Burstiness in Finance and Econophysics

From: scerir (scerir@libero.it)
Date: Thu Jan 18 2001 - 07:35:28 MST


C. Tannous, A. Fessant
"Combustion Models in Finance"
http://xxx.lanl.gov/abs/physics/0101042
Combustion reaction kinetics models are used for the description
of a special class of bursty Financial Time Series.
The small number of parameters they depend upon enable
financial analysts to predict the time as well as the magnitude
of the jump of the value of the portfolio.
Several Financial Time Series are analysed within this framework
and applications are given.

"Physics and the stock market: playing with fire"
http://physicsweb.org/article/news/5/1/6
Financiers have long used statistics to forecast fluctuations on the stock
market, and in recent years they have also employed models from high
energy physics and statistical physics. Now two French physicists have
used theories of combustion to predict the future values of certain shares
with surprising accuracy.
Combustion often appears to occur spontaneously after a long period of
stability. Charbel Tannous and Alain Fessant of the Université de Bretagne
Occidentale in Brest recognised that this is similar to the behaviour of
some stocks that tend to jump in value suddenly after long spells of
stability - a quality known as 'burstiness'. The pair decided to apply their
knowledge of condensed matter physics to the fluctuations of the stock
market. Their findings could help market analysts to predict when stock
values will rise and fall, and by how much.
Simple models of combustion are based on changes in the concentration
of fuel over time. Ignition takes place when the fuel concentration reaches
a certain level. Tannous and Fessant modified the equations, replacing fuel
concentration by the share prices of real companies. The researchers chose
large and small companies from different industrial and economic
backgrounds, and charted the variation in their share prices over five
years.
For all six companies, the researchers found that the predictions of the
combustion-inspired model very accurately matched the actual variation in
the share price. "We would be delighted if financial analysts would consider
deterministic approaches for predicting financial trends", Tannous told
PhysicsWeb. "Combustion theory offers a natural framework for the
description of bursty financial time series."
According to Tannous, this is the first time that combustion models have
been used to predict the performance of companies. Tannous and Fessant do
point out, however, that after the share value has jumped, market conditions
become more important and the analogy with combustion models disappears.

"It seems that to some extent the laws of physics
are chosen by the universe in response to the experiments that
are done by its inhabitants, thanks to a kind of wavefunction
collapse. When a publicly accessible causal loop is introduced
into the system, it turns out that, to avoid certain paradoxes,
the universe responds by making the laws of physics different
from the currently accepted model"
-John Baez

"You think you are joking, but there are systems that work this way.
Take a look at the stock market (but don't look too hard it'll get you
in trouble). If anyone figures out a good theory, one that can make
predictions, look what happens. People use those predictions to make
money, sometimes at first lots of money. But then a strange thing
happens, the theory has a direct effect on the system being studied.
Say the theory correctly predicts one week in advance that a given
stock will go up %10. After the theory is discovered the stock goes up
%10 but does it one week earlier"
-Ralph Hartley

See also
http://arxiv.org/abs/cond-mat?9804111
http://www.iop.org/Physics/News/0282j
http://www.iop.org/Journals/qf/extra/9
http://physicsweb.org/article/world/12/6/1/1
http://physicsweb.org/article/world/12/1/7/1
http://physicsweb.org/article/world/12/9/3/1



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:05:00 MST