Risk management system and method for determining risk characteristics explaining heavy tails of risk factors

• US 7,778,897 B1
• Filed: 01/10/2003
• Issued: 08/17/2010
• Est. Priority Date: 01/11/2002
• Status: Active Grant

• Current Assignee: FinAnalytica, Incorporated (FactSet Research Systems Incorporated)
• Original Assignee: FinAnalytica, Incorporated (FactSet Research Systems Incorporated)
• Inventors: Hristov, Boris Blagoev, Racheva-Iatova, Boryana Svetlozarova, Rachev, Svetlozar Todorov
• Primary Examiner(s): Kazimi; Hani
• Assistant Examiner(s): Vyas; Abhishek
• Application Number:
  US10/339,580
• Time in Patent Office: 2,776 Days
• Field of Search: 705 35- 45
• US Class Current: 705/35
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

A risk management system and method of determining risk characteristics of portfolios by generating risk factor scenarios based on stable Paretian distributions is provided. The system includes a database for storing any needed input, output, and intermediate results data. At least one module from each of the predefined types can be employed to produce a corresponding result—either intermediate or final. The risk management system and method provide possibilities to use flexible multivariate distribution exhibiting heavy-tails, skewness and different dependence structure, and characterizing the whole distribution of a given financial variable, allows users to perform flexible stress tests on all calculation levels, supplies fast analysis due to the possibility to use intermediate results in many subsequent calculations and to employ several calculation modules in parallel, allows for an efficient enterprise-wide risk management because of the possibility intermediate results to be used by many users.

Claims

1. A computer-implemented method for estimating risk characteristics of portfolios of financial instruments, the computer-implemented method comprising the steps of:

maintaining a database of electronically stored information comprising a plurality of scenarios, sets of instruments, sets of risk factors, and portfolio weights for the instruments;

selecting a required number of scenarios, a set of instruments with corresponding amounts, a set of risk factors, and one or more sets of portfolio weights for each of the set of instruments from the database by a user through an input device for input into a memory;

maintaining a network operatively coupled and providing communication between the database, the input device, and a computer processor coupled to the memory;

processing the selected information by the computer processor coupled to the memory, comprising;

transforming historical observations for the set of risk factors to a stationary homogenous time-series comprising transformed historical observations;

obtaining stable Paretian parameters describing a heavy tailed and asymmetric distribution for the transformed historical observations in the set of risk factors by estimating stable Paretian parameters based on the transformed historical observations;

generating a set of stable Paretian scenarios from the heavy tailed and asymmetric distribution for the set of risk factors by simulating dependent stable Paretian random variables based on the estimated Paretian parameters, wherein the distribution is performed in accordance with the equation;

2. The method of claim 1 wherein the step of generating a set of stable Paretian scenarios from the heavy tailed and asymmetric distribution for the set of risk factors comprises the steps of:

converting the set of transformed historical observations for each risk factor in the set of risk factors to a uniform distribution by applying at least one of stable Paretian distribution function with the obtained stable Paretian parameters and empirical distribution function to each observation of the set of transformed historical observations;

obtaining distribution parameters of a chosen class of copula distribution by using estimations based on data resulting from the converting step for the set of transformed historical observations;

obtaining copula scenarios based on the copula distribution by generating random variables following the copula distribution;

generating scenarios with normal one-dimensional marginal distributions and the copula distribution by applying inverse normal distribution function to each component of the obtained copula scenarios;

obtaining stable subordinator simulations for each risk factor in the set of risk factors by generating stable subordinator simulations based on the obtained stable Paretian parameters of the risk factor in the set of risk factors; and

obtaining the set of stable Paretian scenarios for the set of risk factors by multiplying each component of each of the obtained normal scenarios with the corresponding stable subordinator simulation.

3. The method of claim 1 wherein the step of generating a set of stable Paretian scenarios from the heavy tailed and asymmetric distribution for the set of risk factors comprises the steps of:

converting the set of transformed historical observations for each risk factor in the set of risk factors to a uniform distribution by applying at least one of stable Paretian distribution function with the obtained stable Paretian parameters and empirical distribution function to each observation of the set of transformed historical observations;

obtaining distribution parameters of a chosen class of copula distribution, by using estimations based on data resulting from the converting step for the set of transformed historical observations;

obtaining copula scenarios based on the copula distribution by generating random variables following the copula distribution; and

obtaining the set of stable Paretian scenarios for the set of risk factors by applying to each component of each of the obtained copula scenarios the inverse stable Paretian distribution function with the estimated stable Paretian parameters for the corresponding risk factor.

4. The method of claim 1 wherein the step of transforming historical observations is performed in parallel on subsets of the set of risk factors.

5. The method of claim 1 wherein the step of obtaining stable Paretian parameters is performed in parallel on subsets of the set of risk factors.

6. The method of claim 1 wherein the step of transforming the stable Paretian scenarios is performed in parallel on subsets of the set of risk factors.

7. The method of claim 1 wherein the step of estimating instrument characteristics is performed in parallel on subsets of the set of instruments.

8. The method of claim 2 wherein the converting step is performed in parallel on subsets of the set of risk factors.

9. The method of claim 2 wherein the step of obtaining stable subordinator simulations is performed in parallel on subsets of the set of risk factors.

10. The method of claim 2 wherein the step of obtaining the set of stable Paretian scenarios is performed in parallel on subsets of the set of risk factors.

11. The method of claim 2 wherein the step of obtaining copula scenarios is performed in parallel on subsets of the set of scenarios.

12. The method of claim 2 wherein the step of generating scenarios is performed in parallel on subsets of the set of scenarios.

13. The method of claim 2 wherein the step of obtaining stable subordinator simulations is performed in parallel on subsets of the set of scenarios.

14. The method of claim 3 wherein the converting step is performed in parallel on subsets of the set of risk factors.

15. The method of claim 3 wherein the step of obtaining the set of stable Paretian scenarios is performed in parallel on subsets of the set of risk factors.

16. The method of claim 3 wherein the step of obtaining copula scenarios is performed in parallel on different subsets of scenarios.

17. The method of claim 3 wherein the step of obtaining the set of stable Paretian scenarios is performed in parallel on different subsets of scenarios.

18. The method of claim 1 wherein each step is performed by at least two users, each of the two users producing risk characteristics for generally different selected sets of instruments.

19. A computer-implemented risk management system for estimating risk characteristics, comprising:

a database of electronically stored information comprising a plurality of scenarios, sets of instruments, sets of risk factors, and portfolio weights for the instruments;

an input device to receive selections from a user of a required number of scenarios, a set of instruments with corresponding amounts, a set of risk factors, and one or more sets of portfolio weights for each of the set of instruments from the database by a user through an input device for input into a memory;

a network operatively coupled and providing communication between the database, memory, the input device, and a computer processor coupled to the memory, wherein the processor is programmed to execute program modules, the program modules comprising;

at least one Transformation Module from each type of Transformation Module out of a pre-defined non-empty set of Transformation Module types that transforms a set of risk factors'"'"' historical observations into a stationary homogenous time-series;

a Stable Scenarios Module comprising;

at least one Stable Parameters Estimation Module from each type out of a pre-defined non-empty set of Stable Parameters Estimation Module types, that estimates stable Paretian parameters describing a heavy tailed and asymmetric distribution over a set of time-series data;

at least one Copula Parameters Estimation Module from each type out of a pre-defined non-empty set of Copula Parameters Estimation Module types, that estimates copula parameters for a given class of copula distributions over a set of time-series data;

at least one Stable Simulations Module from each type out of a pre-defined non-empty set of Stable Simulations Module types, that generates stable Paretian random scenarios from the heavy tailed and asymmetric distribution, wherein the distribution is performed in accordance with the equation;

20. A risk management system of claim 19, wherein the system comprises at least two of the Transformation Modules, each of the at least two Transformation Modules operating in parallel to transforms risk factors'"'"' historical observations for different sub-sets of risk factors.

21. A risk management system of claim 19, wherein the system comprises at least two of the Inverse Transformation Modules, each of the at least two Inverse Transformation Modules operating in parallel to transforms risk factors'"'"' stable Paretian scenarios comprising the heavy tailed and asymmetric distribution for different sub-sets of risk factors.

22. A risk management system of claim 19, wherein the system comprises at least two of the Stable Parameters Estimation Modules, each of the at least two Stable Parameters Estimation Modules operating in parallel to produce stable Paretian parameters describing a heavy tailed and asymmetric distribution for different sub-sets of risk factors.

23. A risk management system of claim 19, wherein the system comprises at least two of the Stable Simulation Modules, each of the at least two Stable Simulation Modules operating in parallel to produce stable Paretian scenarios for at least one of different sub-sets of the required set of scenarios and different sub-sets of risk factors.

24. A risk management system of claim 19, wherein the system comprises at least two of the Copula Parameters Estimation Modules, each of the at least two Copula Parameters Estimation Modules operating in parallel to produce copula parameters for different sub-sets of risk factors.

25. A risk management system of claim 19, wherein the system comprises at least two of the Copula Simulation Modules, each of the at least two Copula Simulation Modules operating in parallel to produce copula scenarios for different sub-sets of the required set of copula scenario.

26. A risk management system of claim 19, wherein the system comprises at least two of the Normal Distribution Modules, each of the at least two Normal Distribution Modules operating in parallel to transform a random number having uniform distribution to a random number having normal distribution and vice versa for different sub-set random numbers.

27. A risk management system of claim 19, wherein the system comprises at least two of the Instrument Characteristics Estimation Modules, each of the at least two Instrument Characteristics Estimation Modules operating in parallel to produce instrument characteristic scenarios under different sub-sets of stable Paretian scenarios.

28. A risk management system of claim 19, wherein the system comprises at least two of the Risk Characteristics Estimation Modules, each of the at least two Risk Characteristics Estimation Modules operating in parallel to produce risk characteristic under different sub-sets of instrument characteristic scenarios.