Books
-
The Flaw of Averages, by Sam
Savage, describes the pitfalls of average-case analysis and lays out the
necessity of Probability Management. Examples from the book and other related materials can be found at flawofaverages.com.
Where to buy:
Articles
Markowitz 2.0, in Morningstar Advisor, by Paul D. Kaplan and Sam Savage. (2010)
Probability Management in Action, in Analytics Magazine, by Sanjay Saigal. Download PDF. (2009)
Review of The Flaw of Averages in Chance News.
"It is a great book! There are 47 chapters in the book and on average they are gems. " Give Me a Number — Introducing the DIST, in the Big Fat Finance Blog, by Alan Radding. (2009) Probability management: The new arithmetic for risk, at Entrepreneur Corner, by Sam Savage. (2009) Toward a Consolidated Risk Statement, by Sam Savage and Aaron Brown, appears in the December issue of Risk Professional magazine. (Subscribers to the magazine may view it at their site.) It describes the potential of the DIST format to enable the aggregation of risk reports across an organization. (2009)
A two-part foundational article on
Probability Management was published in OR/MS
Today. You can download PDF
(Part
I, Part II)
or view online (Part
I, Part
II). (2006)
Press
Release
Palo Alto, California, November 3, 2008
ProbabilityManagement.org announces the DIST 1.0 Distribution
String standard
The recent turbulence in financial markets is a sobering reminder that
risk and uncertainty are always in the shadows. Yet today’s risk
management is usually based on archaic and outdated nineteenth century
statistical concepts such covariance. This approach cannot capture the
nonlinearities and optionality of the modern economy, nor does it allow
for risks to be easily consolidated across an enterprise.
Now a group of academics, software developers and managers have come
up with a new computer data type, the Distribution String, or DIST, for
short, to correct some of the glaring problems with classical risk management.
History
The concept of Probability Management is based on the work of
Dr. Sam Savage of Stanford University, Stefan
Scholtes of Cambridge University, and Daniel of Zweidler of Merck &
Co., Inc. In 2005 it was first applied on an industrial scale to consolidate
the portfolio of petroleum prospects at Shell International Exploration
and Production, when Zweidler was at Shell. In 2006 the approach was greatly
enhanced through the introduction of the Risk Solver from Frontline Systems,
headed up by Dan Fylstra, a co-founder of VisiCorp, which marketed the
first spreadsheet, VisiCalc. The basic approach prescribed by Probability
Management is now also supported by JMP from SAS Institute, Crystal Ball
of Oracle Corp, and the Vanguard System from Vanguard Software Corp.
Distribution Strings
Although the Risk Solver technology greatly improved portfolio modeling
at Shell in 2007, the approach still used vast quantities of data to model
tens of millions of possible uncertain outcomes. Then in 2008, Savage
devised a new data type: the Distribution String, which packs together
a thousand or so numeric outcomes into a single character string. This
allows a probability distribution to be represented as a single data element
in a spreadsheet or database, and furthermore greatly speeds up computation.
Open Standard
Savage believed the best way for this idea to flourish was to publish
it as open source, and in 2008, he formed an ad hoc committee
to establish a useful format standard. Members of the committee include
Nobel Laureate Harry Markowitz at the University of California in San
Diego; John Sall, the co-founder of SAS; Dan Fylstra of Frontline Systems;
Daniel Zweidler of Merck and Co. Inc.; Bryan Baker of Shell International
Exploration and Production; Eric Wainwright of Oracle Corp.; Rob Suggs,
CEO of Vanguard Software Corporation; David Cawlfield of Olin; Robert
Ameo of Market Modelers; Doug Hubbard of Hubbard Research; and Darren
Johnson of Genentech, among others.
The resulting standard benefited greatly from this collaboration, and
is now available. For more information
on the DIST standard, email us at DistInfo@ProbabilityManagement.org.
|
