[p2p-research] How the Derivatives Market Jumped of a Cliff On Its Way to a Party

[marc fawzi]

How the Derivatives Market Jumped of a Cliff On Its Way to a Party

This is based on something I read recently but cannot locate atm....

Someone had come up with a probabilistic formula to estimate the probability
of risk associated with a given random pool of derivative securities. It
worked 99% of the time in predicting the probability of risk by making the
flawed assumption that a high degree of correlation between independent
random variables (across different types of derivatives) was actually
expressing some hidden causality (some magic in the way nature worked that
no one had yet understood) when in fact all it was expressing was a very
high degree of correlation between completely independent random variables.


When statisticians play physicist they use the term "copula" which is the
bundling of independent random variables with high correlation into a
constant. This "copula" method assumes that there is some hidden causality
that no one really understands but that can be used to calculate with high
certainty the probability of risk associated with a random pool of
derivative securities each with it's own set of independent random
variables, without analyzing each derivative security individually, and
without having to put only derivatives of the same type in the same pool. As
if statistics had found some hidden causality in nature, which happens all
the time, e.g. Gaussian distribution is common in nature but not universal
yet it is assumed to be universal by some statisticians as if it's a
physical law. In other words, high degree of correlation is often confused
with causation or is taken to be hidden causation in order to simplify some
probabilistic calculation (at the expense of misleading the user.)

So bankers were able to bundle all sorts of derivatives into the same pools
and rate them together using the said magic formula, which allowed a massive
increase in the volume of trading in securities (went from double digit
billions to double digit trillions in just over 10 years.)

When the market conditions changed this copula (or assumed hidden causality)
went out of the window and the magic formula's ability to predict the
probability of risk associated with pools of derivatives went out of the
window with it. So what we ended up with were credit securities pools that
were rated AAA but had dog shit in them.

~~

That satisfies my personal curiosity as to what happened.

It could have been avoided if the person responsible for the now-infamous
derivatives valuation formula did not treat causality (or lack of) in such a
shallow and stupid way.

Some kids have an imaginary friend. Some statisticians have an imaginary
universe.

Marc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listcultures.org/pipermail/p2presearch_listcultures.org/attachments/20090227/4ea9a9a7/attachment.html>