Interactive simulation packages can now process thousands of scenarios be-
fore the user’s finger has left the ‘Enter’ key. They provide the same intuitive
feel for probabilistic models that the spreadsheet did for numerical models.
other upcoming technologies, gives companies the ability to
capture, share and manage the uncertainties that are affecting
their businesses. At Oracle, we are actively working to support
the new DIST standard in future software releases.”
In addition to capturing the interrelationships between uncertainties, this approach can essentially model any type of
distribution, including historical. “The DIST is a significant
advance in the representation of probability distributions,”
says Paul Kaplan, vice president of quantitative research at
Morningstar Inc. “At Morningstar and Ibbotson, we have already begun to explore using it to enhance our capabilities in
Monte Carlo simulation and portfolio optimization when return distributions are highly non-normal.”
Because the strings are stored in a representation called
extended markup language (XML7), they fit well into existing
corporate information infrastructure. In fact, nothing would
prevent nesting them within higher level data structures such
as the aforementioned EconSML, which is also XML based.
Interactive Simulation
A new generation of “interactive” simulation packages can
now process thousands of scenarios before the user’s finger has
left the “Enter” key. They provide the same “intuitive feel” for
probabilistic models that the original spreadsheet did for numerical models. The most notable of these new programs is the
Risk Solver from Frontline Systems, which provides interactive
simulation in Excel, and was the first commercially available
software to support the DIST 1.0 format. (See “Tech Options”
on pg. 39 for more information on software resources.)
A Sample Model
We have provided a small sample of an interactive DIST-based model, shown in Figure 3, and downloadable at www.
ProbabilityManagement.org. You may wish to open the file in
Excel before continuing. Be sure to enable macros in Excel,
and look at the instruction page.
Consider a risk manager charged with monitoring and
hedging firmwide market risk for a large, proprietary trading
organization. Traditional approaches to this problem involve
massive IT infrastructure and long-term projects with price
tags in the billions of dollars. Large batch runs must be coordinated by complex middleware, preventing timely access to integrated data (you can get it fast, or you can get it right, or you
can get it aligned with other data; but only one of the three).
Users are often frustrated with the rigid, opaque results, the
difficulty of correcting errors and introducing new products
and, perhaps worst of all, ill-defined overlapping responsibilities that lead to finger-pointing and bad feelings.
Recently, the technology has been introduced to do the entire project in spreadsheets, with manageable modular jobs, inexpensive, fast, flexible development and user-friendly output.
Figure 3: The Interactive DIST-Based
Sample Model
Visualization
In 1977, John W. Tukey published his influential Exploratory
Data Analysis8, a book on interpreting statistical data. This
and other work eventually led to a category of statistical visualization software that allows non-statisticians to explore
data, draw inferences, and communicate results to other
non-statisticians. Now more powerful than ever, these packages originally designed for statistical data are perfect tools
for visualizing the distributions and relationships within scenario libraries.
