Risk Professional - December 2009

Crystal Ball
( http://www.oracle.com/appserver/business-intelli-
gence/crystalball/ index.html)
Crystal Ball has been one of the leading Monte Carlo simulation pack-
ages for Excel for decades. It was recently purchased by Oracle Corp.,
where it has the potential to become integrated into numerous Oracle
applications. Members of the Crystal Ball team contributed signifi-
cantly to the development of the DIST 1.0 standard, and were pio-
neers in promoting libraries of probability distributions for increased
standardization.

Risk Solver ( www.Solver.com)

Introduced in 2006 by Frontline Systems, Risk Solver was the first commercially available interactive simulation package for Excel. That is, instead of requiring a user to tell it to run a simulation, it runs one automatically and nearly instantly as your finger leaves the “Enter” key. In 2008, it became the first package to support the DIST 1.0 data type. Frontline Systems also provides a suite of more powerful optimization engines that are now tied to Risk Solver, enabling stochastic optimization. This optimization may be driven stochastic scenarios generated at run time, or by previously generated DISTs.

XLSim 3.0 ( www.Probili Tech.com)

XLSim 3.0 grew out of Monte Carlo software developed by Sam Savage for teaching. It is currently the only package besides Risk Solver that supports interactive simulation with DISTs. Although not for industrial applications, it is an inexpensive way to learn about the DIST format and to develop small models. It can also translate DISTs for use in other Windows software packages.

Visualization
Data Desk ( www.datadesk.com)
Data Desk was one of the first programs inspired by Tukey’s explor-
atory data analysis. It was developed in 1985 by one of his former stu-
dents — a statistics professor at Cornell. It was released for Macintosh
in 1986 and Windows in 1997.

JMP ( www.JMP.com)

JMP is an interactive statistical exploration program from SAS Institute. Although it is a standalone desktop application, it also integrates tightly to SAS industrial databases. JMP provides a wide variety of graphing capabilities, but, unlike most visualization packages, it also contains many powerful analytical tools. It can run interactive simulation on applications built within its own modeling environment.

Spotfire ( www.spotfire.com)

Spotfire is a visualization package aimed at the business intelligence market. It links to numerous data sources and provides excellent interactive graphs. It has been used in industries as diverse as energy, pharmaceuticals and finance.

Figure 5: A Scatter Plot Matrix of the SPX, CLV9
and TYZ9 DISTs

The two sets of price assumptions in the sample model are selected with buttons, which load the appropriate DIST library. This way the statistical results under each set of assumptions can be toggled back and forth for comparison.

There are four business units in our sample file, each with a simple pricing model — simple because we gave each desk only one or two positions. It is important to note that although these pricing models are all in the same worksheet in our example, they could just as easily have been separate models ( because of the coherence of the input DISTs), running on different computers in different programming environments.

In a normal spreadsheet, the output cells would contain numbers, perhaps the average P&L for each desk under each assumption. This leads to the “Flaw of Averages,” a set of systematic errors that occur when single numbers (usually averages) are plugged into models as proxies for uncertain assumptions (see www.FlawOfAverages.com).