System and method for financial instrument modeling and using monte carlo simulation

• US 20020010667A1
• Filed: 01/03/2001
• Published: 01/24/2002
• Est. Priority Date: 08/21/1997
• Status: Active Grant

• Current Assignee: Curt Randall, Elaine Kant
• Original Assignee: Curt Randall, Elaine Kant
• Inventors: Kant, Elaine, Randall, Curt
• Granted Patent: US 6,772,136 B2
• Time in Patent Office: Days
• US Class Current: 705/35
• CPC Class Codes: G06F 17/13 Differential equations using digital differential analysers G06F7/64
• CPC Class Codes: G06F 8/10 Requirements analysis; Specification techniques
• CPC Class Codes: G06F 8/30 Creation or generation of source code
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes

Abstract

The software synthesis method and system of the present invention provides a problem solving environment for Monte Carlo simulations (or other concise mathematical description), common in engineering, finance, and science, which automatically transforms a problem description into executable software code. The method and system uses a specification language to support a user'"'"'s natural description of the geometry and mathematics of the problem and solution strategies. The natural description is concisely expressed using general coordinates and dimensionless parameters, using domain specific keywords as appropriate. The user'"'"'s problem description is compared with the system'"'"'s knowledge base to refine the problem—i.e., identifying constraints, applying heuristics and defaults, and applying elaboration rules. The software synthesis method and system uses planning process, computer algebra, and templates to analyze and optimize the problem description, choose and customize data structures, and generate pseudo-code. The pseudo-code is translated into the desired target language source code. The software synthesis system and method therefore provides the ability to describe a problem and possible solution strategies at a high level, and outputs target language code that implements a solution. The software synthesis system and method is particularly useful modeling options where a Monte Carlo simulation is used.

Claims

1. A method for automatically generating software code in a target language for modeling and valuing financial instruments, comprising the steps of:

a) developing a problem specification for the financial instrument including—

defining the financial instrument as problem specification, comparing said problem specification with constraints within a knowledge base, invoking heuristics within said knowledge base to further define said problem specification;

b) formulating a financial model of the problem specification, including—

producing one or more templates associated with a Monte Carlo Simulation describing said financial model;

c) generating software code in said target language based on said templates; and

d) running said software code on a computer with user-defined variables to value said financial instrument.

2. The method of claim 1, said invoking heuristics substep comprising using the heuristics within said knowledge base unless in conflict with said problem specification.

3. The method of claim 1, said template also expressing procedural control structures and data declarations.

4. The method of claim 1, said knowledge base including a number of templates arranged as a hierarchy of classes.

5. The method of claim 4, said template classes including solvers, time evolution, interpolation routines, optimization routines, and time-stepping algorithms, singular value decomposition routines, pseudo-random and quasi-random number generation routines, routines for generation of samples from various statistical distributions, Brownian Bridge path generation routines, Latin hypercube sampling routines, single and multi-variate regression and back-substitution routines.

6. The method of claim 1, said developing step (a) including comparing the problem specification with heuristics within said knowledge base to determine if the problem specification is complete.

7. The method of claim 6, said developing step (a) including invoking user feedback if the problem specification is incomplete and if a constraint is violated.

8. The method of claim 1, said formulating step (b) including specifying an evolution algorithm in an evolution template.

9. The method of claim 8, said evolution template comprising a fixed time step.

10. The method of claim 8, said evolution template comprising a variable time step.

11. The method of claim 8, said evolution template comprising a time discretization.

12. The method of claim 1, said generating step (c) includes developing a data structure using said problem specification and said template, and generating pseudo-code describing said user problem using said template and data structure.

13. The method of claim 12, including translating said pseudo-code into executable software code in said target language.

14. A system for valuing a financial instrument which automatically generates software code based on a user-defined specification of said financial instrument, comprising:

a knowledge base containing constraints, heuristics and defaults for developing a finance problem specification based on said user-defined specification and said constraints, heuristics and defaults;

a computer algebra system for writing at least discrete portions of a Monte Carlo Simulation indicative of said finance problem specification;

one or more templates describing said stochastic differential equations, describing evolution of said stochastic differential equations, and generating pseudo-code reflecting said Monte Carlo Simulation; and

a code generator which generates software code in a target language from said pseudo-code for valuing a financial instrument based on said user-defined specification.

15. A method for modeling a finance problem useful for valuing financial instruments associated with the problem, comprising the steps of:

assembling a set of algorithm templates some of which represent a Monte Carlo Simulation describing the finance problem,

the templates being human readable declarations of said Monte Carlo Simulation independent of data structure and target language specifics,

one or more of said templates including pseudo-code expressions including matrix operators, scalar operators, or Boolean operators;

filling in the problem attributes in said templates including invoking a knowledge base having constraints, heuristics, and defaults to refine design decisions of the finance problem; and

building pseudo-code from said templates which models said finance problem.

16. The method of claim 15, each template being represented as a class in a hierarchy of template classes in a knowledge base.

17. The method of claim 15, names for variables and place-holders in a template being drawn from other templates.

18. The method of claim 15, said constraints comprising requirements for the finance problem.

19. The method of claim 15, said heuristics comprising suggestions for the finance problem.

20. The method of claim 15, said defaults comprising problem attributes used for the finance problem in the absence of user definitions or heuristics.