CN104463662A - Financial data calculation method and device - Google Patents
Financial data calculation method and device Download PDFInfo
- Publication number
- CN104463662A CN104463662A CN201410722408.XA CN201410722408A CN104463662A CN 104463662 A CN104463662 A CN 104463662A CN 201410722408 A CN201410722408 A CN 201410722408A CN 104463662 A CN104463662 A CN 104463662A
- Authority
- CN
- China
- Prior art keywords
- data
- finance
- finance data
- parallel processing
- monte carlo
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention discloses a financial data calculation method. The financial data calculation method comprises the steps that financial data are obtained; whether the financial data accord with a preset data rule is verified; when the financial data accord with the preset data rule, parallel processing is conducted in the Monte Carlo calculation process of the data by means of multiple java parallel processing framework JPPF nodes. The invention further discloses a financial data calculation device. By the adoption of the financial data calculation method and device, parallel processing is conducted in the Monte Carlo calculation process of the data by means of the multiple java parallel processing framework JPPF nodes, in this way, time for frequent Monte Carlo calculation is shortened when the data volume is large, and mass financial data can be calculated rapidly and accurately.
Description
Technical field
The present invention relates to capital and calculate field, particularly relate to a kind of finance data computing method and device.
Background technology
Economic capital is in certain level of confidence, certain hour, in order to make up bank Unexpected Losses required for capital, be bank be the capital amount needed of accepting the risk, the feature of risk of bank self can be reflected completely.
It is based on the Output rusults of apparatus of grading (PD LGD the parameter such as EAD) that economic capital calculates, in conjunction with the measurement results of the assets correlativity based on external data, measure credit losses distribution from bottom to up, calculate the credit risk economic capital needed for whole asset portfolio further.Economic capital calculates based on promise breaking pattern model, economic capital and the expected loss of this loan portfolio is calculated by Monte Carlo simulation, decompose each liability by economic capital, gather according to each dimension, calculate economic capital needed for each dimension.
Monte-carlo Simulation Method is a kind of take Probability Statistics Theory as the very important numerical computation method of a class instructed, the step once calculated is comparatively loaded down with trivial details, when carrying out the calculating of economic capital, data volume is usually very large, utilizes Monte-carlo Simulation Method directly to carry out the calculating of big data quantity very consuming time.
Summary of the invention
In view of this, fundamental purpose of the present invention is to provide a kind of finance data computing method and device, can calculate finance data rapidly and accurately.
For achieving the above object, the invention provides a kind of finance data computing method, comprising:
Obtain finance data;
Verify whether described finance data meets preset data rule;
When described finance data meets described preset data rule, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
Preferably, described acquisition finance data comprises:
Extracted the Output rusults data of grading apparatus by data pick-up, conversion, loading ETL mode, obtain described finance data.
Preferably, utilize multiple java parallel processing frame J PPF node, the Monte Carlo calculations of parallel processing to described data is crossed Cheng Qian and is also comprised:
The parameter that the finance judging whether to revise acquiescence calculates, if so, then the parameter that the finance changing acquiescence calculates, if not, is then defined as target component by the parameter that the finance of acquiescence calculates.
Preferably, described Monte Carlo calculations process comprises:
Determine cycle index N and number realization K, calculate promise breaking threshold value, carry out K simulation, the loss of whole liability is obtained according to K analog result, according to whole liability costing bio disturbance expected loss, according to the fractile of whole liability costing bio disturbance statistic fiducial interval, fractile deducts expected loss and obtains result;
Described simulation comprises:
The tchaikovsky carrying out assets correlativity decomposes, and generates a presetting numerical value standard normal substep random number, produces the random number sequence that assets are relevant, carry out N cycle calculations, calculate the loss of whole liability;
Described cycle calculations comprises:
Generate a standard normal substep random number, calculate Return on Assets, according to Return on Assets and the loss of promise breaking threshold value determination liability.
Preferably, utilize multiple java parallel processing frame J PPF node, parallel processing carries out importance sampling simulation to also comprising after the Monte Carlo calculations process of described data.
Preferably, also comprise after carrying out importance sampling simulation according to described parameter and the finance data calculated is distributed.
Preferably, also comprise after the finance data calculated being distributed and result of calculation is shown.
Present invention also offers a kind of finance data calculation element, comprising:
Data capture unit, for obtaining finance data;
Authentication unit, for verifying whether described finance data meets preset data rule;
Processing unit, during for meeting described preset data rule when described finance data, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
Preferably, described finance data calculation element also comprises parameter configuration unit, for configuring the parameter of carrying out finance data calculating.
Preferably, described finance data calculation element also comprises result display unit, for showing result of calculation.
Apply finance data computing method provided by the invention and device, utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data, the Monte Carlo calculations time that when saving big data quantity, number of times is various, a large amount of finance data can be calculated rapidly and accurately.
Accompanying drawing explanation
In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only embodiments of the invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to the accompanying drawing provided.
Fig. 1 is the process flow diagram of a kind of finance data computing method of the present invention;
Fig. 2 is the another process flow diagram of a kind of finance data of the present invention computing method;
Fig. 3 is the another process flow diagram of a kind of finance data of the present invention computing method;
Fig. 4 is the process flow diagram of the present invention one specific embodiment;
Fig. 5 is the theory diagram of a kind of finance data computing method of the present invention specific embodiment;
Fig. 6 is a kind of finance data computing method of the present invention specific embodiment economic capital gauge assembly interaction figure;
Fig. 7 is the schematic diagram of the Monte Carlo calculations process of a kind of finance data computing method of the present invention;
Fig. 8 is the structural representation of a kind of finance data calculation element of the present invention;
Fig. 9 is the another structural representation of a kind of finance data of the present invention calculation element;
Figure 10 is the another structural representation of a kind of finance data of the present invention calculation element.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
The invention provides a kind of finance data computing method, as shown in Figure 1, be the process flow diagram of finance data computing method embodiment of the present invention, it is characterized in that, comprising:
Step S101: obtain finance data;
Step S102: verify whether described finance data meets preset data rule;
Step S103: when described finance data meets described preset data rule, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
The finance data computing method that application the present embodiment provides, utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data, the Monte Carlo calculations time that when saving big data quantity, number of times is various, a large amount of finance data can be calculated rapidly and accurately.
As shown in Figure 2, be the method flow diagram of the specific embodiment of the invention, be applied in the economic capital calculating of bank, corresponding to Fig. 1, obtain finance data and specifically comprise:
Step S201: the Output rusults data being extracted grading apparatus by data pick-up, conversion, loading ETL mode, obtain described finance data.
Also add step S204: the parameter of Monte Carlo calculations is carried out in configuration.
As shown in Figure 3, be the method flow diagram of further embodiment of this invention, be applied in the economic capital calculating of bank, corresponding to Fig. 2, the parameter that Monte Carlo calculations is carried out in configuration specifically comprises:
Step S304: the parameter judging whether the Monte Carlo calculations revising acquiescence, if so, then enters step S305, if not, then enter step S306.
Step S305: the parameter of the Monte Carlo calculations of change acquiescence;
Step S306: the parameter of the Monte Carlo calculations of acquiescence is defined as target component.
Also add step S308: result of calculation is shown.
As described in Figure 4, for the method flow diagram of the another specific embodiment of the present invention, the economic capital of bank is calculated, the present embodiment economic capital calculates based on the Output rusults of device system of grading (PD LGD the parameter such as EAD), the preliminary work calculating data has been needed before calculating, extracted the data required for economic capital calculating of up-stream system propelling movement by ETL (i.e. the process of data pick-up, conversion, loading), and complete data check according to data rule verification engine; The initiation of economic capital calculation task can be carried out afterwards.
Economic capital calculation procedure:
1) Monte Carlo simulation, this part is the population distribution in order to obtain liability assembling loss, thus obtains the penalty values L of corresponding economic capital objective degrees of confidence
0.Simultaneously based on this loss distribution, determine a penalty values calculated for next step, and a series of (n) threshold value L
i, i=1,2 ..., n.These threshold values will be used for the afterbody determining the loss distribution that importance sampling is simulated.Under normal circumstances, L
μl can be less than
i, L
ibe less than L
0;
2) calculate μ value, μ value is the important parameter for importance sampling simulation, the intermediate value of his systemic factor stochastic distribution under illustrating probability metrics new in importance sampling analogue technique;
3) importance sampling simulates this part is by importance sampling technology, simulates the afterbody of liability loss distribution, namely corresponding with the threshold value Li in first step loss probability Probi.According to this series of loss probability and corresponding threshold value, can the penalty values L. of corresponding economic capital objective degrees of confidence be obtained by difference and then obtain the economic capital of whole liability combination;
4) economic capital distributes, and the economic capital calculated is decomposed every liability, distributes economic capital: based on risk collocation method and the contributrion margin method of standard deviation, economic capital project mainly adopts the risk collocation method of standard deviation by two methods.
The present embodiment economic capital calculating realization mainly have employed Hessian lightweight frame and Spring batch processing framework is integrated, calculate the execution of batch processing task, Spring batch processing framework carries out batch processing by using the dependence of Spring to inject (dependency injection) to economic capital calculation task, and Spring is an Open Framework in order to the complicacy solving enterprise's application and development creates.Computing engines instrument JPPF parallel processing framework is have invoked for common model Carlow analog computing system, JPPF, i.e. java parallel processing framework, it is the Grid Computing Framework of an open source code, it can run multiple java simultaneously apply in a distribution execution environment, JPFF framework by client, service end and node three part form, the principle of work is that it performs multiple task matching to multiple node goes, by opening multiple JPPF nodal parallel process model Carlows computation process.JPPF provides load balance, the service such as Failure Transfer and Fault recovery, also provides one based on the supervisor console of JMX, and it both can monitor that node also can manage the task of execution, can long-range cancellation and restart task, or configuration makes cut-off date or the time interval of its time-out.
The present embodiment has carried out technical tuning process to JPPF, by opening multiple JPPF nodal parallel process model Carlows computation process, the analog computation time in common model's card of 3,000,000 times was compressed to 1 current hours from 72 several hours, and theory diagram as shown in Figure 5.
Be illustrated in figure 6 the present embodiment economic capital gauge assembly interaction figure, the present embodiment economic capital gauge assembly adopts the Design Mode towards interface, adopt interfaceization and rely on the method for designing injected, different functions is completed by the different implementors of interface, namely same interface difference realizes, that can find out in figure that dependence (calling) task parameters interface completes economic capital metering tasks model parameter arranges work, dependence task executive's interface is finished the work and is performed the read work of step, for preliminary work has been done in economic capital metering.Complete by calling economic capital metering batch processing service interface work is called to economic capital metering modules after preliminary work completes.
As shown in Figure 7, the economic capital method flow diagram that in calculating, common Monte Carlo simulation calculates, determine client's number N and the times N will carrying out Monte Carlo simulation calculating, the method step that common Monte Carlo simulation calculates comprises:
Step S401: the promise breaking threshold values calculating each client, calls NORMSINV (pd) (inverse function of Standard Normal Distribution) formulae discovery;
Start the computation process of K Monte Carlo simulation after completing steps S401, all clients all will be simulated one time by each simulation:
Step S402: the tchaikovsky of assets correlativity decomposes, and calls Cholesky () formula and matrix transpose is become lower triangular matrix;
Step S403: the standardized normal distribution random number generating d industry, calls ORMINV (RAND (), mean, standard_dav) formulae discovery;
Step S404: be multiplied with d industry random number by the matrix after transposition, produces the random number sequence that assets are relevant;
The all clients of completing steps S404 Posterior circle (N number of):
Step S405: the standardized normal distribution random number ε generating Customer Personality;
Step S406: the Return on Assets calculating client, ξ
k=a
kβ Z+b ε;
Step S407: if customer capital earning rate < customer default threshold values, then under this client, liability loss is:
(N is liability number under client);
Wherein: when not considering LGD undulatory property, LGD
jfor the loss given default of liability.
When considering LGD undulatory property, LGD
j=BETAINV (Rnd (), Alpha, Beta)
Alpha＝LGD*(LGD*(1-LGD)/LGD
vol/LGD
vol-1)
Otherwise loss is 0.
Step S408: calculate this simulation penalty values TL=L1+L2+.....Ln
Terminate this simulation after completing the circulation of N number of client, start next simulation process;
Terminate K simulation afterwards;
Step S409: the distribution results obtaining full row simulation loss, sorts from high to low according to the loss amount of money
Calculate full row expected loss: EL=PD × LGD × EaD;
Step S410: according to the fractile of loss result compute statistics fiducial interval, calls PERCENTILE function and calculates, by fractile as total loss result;
Step S411: be total economic capital after deducting expected loss EL with the total losses that fractile calculates.
Present invention also offers a kind of finance data calculation element, as shown in Figure 8, be finance data calculation element example structure schematic diagram, comprise:
Data capture unit 101, for obtaining finance data;
Authentication unit 102, is connected with described data capture unit 101, for verifying whether described finance data meets preset data rule;
Processing unit 103, is connected with described authentication unit 102, and during for meeting described preset data rule when described finance data, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
Apply finance data calculation element provided by the invention, utilize multiple java parallel processing frame J PPF node, parallel processing, to the Monte Carlo calculations process of described data, the Monte Carlo calculations time that when saving big data quantity, number of times is various, can calculate a large amount of finance data rapidly and accurately.
As shown in Figure 9, be a specific embodiment of finance data calculation element of the present invention, corresponding to Fig. 8, also comprise:
Parameter configuration unit 104, is all connected with processing unit 103 with described authentication unit 102, for configuring the parameter of carrying out Monte Carlo calculations.
As shown in Figure 10, be the another specific embodiment of finance data calculation element of the present invention, corresponding to Fig. 9, also comprise:
Result display unit 105, is connected with described processing unit 103, for showing result of calculation.
It should be noted that, each embodiment in this instructions all adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar part mutually see.For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part illustrates see the part of embodiment of the method.
Finally, also it should be noted that, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or equipment and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or equipment.When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the equipment comprising described key element and also there is other identical element.
Above finance data computing method provided by the present invention and device are described in detail, apply specific case herein to set forth principle of the present invention and embodiment, the explanation of above embodiment just understands method of the present invention and core concept thereof for helping; Meanwhile, for one of ordinary skill in the art, according to thought of the present invention, all will change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (10)
1. finance data computing method, is characterized in that, comprising:
Obtain finance data;
Verify whether described finance data meets preset data rule;
When described finance data meets described preset data rule, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
2. finance data computing method according to claim 1, is characterized in that, described acquisition finance data comprises:
Extracted the Output rusults data of grading apparatus by data pick-up, conversion, loading ETL mode, obtain described finance data.
3. finance data computing method according to claim 1, is characterized in that, utilize multiple java parallel processing frame J PPF node, the Monte Carlo calculations of parallel processing to described data is crossed Cheng Qian and also comprised:
The parameter that the finance judging whether to revise acquiescence calculates, if so, then the parameter that the finance changing acquiescence calculates, if not, is then defined as target component by the parameter that the finance of acquiescence calculates.
4. finance data computing method according to claim 3, is characterized in that, described Monte Carlo calculations process comprises:
Determine cycle index N and number realization K, calculate promise breaking threshold value, carry out K simulation, the loss of whole liability is obtained according to K analog result, according to whole liability costing bio disturbance expected loss, according to the fractile of whole liability costing bio disturbance statistic fiducial interval, fractile deducts expected loss and obtains result;
Described simulation comprises:
The tchaikovsky carrying out assets correlativity decomposes, and generates a presetting numerical value standard normal substep random number, produces the random number sequence that assets are relevant, carry out N cycle calculations, calculate the loss of whole liability;
Described cycle calculations comprises:
Generate a standard normal substep random number, calculate Return on Assets, according to Return on Assets and the loss of promise breaking threshold value determination liability.
5. finance data computing method according to claim 4, is characterized in that, utilize multiple java parallel processing frame J PPF node, parallel processing carries out importance sampling simulation to also comprising after the Monte Carlo calculations process of described data.
6. finance data computing method according to claim 5, is characterized in that, also comprise being distributed by the finance data calculated according to described parameter after carrying out importance sampling simulation.
7. finance data computing method according to claim 6, is characterized in that, also comprise and show result of calculation after being distributed by the finance data calculated.
8. a finance data calculation element, is characterized in that, comprising:
Data capture unit, for obtaining finance data;
Authentication unit, for verifying whether described finance data meets preset data rule;
Processing unit, during for meeting described preset data rule when described finance data, then utilize multiple java parallel processing frame J PPF node, parallel processing is to the Monte Carlo calculations process of described data.
9. finance data calculation element according to claim 8, is characterized in that, also comprise parameter configuration unit, for configuring the parameter of carrying out finance data calculating.
10. finance data calculation element according to claim 9, is characterized in that, also comprises result display unit, for showing result of calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410722408.XA CN104463662A (en) | 2014-12-02 | 2014-12-02 | Financial data calculation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410722408.XA CN104463662A (en) | 2014-12-02 | 2014-12-02 | Financial data calculation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104463662A true CN104463662A (en) | 2015-03-25 |
Family
ID=52909653
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410722408.XA Pending CN104463662A (en) | 2014-12-02 | 2014-12-02 | Financial data calculation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104463662A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220892A (en) * | 2017-05-27 | 2017-09-29 | 国家计算机网络与信息安全管理中心 | One kind melts data intelligence pretreating tool and method applied to magnanimity P2P net monetary allowances |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070174236A1 (en) * | 2006-01-20 | 2007-07-26 | Daniel Pagnussat | Technique for supplying a data warehouse whilst ensuring a consistent data view |
| CN101221647A (en) * | 2008-02-01 | 2008-07-16 | 中国建设银行股份有限公司 | Finance data processing method and system |
-
2014
- 2014-12-02 CN CN201410722408.XA patent/CN104463662A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070174236A1 (en) * | 2006-01-20 | 2007-07-26 | Daniel Pagnussat | Technique for supplying a data warehouse whilst ensuring a consistent data view |
| CN101221647A (en) * | 2008-02-01 | 2008-07-16 | 中国建设银行股份有限公司 | Finance data processing method and system |
Non-Patent Citations (2)
| Title |
|---|
| 亓慧艳: "面向金融风险管理的构件技术及应用研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 张雪峰: "欧美商业银行经济资本管理研究", 《中国博士学位论文全文数据库 经济与管理科学辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220892A (en) * | 2017-05-27 | 2017-09-29 | 国家计算机网络与信息安全管理中心 | One kind melts data intelligence pretreating tool and method applied to magnanimity P2P net monetary allowances |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pakes et al. | Computing markov perfect nash equilibria: Numerical implications of a dynamic differentiated product model | |
| Singh et al. | Entropy based software reliability analysis of multi-version open source software | |
| De Saporta et al. | Numerical methods for simulation and optimization of piecewise deterministic Markov processes: application to reliability | |
| CN101561904B (en) | A kind of software project cost assay method of Kernel-based methods data and system | |
| Stutzke et al. | A stochastic model of fault introduction and removal during software development | |
| Vallecillo et al. | Expressing measurement uncertainty in software models | |
| Pesenti et al. | Scenario Weights for Importance Measurement (SWIM)–an R package for sensitivity analysis | |
| CN104463662A (en) | Financial data calculation method and device | |
| Prasetyawan et al. | The integration between business model canvas and manufacturing system design | |
| Urgilés et al. | Methods for quantitative risks analysis of cost and deadline overruns in complex projects | |
| CN101674194A (en) | Cluster load model based on log feature analysis and modeling method thereof | |
| Anand et al. | Resource allocation problem for multi versions of software system | |
| JP2021077364A (en) | Project management system, project management method, and program | |
| Ramacharan et al. | Software effort estimation of GSD projects using calibrated parametric estimation models | |
| Bosse et al. | Predicting availability and response times of IT services | |
| Liang et al. | From cadcad to cascad2: A mechanism validation and verification system for decentralized autonomous organizations based on parallel intelligence | |
| Toledano et al. | On Markov reward approach to failure criticality importance assessment for aging multi-state system | |
| Attari-Shendi et al. | A comprehensive mathematical model for sequencing interrelated activities in complex product development projects | |
| CN106528998A (en) | Simulation based acquisition integrated model description method and simulation based acquisition integrated model realization system | |
| Rostampour et al. | Measures of structural complexity and service autonomy | |
| Cao | Existence of generalized recursive equilibrium in Krusell and Smith (1998) | |
| Gandhi et al. | Estimating reliability for OSS: An approach with change-point in operational phase | |
| KR20130124985A (en) | Risk management device | |
| Amjady et al. | Optimal reliable operation of hydrothermal power systems with random unit outages | |
| Potok et al. | A model of correlated team behaviour in a software development environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150325 |