System and method for providing reallocation and reverse optimization of a financial portfolio using a parametric leptokurtic distribution

• US 7,890,409 B2
• Filed: 05/03/2010
• Issued: 02/15/2011
• Est. Priority Date: 07/11/2003
• Status: Active Grant

• Current Assignee: FinAnalytica, Incorporated (FactSet Research Systems Incorporated)
• Original Assignee: FinAnalytica, Incorporated (FactSet Research Systems Incorporated)
• Inventors: Stoyanov, Stoyan Veselinov, Racheva-Iotova, Boryana S., Rachev, Svetlozar Todorov, Martin, Richard Douglas
• Primary Examiner(s): Kalinowski; Alexander
• Assistant Examiner(s): Vyas; Abhishek
• Application Number:
  US12/772,936
• Publication Number:
  US 20100332411A1
• Time in Patent Office: 288 Days
• Field of Search: 705 35- 37
• US Class Current: 705/36.00R
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/04 Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis

Abstract

A system and method for providing reallocation and reverse optimization of a financial portfolio using a parametric leptokurtic distribution are provided. A time series including risk factors applicable over at least one time horizon, a portfolio including financial assets, a quantile, and one or more risk adjusted return points for the financial assets is stored. The financial assets are associated with the risk factors. A subordinated parametric distribution model exhibiting leptokurtic behavior is generated. A function of expected tail loss for the quantile based on the subordinated parametric distribution model exhibiting leptokurtic behavior is expressed. A set of portfolio asset weight changes for each of the financial assets is determined based on the expected tail loss at each such time horizon and for each risk adjusted return point. Reallocation of the portfolio is provided based on the portfolio asset weight changes for each risk adjusted return point for the portfolio.

Claims

1. A computer-implemented system for providing reallocation and reverse optimization of a financial portfolio using a parametric leptokurtic distribution, comprising:

a memory containing a database configured to store a time series comprising a plurality of risk factors applicable over at least one time horizon, a portfolio comprising a plurality of financial assets, a quantile, and one or more risk adjusted return points for the financial assets;

an input device to receive input from a user;

a network operatively coupled and providing communication between the memory, the input device, an output device, and a processor;

where the processor is programmed to execute program modules, the program modules comprising;

an association module configured to associate the financial assets with the risk factors;

a risk determining module configured to generate a subordinated parametric distribution model exhibiting leptokurtic behavior;

an expected tail loss determining module configured to express a function of expected tail loss for the quantile based on the parametric distribution model exhibiting leptokurtic behavior, wherein the expected tail loss, ETL(x,ε

), is defined in accordance with the equation;

2. A system according to claim 1, wherein the portfolio assets weights are determined using Marginal Contribution to ETL (MCETL) and Implied Return (IR) based on the parametric leptokurtic distribution.

3. A system according to claim 2, wherein the IR is subject to one or more ETL risk-adjusted performance ratios.

4. A system according to claim 3, wherein the one or more ETL risk-adjusted performance ratios are selected from one of STARR, Rachev Ratio, and Generalized Rachev Ratio.

5. A system according to claim 2, wherein the MCETL is defined according to the equation:

MCETL_i=E(L_i|L_p≧

VaR<sub>α

%</sub>)

where L_p is the portfolio loss, L_i is the loss of the i-th portfolio constituent VaR<sub>α

%</sub> is the portfolio VaR at the α

% quantile, and E is the conditional expected value of L_i given that L_p is greater or equal to VaR<sub>α

%</sub>.

6. A system according to claim 5, wherein the IR is defined according to the equation:

IR_i=λ

×

MCETL_i

where IR_i is one of implied return of the i-th asset and expected implied return of the i-th asset over the threshold, and λ

equals one of the STARR ratio of the portfolio and the Rachev ratio or Geheralized Rachev ratio of the portfolio, respectively.

7. A system according to claim 1, wherein the subordinated parametric distribution model is selected from the group comprising an alpha-Stable distribution, T-distribution, sub-Gaussian Stable distribution, and mixture of multivariate Normal distributions.

8. A system according to claim 1, wherein the subordinated parametric distribution model further has an asymmetric behavior.

9. A system according to claim 1, further comprising:

a clustering of volatility function, wherein the system is configured to apply the clustering of volatility function along a time series of returns of the risk factors.

10. A system according to claim 9, wherein the clustering of volatility function is selected from the group comprising a multivariate ARMA-GARCH, comprising an exponential weighted moving average algorithm, applied to a time series of risk factor returns and producing innovations having leptokurtotic distributions.

11. A system according to claim 1, wherein the risk factors are selected from the group comprising market price returns, derived values and exogenous factors.

12. A system according to claim 1, wherein a plurality of time horizons are specified.

13. A system according to claim 1, wherein the portfolio is structured into a hierarchy of subportfolios.

14. A system according to claim 1, wherein estimates of returns are adjusted for returns of a risk free asset or a benchmark portfolio.

15. A system according to claim 1, wherein a penalty function is used to adjust returns.

16. A computer-implemented method for providing reallocation and reverse optimization of a financial portfolio using a parametric leptokurtic distribution, the computer-implemented method comprising the steps of:

storing a time series comprising a plurality of risk factors applicable over at least one time horizon, a portfolio comprising a plurality of financial assets, a quantile, and one or more risk adjusted return points for the financial assets in a database contained in a memory;

maintaining a network operatively coupled and providing communication between the memory, an input device, an output device, and a processor,

wherein the processor is configured to execute computer executable code stored in program modules, comprising;

associating the financial assets with the risk factors;

generating a subordinated parametric distribution model exhibiting leptokurtic behavior;

expressing a function of expected tail loss for the quantile based on the subordinated parametric distribution model exhibiting leptokurtic behavior, wherein the expected tail loss, ETL(x,ε

) is defined in accordance with the equation;

17. A method according to claim 16, wherein the portfolio assets weights are determined using Marginal Contribution to ETL (MCETL) and Implied Return (IR) based on the parametric leptokurtic distribution.

18. A method according to claim 17, wherein the IR is subject to one or more ETL risk-adjusted performance ratios.

19. A method according to claim 18, wherein the one or more ETL risk-adjusted performance ratios are selected from one of STARR, Rachev Ratio, and Generalized Rachev Ratio.

20. A method according to claim 17, wherein the MCETL is defined according to the equation:

MCETL_i=E(L_i|L_p≧

VaR<sub>α

%</sub>)

where L_p is the portfolio loss, L_i is the loss of the i-th portfolio constituent, VaR<sub>α

%</sub> is the portfolio VaR at the α

% quantile, and E is the conditional expected value of L_i given that L_p is greater or equal to VaR<sub>α

%</sub>.

21. A method according to claim 20, wherein the IR is defined according to the equation:

IR_i=λ

×

MCETL_i

where IR_i is one of implied return of the i-th asset and expected implied return of the i-th asset over the threshold, and λ

equals one of the STARR ratio of the portfolio and the Rachev ratio or Generalized Rachev ratio of the portfolio, respectively.

22. A method according to claim 16, wherein the subordinated parametric distribution model is selected from the group comprising an alpha-Stable distribution, T-distribution, sub-Gaussian Stable distribution, and mixture of multivariate Normal distributions.

23. A method according to claim 16, wherein the subordinated parametric distribution model further has an asymmetric behavior.

24. A method according to claim 16, further comprising:

specifying a clustering of volatility function; and

applying the clustering of volatility function along a time series of returns of the risk factors.

25. A method according to claim 24, wherein the clustering of volatility function is selected from the group comprising a multivariate ARMA-GARCH, comprising an exponential weighted moving average algorithm, applied to a time series of risk factor returns and producing innovations having leptokurtotic distributions.

26. A method according to claim 16, wherein the risk factors are selected from the group comprising market price returns, derived values and exogenous factors.

27. A method according to claim 16, wherein a plurality of time horizons are specified.

28. A method according to claim 16, further comprising:

structuring the portfolio into a hierarchy of subportfolios.

29. A method according to claim 16, wherein estimates of returns are adjusted for returns of a risk free asset or a benchmark portfolio.

30. A method according to claim 16, wherein a penalty function is used to adjust returns.