Adjusted Factor-Based Performance Attribution

• US 20150081592A1
• Filed: 07/21/2014
• Published: 03/19/2015
• Est. Priority Date: 08/23/2013
• Status: Abandoned Application

• Current Assignee: Axioma, Inc. (Deutsche Borse AG)
• Original Assignee: Axioma, Inc. (Deutsche Borse AG)
• Inventors: Stubbs, Robert A., Jeet, Vishv
• Application Number:
  US14/336,123
• Publication Number:
  US 20150081592A1
• Time in Patent Office: Days
• US Class Current: 705/36.R
• CPC Class Codes: G06Q 10/067 Enterprise or organisation modelling
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis

Abstract

Performance attribution results of investment portfolios are often misleading due to correlation between the factor and specific contributions. This correlation is not correctly accounted for in standard factor-based attribution thus leading to potentially erroneous results. The present invention produces an adjusted factor-based performance attribution methodology that moves a portion of the specific return that is correlated with the factor contributions into the factor portion. This methodology adjusts the contribution to a subset of factors and to the specific contributions such that the resulting factor and specific contributions have small correlation.

Claims

7. A computer-implemented system for computing and reporting the performance attribution of a set of portfolio holdings over time comprising:

a memory for storing data for a set of dates defining an attribution time horizon to be performed;

a processor executing software to retrieve data for historical portfolios of holdings having investment weights in a set of investible assets at each date;

a processor executing software to retrieve data for a set of factors and a set of factor exposures for each investible asset in the historical portfolio of holdings as of that date;

a processor executing software to retrieve data or compute data for a factor return for each factor exposure as of that date;

a processer executing software to retrieve data or compute data for a specific return for all investible assets in the portfolio as of that date;

computing on the processor the factor contributions for each factor by combining the investment weights of the historical portfolios, the factor exposures, and the factor returns for each date;

computing on the processor the specific contributions by combining the investment weights of the historical portfolios and the specific returns for each date;

computing on the processor one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting on the processor a preferred mathematical model from those computed;

computing on the processor an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model for each date;

computing on the processor a performance attribution for the historical portfolios of holdings based on the adjusted set of factor and specific contributions; and

electronically outputting the performance attribution results on an output device.

8. The system of claim 7 in which the time series regression model is a linear function of a set of factor contributions.

9. The system of claim 8 in which a sequence of mathematical time series regression models is constructed that removes statistically insignificant factor contributions from the model at each iteration of the sequence.

10. The system of claim 7 in which an adjusted factor risk estimate is computed.

11. The system of claim 7 in which the factor exposures, factor returns, and specific returns are derived from a factor risk model.

12. The system of claim 11 in which a modified factor risk model is estimated using the adjusted factor and specific returns.

13. A computer-implemented method for computing and reporting factor and specific contributions for a set of portfolio holdings over time comprising:

electronically receiving and storing by the programmed computer a set of dates defining a time horizon for the computation;

for each date, electronically receiving and storing by the programmed computer a historical portfolio of holdings having investment weights in a set of investible assets;

for each date, electronically receiving and storing by the programmed computer a factor risk model comprising a set of factors, a set of factor exposures for each investible asset in the historical portfolio of holdings, factor returns for each factor, and specific returns for each investible asset in the historical portfolio of holdings as of that date;

for each date, computing a first set of factor contributions by combining the investment weights of the historical portfolios, the factor exposures, and the factor returns as of that date;

for each date, computing a first set of specific contributions by combining the investment weights of the historical portfolios and the specific returns of the assets in the historical portfolio as of that date;

computing one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting a preferred mathematical model from those computed;

computing an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model; and

electronically outputting the adjusted set of factor and specific contributions using an output device.

14. The method of claim 13 in which the time series regression model is a linear function of a set of factor contributions.

15. The method of claim 14 in which a sequence of mathematical time series regression models is constructed that identifies the most statistically significant factor contributions from the model at each iteration of the sequence.

16. The method of claim 15 in which the adjusted factor and specific contributions are used to produce a performance attribution for the historical portfolios.

17. The method of claim 16 in which an adjusted factor risk estimate is computed.

18. The method of claim 15 in which a modified factor risk model is estimated using the adjusted factor and specific contributions.

19. A computer-implemented system for computing and reporting the performance attribution of a set of portfolio holdings over time comprising:

a memory for storing data for a set of dates defining an attribution time horizon to be performed;

a processor executing software to retrieve data for a historical portfolio of holdings having investment weights in a set of investible assets at each date;

a processor executing software to retrieve data for a factor risk model comprising a set of factors, a set of factor exposures for every asset in the historical portfolio, factor returns for every factor, and asset specific returns for every asset in the historical portfolio of holdings as of that date;

computing on the processor factor contributions by combining the investment weights of the historical portfolio, the factor exposures, and the factor returns as of that date;

computing specific contributions by combining the weights of the historical portfolio and the specific returns as of that date;

computing on the processor one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting on the processor a preferred mathematical model from those computed;

computing on the processor an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model for each date;

electronically outputting the adjusted factor and specific contributions on an output device.

1. A computer-implemented method for computing and reporting the performance attribution of a set of portfolio holdings over time comprising:

electronically receiving and storing by the programmed computer a set of dates defining an attribution time horizon to be analyzed;

for each date, electronically receiving and storing by the programmed computer a historical portfolio of holdings having investment weights in a set of investible assets;

for each date, electronically receiving and storing by the programmed computer a set of factors and a set of factor exposures for each investible asset in the historical portfolio of holdings as of that date;

for each date, electronically receiving and storing or calculating and storing by the programmed computer a factor return for each factor exposure as of that date;

for each date, electronically receiving and storing or calculating and storing by the programmed computer specific returns for all investible assets in the portfolio as of that date;

for each date, computing factor contributions by combining the investment weights of the historical portfolio, the factor exposures and the factor returns as of that date;

for each date, computing specific contributions by combining the investment weights of the historical portfolio and the specific returns as of that date;

computing one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting a preferred mathematical model from those computed;

computing an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model;

computing a performance attribution for the historical portfolios of holdings based on the adjusted set of factor and specific contributions; and

electronically outputting the performance attribution results using an output device.

2. The method of claim 1 in which the time series regression model is a linear function of a set of factor contributions.

3. The method of claim 2 in which a sequence of mathematical time series regression models is constructed that removes statistically insignificant factor contributions from the model at each iteration of the sequence.

4. The method of claim 1 in which an adjusted factor risk estimate is computed.

5. The method of claim 1 in which the factor exposures, factor returns, and specific returns are derived from a factor risk model.

6. The method of claim 5 in which an adjusted factor risk model is estimated using the adjusted factor and specific returns.

20. The system of claim 19 in which the time series regression model is a linear function of a set of factor contributions.

21. The system of claim 20 in which a sequence of mathematical time series regression models is constructed that removes statistically insignificant factor contributions from the model at each iteration of the sequence.

21-1. The system of claim 19 in which an adjusted factor risk estimate is computed.

22. A computer-implemented method for computing and reporting the performance attribution of a set of portfolio holdings over time comprising:

electronically receiving and storing by the programmed computer a set of dates defining an attribution time horizon to be performed;

for each date, electronically receiving and storing by the programmed computer historical portfolios of holdings having investment weights in a set of investible assets;

for each date, electronically receiving and storing by the programmed computer a set of factor exposures for each investible asset in the historical portfolio of holdings as of that date;

for each date either electronically receiving and storing or computing factor returns for each factor, and specific returns for each assets in the portfolio as of that date;

for each date, computing a factor contribution for each factor by combining the investment weights, the factor exposures, and the factor returns of that date;

for each date, computing a specific contribution of the historical portfolio by combining the investment weights and the specific returns as of that date;

computing a correlation of the time series of factor contributions and specific returns;

selecting a pre-defined correlation magnitude limit;

if the magnitude of the correlation exceeds the pre-defined limit, then

computing one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting a preferred mathematical model from those computed;

computing an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model;

computing a performance attribution for the historical portfolio of holdings based on the adjusted set of factor and specific contributions; and

electronically outputting the performance attribution results using an output device.

23. The method of claim 22 in which the time series regression model is a linear function of a set of factor contributions.

24. The method of claim 23 in which a sequence of mathematical time series regression models is constructed that removes statistically insignificant factor contributions from the model at each iteration of the sequence.

25. The method of claim 22 in which an adjusted factor risk estimate is computed.

26. The method of claim 22 in which the factor exposures, factor returns, and specific returns are derived from a factor risk model.

27. The method of claim 26 in which an adjusted factor risk model is estimated using the adjusted factor and specific returns.

28. A computer-implemented system for computing and reporting the performance attribution of a set of portfolio holdings over time comprising:

a memory for storing data for a set of dates defining an attribution time horizon to be performed;

a processor executing software to retrieve data for a historical portfolio of holdings having investment weights in a set of investable assets at each date;

a processor executing software to retrieve data for a set of factors and a set of factor exposures for every asset in the historical portfolio of holdings as of that date;

a processor executing software that either receives and stores data or computes data for a set of factor returns for every factor as of that date;

a processor executing software that either receives and stores data or computes data for specific returns for every asset in the portfolio as of that date;

computing on the processor a set of factor contributions by combining the investment weights of the historical portfolios, the factor exposures, and the factor returns as of that date;

computing on the processor a set of specific contributions by combining the investment weights of the historical portfolios and the specific returns for each date;

computing a correlation of the time series of factor contributions and specific returns;

selecting a pre-defined correlation magnitude limit;

if the magnitude of the correlation exceeds the pre-defined limit, then

computing on the processor one or more mathematical models using time series regression that describes a relationship between a time series of specific contributions as a function of the time series of factor contributions;

selecting on the processor a preferred mathematical model from those computed;

computing on the processor an adjusted set of factor contributions and specific contributions utilizing the preferred mathematical model for each date;

computing on the processor a performance attribution for the historical portfolio of holdings based on the adjusted set of factor and specific contributions; and

electronically outputting the performance attribution results on an output device.

29. The system of claim 28 in which the time series regression model is a linear function of a set of factor contributions.

30. The system of claim 29 in which a sequence of mathematical time series regression models is constructed that removes statistically insignificant factor contributions from the model at each iteration of the sequence.

31. The system of claim 28 in which an adjusted factor risk estimate is computed.

32. The system of claim 28 in which the factor exposures, factor returns, and specific returns are derived from a factor risk model.

33. The system of claim 32 in which a modified factor risk model is estimated using the adjusted factor and specific returns.