|
PureBytes Links
Trading Reference Links
|
Concerning discussion of the article "A Multifractal walk down Wall Street" by Benoit
Mandelbrot in the February 1999 issue of Scientific America:
To save space here "M" is Mandelbrot.
Underlying the article seems to be that a key aspect of portfolio management is the
assessment of risk, and that Mandelbrot's (M's) models based on using "fractal
generators" more accurately reflect the risk of drastic changes in financial prices than
conventional risk models. M says that variations in financial prices can be accounted
for by his work with fractals, which new theory refers to as multifractals. Although I
don't know the theory of fractals what M does is generate models of daily changes in
financial prices using "fractal generators." I'm not sure what's so new about it, since
the generation and analysis of fractal objects has been around a long time (a typical
example is a fictitious asteroid or planet in a science fiction movie.) Perhaps it's the
use of the new multifractals.
M says that he can use his fractal generator models to "stress-test a portfolio" much
better than conventional portfolio theory. He doesn't say it but it seems if for example
Long Term Capital Management were using M's fractal generators, they could have
potentially better estimated the risk of their positions going drastically wrong and as a
result could have either reduced their risk exposure or used derivatives to hedge their
risk. As M says "modern portfolio theory poses a risk to those who believe in it too
strongly."
M has an interesting sub-topic called "Pick the Fake" with several time series of changes
in daily prices of different instruments. Some time series are generated by fractal
generators and some are real changes in daily prices. It's impossible to tell which time
series is real and which isn't.
Although M's emphasis is not on predicting the market, he seems to say his fractal
generators can help give probabilities of various future moves. For example his fractal
generators might give the probability that high and low for the coming week will be
drastically different from this week.
M also has a new book referenced in the article (Benoit B. Mandelbrot):
Multi-Fractals and 1/F Noise: Wild self-affinity in physics, Springer-Verlag, 1999
--- Chris
-----Original Message-----
From: Gaius Marius <magnus@xxxxxxxxxxx>
To: Omega List <omega-list@xxxxxxxxxx>
Date: Thursday, February 04, 1999 12:58 PM
Subject: Mandelbrot's article in Scientific American
>Anyone care to discuss Benoit Mandelbrot's article in Scientific American,
>"A Multifractal Walk down Wall St." ?
|