Multi-affine analysis of typical currency exchange rates

摘要

For foreign currency exchange rates, multi-affine analysis can put quantitatively into evidence the differences between correlated (daily closing market) values and random walks in time dependent data. The H(q) spectrum is presented and discussed here for the USD/DEM and JPY/USD exchange rates. The time-evolution of these ratios is found to be multi-affine. The h(γ)-curve describing the hierarchy of exponents is numerically obtained. Our findings suggest that the modelling of exchange rate time-evolution from day to day is possible within the framework of modern statistical physics and related to models of turbulence in the physics of fluids. Finally, we argue that there is a multiplicity of information levels in the foreign exchange market such that the "efficient market theory" is a crude oversimplification indeed.