Lecturer:  Thomas Guhr, Faculty of Physics, University of Duisburg-Essen
Abstract:
The basic models of stock markets assume purely Markovian
dynamics in the stock markets, i.e. lack of memory. This
cannot be completely true, one of the reasons being that
some traders use strategies. We study a fully correlated
stock market and quantitatively extract the deviations
from purely Markovian behavior in intraday trading.

Non-stationarity is one of the most important features
of stock markets or complex systems in general. We show
how to identify and determine, on long time scales,
quasistationary states of the correlated stock markets.
The time evolution of the system is along trajectories
in the space spanned by these quasistationary states.
This is, on long time scales, another non-Markovian
effect.