unit 302 is established to identify key risk factors in budget decision data, determine the risk factor probability distribution of key risk factors, and establish a joint risk factor probability model that includes the joint risk factor probability distribution of multiple joint key risk factors.

[0168] Unit 303 is used to generate a joint probability distribution through a joint risk factor probability model, and to determine the key risk factor values ​​of key risk factors based on the joint probability distribution.

[0169] The generation unit 304 is used to determine the fluctuation range of key risk factors based on key risk factor values ​​through Monte Carlo simulation, and to simulate and generate financial indicator distribution data.

[0170] Result unit 305 is used to input financial indicator distribution data into the budget constraint model. The budget constraint model determines the elasticity range of budget indicators based on the preset objective function and fluctuation range.

[0171] Preferably, the system includes a hedging unit for:

[0172] Monitor real-time data of key risk factors and calculate the conditional value of risk of key risk factors based on the real-time data;

[0173] When the conditional risk value of a key risk factor exceeds a preset risk threshold, a risk hedging decision is provided through the trained risk hedging model.

[0174] Preferably, the algorithm for the risk hedging model includes: random forest algorithm or reinforcement learning algorithm;

[0175] The hedging unit is used to assess the risk exposure of key risk factors, determine the impact of changes in key risk factors on the distribution data of financial indicators, and obtain the risk exposure assessment results.

[0176] The risk exposure assessment results, risk hedging tool library parameters, and budget constraints are used as inputs to train the risk hedging model until the hedging costs and benefits of the risk hedging decision reach the preset target values.

[0177] Preferably, the probability distribution of the risk factors includes: normal probability density, log-normal probability density, and t-distribution;

[0178] The unit is also used to determine the parameters in the Gaussian Copula function through the maximum likelihood estimation method, and to determine the joint variation relationship between joint risk factors by determining the Gaussian Copula function with determined parameters, thereby establishing a joint risk factor probability model.

[0179] Preferably, the generation unit 304 is used to determine the fluctuation range of key risk factors based on key risk factor values ​​using Monte Carlo simulation, and to simulate and generate financial indicator distribution data, and is also used to:

[0180] Financial indicator distribution data is generated through iterative iterations using a preset maximum number of iterations.

[0181] Risk factor values ​​are obtained from the joint probability distribution, and financial indicator distribution data are calculated and recorded using the risk factor values.

[0182] Continue iterating to generate financial indicator distribution data until the maximum number of iterations is reached.

[0183] Preferably, the generating unit 304 is also used to perform statistical analysis on the financial indicator distribution data to determine the statistical values ​​of the financial indicator distribution data;

[0184] Determine the data characteristics of financial indicator distribution data based on statistical values ​​of financial indicator distribution data;

[0185] Based on data characteristics, assess the financial performance of financial indicator distribution data.

[0186] Preferably, the internal data of the target unit includes: historical purchase prices, sales data, and inventory turnover rate;

[0187] External market data includes: copper prices, central bank exchange rate midpoint, and industry policy documents;

[0188] Preprocessing is performed on the internal data of the target unit and the external market data, including: missing value imputation, outlier correction, and time series decomposition.

[0189] Preferably, the establishing unit is used to establish a joint risk factor probability model that includes the probability distribution of multiple joint key risk factors, and is also used to:

[0190] The distribution of each key risk factor was determined using the KS test.

[0191] Based on the distribution of each key risk factor, a joint risk factor probability model is established using the Copula function, and the parameters in the joint risk factor probability model are determined using the maximum likelihood estimation method.

[0192] The risk quantification-driven elastic budget model and intelligent decision support system proposed in this invention correspond to the risk quantification-driven elastic budget model and intelligent decision support method proposed in another embodiment of this invention, and will not be described in detail here.

[0193] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0194] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0195] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0196] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0197] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0198] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0199] The invention has been described with reference to a few embodiments. However, as will be known to those skilled in the art, and as defined in the appended claims, other embodiments besides those disclosed above fall equivalently within the scope of the invention.

[0200] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the art, unless otherwise expressly defined herein. All references to “a / the / the [device, component, etc.]” ​​are openly interpreted as at least one instance of said device, component, etc., unless otherwise expressly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed unless explicitly stated otherwise.

Claims

1. A risk quantification-driven elastic budgeting model and intelligent decision support method, the method comprising: Obtain initial budget-related data, including internal data of the target unit and external market data; The initial data is preprocessed to generate budget decision data; Identify the key risk factors in the budget decision data, determine the risk factor probability distribution of the key risk factors, and establish a joint risk factor probability model that includes the joint risk factor probability distribution of multiple joint key risk factors. A joint probability distribution is generated using the joint risk factor probability model, and the key risk factor value of the key risk factor is determined based on the joint probability distribution. The fluctuation range of the key risk factor is determined based on the value of the key risk factor using Monte Carlo simulation, and the distribution data of financial indicators is generated by simulation. The financial indicator distribution data is input into the budget constraint model, which determines the elasticity range of the budget indicators based on a preset objective function and the fluctuation range.

2. The method according to claim 1, wherein the method comprises: Monitor real-time data of the key risk factors and calculate the conditional value of risk of the key risk factors based on the real-time data; When the conditional risk value of the key risk factor exceeds a preset risk threshold, a risk hedging decision is provided through the trained risk hedging model.

3. The method according to claim 2, wherein the algorithm of the risk hedging model includes: Random forest algorithm or reinforcement learning algorithm; Risk exposure assessment is performed on the key risk factors to determine the impact of changes in the key risk factors on the distribution data of the financial indicators, and the risk exposure assessment results are obtained. The risk exposure assessment results, risk hedging toolkit parameters, and budget constraints are used as inputs to train the risk hedging model until the hedging costs and benefits of the risk hedging decision reach the preset target values.

4. The method according to claim 1, wherein the probability distribution of the risk factors comprises: Normal probability density, log-normal probability density, and t-distribution.

5. The method according to claim 1, wherein determining the fluctuation range of the key risk factor based on the key risk factor value using Monte Carlo simulation and simulating the distribution data of financial indicators includes: Financial indicator distribution data is generated through iterative iterations using a preset maximum number of iterations. Risk factor values ​​are obtained from the joint probability distribution, financial indicator distribution data are calculated using the risk factor values, and the financial indicator distribution data is recorded. Continue iterating to generate financial indicator distribution data until the maximum number of iterations is reached.

6. The method according to claim 5 further includes performing statistical analysis on the financial indicator distribution data to determine the statistical values ​​of the financial indicator distribution data; The data characteristics of the financial indicator distribution data are determined based on the statistical values ​​of the financial indicator distribution data. Based on the data characteristics, the financial performance of the financial indicator distribution data is determined.

7. The method according to claim 1, wherein the internal data of the target unit includes: Historical purchase prices, sales data, and inventory turnover rate; The external market data includes: copper prices, the central bank's exchange rate midpoint, and industry policy documents; The target unit's internal data and the external market data are preprocessed, including missing value imputation, outlier correction, and time series decomposition.

8. The method according to claim 1, establishing a joint risk factor probability model comprising a joint risk factor probability distribution including multiple joint key risk factors, comprising: The distribution of each key risk factor was determined using the KS test. Based on the distribution of each key risk factor, a joint risk factor probability model is established using the Copula function, and the parameters in the joint risk factor probability model are determined using the maximum likelihood estimation method.

9. A risk quantification-driven flexible budgeting model and intelligent decision support system, the system comprising: An initial unit is used to acquire initial budget-related data, including internal data of the target unit and external market data. The initial data is preprocessed to generate budget decision data; A unit is established to identify key risk factors in the budget decision data, determine the risk factor probability distribution of the key risk factors, and establish a joint risk factor probability model that includes the joint risk factor probability distribution of multiple joint key risk factors. The determining unit is used to generate a joint probability distribution through the joint risk factor probability model, and determine the key risk factor value of the key risk factor based on the joint probability distribution; The generation unit is used to determine the fluctuation range of the key risk factor based on the key risk factor value through Monte Carlo simulation, and to simulate and generate financial indicator distribution data. The result unit is used to input the financial indicator distribution data into the budget constraint model, which determines the elasticity range of the budget indicator based on a preset objective function and the fluctuation range.

10. The system of claim 9, wherein the system includes a hedging unit for: Monitor real-time data of the key risk factors and calculate the conditional value of risk of the key risk factors based on the real-time data; When the conditional risk value of the key risk factor exceeds a preset risk threshold, a risk hedging decision is provided through the trained risk hedging model.

11. The system according to claim 10, wherein the algorithm of the risk hedging model includes: Random forest algorithm or reinforcement learning algorithm; The hedging unit is used to assess the risk exposure of the key risk factors, determine the impact of changes in the key risk factors on the distribution data of the financial indicators, and obtain the risk exposure assessment results. The risk exposure assessment results, risk hedging toolkit parameters, and budget constraints are used as inputs to train the risk hedging model until the hedging costs and benefits of the risk hedging decision reach the preset target values.

12. The system according to claim 9, wherein the risk factor probability distribution comprises: Normal probability density, log-normal probability density, and t-distribution; The establishment unit is also used to determine the parameters in the Gaussian Copula function by the maximum likelihood estimation method, determine the joint variation relationship between joint risk factors by the Gaussian Copula function with determined parameters, and establish a joint risk factor probability model.

13. The system according to claim 9, wherein the generation unit is configured to determine the fluctuation range of the key risk factor based on the key risk factor value using Monte Carlo simulation, and to simulate and generate financial indicator distribution data, and is further configured to: Financial indicator distribution data is generated through iterative iterations using a preset maximum number of iterations. Risk factor values ​​are obtained from the joint probability distribution, financial indicator distribution data are calculated using the risk factor values, and the financial indicator distribution data is recorded. Continue iterating to generate financial indicator distribution data until the maximum number of iterations is reached.

14. The system according to claim 13, wherein the generating unit is further configured to perform statistical analysis on the financial indicator distribution data to determine the statistical values ​​of the financial indicator distribution data; The data characteristics of the financial indicator distribution data are determined based on the statistical values ​​of the financial indicator distribution data. Based on the data characteristics, the financial performance of the financial indicator distribution data is determined.

15. The system according to claim 9, wherein the internal data of the target unit includes: Historical purchase prices, sales data, and inventory turnover rate; The external market data includes: copper prices, the central bank's exchange rate midpoint, and industry policy documents; The target unit's internal data and the external market data are preprocessed, including missing value imputation, outlier correction, and time series decomposition.

16. The system according to claim 9, wherein the establishing unit is configured to establish a joint risk factor probability model including a joint risk factor probability distribution of multiple joint key risk factors, and is further configured to: The distribution of each key risk factor was determined using the KS test. Based on the distribution of each key risk factor, a joint risk factor probability model is established using the Copula function, and the parameters in the joint risk factor probability model are determined using the maximum likelihood estimation method.

Images