Data processing: financial – business practice – management – or co – Automated electrical financial or business practice or... – Finance
Reexamination Certificate
1998-09-11
2001-09-18
Trammell, James P. (Department: 2161)
Data processing: financial, business practice, management, or co
Automated electrical financial or business practice or...
Finance
Reexamination Certificate
active
06292787
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to the field of financial advisory services. More particularly, the invention relates to a portfolio optimization process that diversifies model risk by favoring a more diversified portfolio over other portfolios with similar characteristics.
BACKGROUND OF THE INVENTION
From a set of N financial products (N>1), an infinite number of portfolios are available for investment. Existing computer financial analysis systems (also referred to as “portfolio optimizers”) purport to help individuals select portfolios to meet their needs. These systems typically implement mathematical models based upon standard optimization techniques involving mean-variance optimization theory. According to the mean-variance approach to portfolio selection, an optimal portfolio of financial products may be identified with reference to an investor's preference for various combinations of risk and return and the set of efficient portfolios (also referred to as the efficient set or the efficient frontier).
FIG. 1
illustrates a feasible set of portfolios that represents all the portfolios that maybe formed from a particular set of financial products. The arc AC represents an efficient set of portfolios that each provide the highest expected return for a given level of risk. A portfolio's risk is typically measured by the standard deviation of returns. In general, there are many portfolios that have almost the same expected return and about the same level of risk as any efficient portfolio (e.g., portfolio B and portfolio E). Since statistical estimates of expected returns and risk are used to calculate efficient portfolios, the calculated efficient set could deviate from the true efficient set. When “model risk” is considered, portfolios in an error space surrounding an optimal portfolio are virtually indistinguishable. By “model risk,” what is meant is the uncertainty/risk in the mathematical models employed and errors that may be introduced when estimating the properties of the financial products based upon historical data which may contain inaccuracies, such as statistical noise or measurement error, for example. An example of a problem induced by measurement error is the potential for highly concentrated estimated efficient portfolios. For instance, consider an asset that has a large positive error in its expected return estimate. Efficient portfolios constructed ignoring the possibility of this large positive error may yield portfolios with highly concentrated positions in this asset.
Existing portfolio optimizers typically ignore model risk, likely because of the great amount of processing that is thought to be required to identify and select from the many indistinguishable portfolios. Prior art portfolio optimizers are notorious for recommending portfolios that have counterintuitive properties, such as highly concentrated positions in individual assets or asset classes. For example, the typical portfolio optimizer, having ignored portfolio E because it is not in the efficient set, would suggest portfolio B which may include highly concentrated holdings in one of the underlying N assets. Such recommendations make users skeptical of the results of traditional portfolio optimizers and discourage adoption of such investment tools.
One way investment managers have traditionally attempted to compensate for the inadequacies of portfolio optimizers is by imposing constraints or bounds on the optimizer in one or more dimensions. For example, an investment manager may limit exposures to certain asset classes, limit short positions, etc. While these manual constraints can be implemented with knowledge of the bounded universe from which the portfolio will ultimately be built, they have several limitations. First, these manual techniques do not take the cost of imposing constraints on the optimization process into account. Additionally, manual solutions are typically only practical when the universe from which the portfolio can be drawn is limited to one set of mutual funds, asset classes, or financial products.
In view of the foregoing, what is needed is a generalized portfolio diversification approach that produces recommended portfolios that take into account inherent model risk and with which users will be intuitively comfortable, thereby fostering the adoption of optimization tools. Additionally, rather than arbitrarily spreading assets out, it is desirable for the decision to pursue more diversity in a portfolio to consider the cost of such diversity, in terms of its effect on expected return, risk, and/or utility, for example. Finally, it would be advantageous for the diversification approach to be broadly applicable to the universe of financial products.
SUMMARY OF THE INVENTION
A portfolio optimization process that diversifies model risk by favoring a more diversified portfolio over other portfolios with similar characteristics is described. Broadly stated the present invention involves determining an initial portfolio, performing diversification processing to identify one or more alternative portfolios having increased diversification, and selecting a recommended portfolio from the initial portfolio or the one or more alternative portfolios based upon a set of one or more criteria.
According to one aspect of the present invention, a more diverse portfolio may be selected over an initial portfolio in order to diversify model risk with reference to a predetermined diversity budget. An initial portfolio of financial products is determined from an available set of financial products. One or more dimensions of an error space are searched for an alternate portfolio that is more diverse than the initial portfolio. A cost associated with the alternate portfolio is then calculated by comparing the difference between a characteristic of the initial portfolio and a corresponding characteristic of the alternate portfolio. Finally, the alternate portfolio is selected as the recommended portfolio if the cost is less than or equal to the predetermined diversity budget.
According to another aspect of the present invention an intelligent search is performed for a diverse portfolio that meets a predetermined diversity budget. An initial portfolio is determined based upon an available set of financial products. The cost associated with more diversified portfolios compared to the initial portfolio is considered and one of the more diversified portfolios is selected that has an associated cost that is less than or equal to the predetermined diversity budget.
Other features of the present invention will be apparent from the accompanying drawings and from the detailed description which follows.
REFERENCES:
patent: 3634669 (1972-01-01), Soumas et al.
patent: 3697693 (1972-10-01), Deschenes et al.
patent: 4007355 (1977-02-01), Moreno
patent: 4334270 (1982-06-01), Towers
patent: 4346442 (1982-08-01), Musmanno
patent: 4376978 (1983-03-01), Musmanno
patent: 4597046 (1986-06-01), Musmanno et al.
patent: 4642767 (1987-02-01), Lerner
patent: 4648037 (1987-03-01), Valentino
patent: 4722055 (1988-01-01), Roberts
patent: 4742457 (1988-05-01), Leon et al.
patent: 4752877 (1988-06-01), Roberts et al.
patent: 4774663 (1988-09-01), Musmanno et al.
patent: 4868376 (1989-09-01), Lessin et al.
patent: 4876648 (1989-10-01), Lloyd
patent: 4885685 (1989-12-01), Wolfberg et al.
patent: 4910676 (1990-03-01), Alldredge
patent: 4933842 (1990-06-01), Durbin et al.
patent: 4953085 (1990-08-01), Atkins
patent: 4989141 (1991-01-01), Lyons et al.
patent: 5025138 (1991-06-01), Cuervo
patent: 5101353 (1992-03-01), Lupien et al.
patent: 5126936 (1992-06-01), Champion et al.
patent: 5132899 (1992-07-01), Fox
patent: 5148365 (1992-09-01), Dembo
patent: 5220500 (1993-06-01), Baird et al.
patent: 5222019 (1993-06-01), Yoshino et al.
patent: 5227967 (1993-07-01), Bailey
patent: 5237500 (1993-08-01), Perg et al.
patent: 5454104 (1995-09-01), Steidlmayer et al.
patent: 5471575 (1995-11-01), Giansante
patent: 5523942 (1996-06-01), Tyler et al.
patent: 5563783 (1996-10-01), Stolfo et al.
patent:
Jones Christopher L.
Scott Jason S.
Shearer James W.
Watson John G.
Blakely , Sokoloff, Taylor & Zafman LLP
Dixon Thomas A.
Financial Engines, Inc.
Trammell James P.
LandOfFree
Enhancing utility and diversifying model risk in a portfolio... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Enhancing utility and diversifying model risk in a portfolio..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Enhancing utility and diversifying model risk in a portfolio... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2445968