Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe
AKSOE: Physik sozio-ökonomischer Systeme
AKSOE 1: Finanzm
ärkte und Risikomanagement I
AKSOE 1.3: Vortrag
Montag, 24. März 2003, 10:30–11:00, BAR/205
Financial Markets: Memory Effekts and Forecasting — •Michael Schulz — Abteilung Theoretische Physik, Uni Ulm, 89069 Ulm, Germany
Financial data such as asset prices show remarkable memory effects. A well known phenomena is the volatility correlation function with a characteristic relaxation time of an order of magnitude 102 trading days. But also the price autocorrelation shows at short time scales a significant memory. It will be demonstrated that this behavior is a consequence of the complex dynamics at the financial markets which can be described by a nonlinear Fokker-Planck equation allowing the determination of the conditional probability distribution function and some related critical exponents as well as the scaling behavior observed in financial data.
Furthermore, it will be demonstrated that the memory effects can be used for the prediction of the evolution of financial time series. Especially the combination of Bayesian statistics and neural networks allows a real forecasting of the volatility and the trend over 1−102 trading days.