# From scaling to forecasting in finance

ARGOMENTI: Convegni

SEMINARI DI CALCOLO DELLE PROBABILITA

Il Prof. Attilio Stella, del Dipartimento di Fisica dell'Università degli Studi di Padova, terràil 29 novembre alle ore 11:00 nell'Aula 2A/45 della Torre Archimede, un seminario dal titolo "FROM SCALING TO FORECASTING IN FINANCE".

- Abstract
An empirical property of financial assets like stocks or market indexes is the non-Gaussian scaling with time of the single return probability density function due to long-range temporal correlations. On time scales ranging, e.g., from days to months, we show that this scaling can be well approximated by a simple form, involving a single anomalous exponent. We also demonstrate that this scaling form and the symmetries due to the efficiency of the market allow one to predict conditional probabilities for successive returns in contiguous time intervals. These probabilities are remarkably consistent with financial data. We incorporate strong correlations by means of a novel product operator applied to the Fourier transforms of single return probability density functions. The same product allows a generalization of the central limit theorem to sums of many strongly correlated random variables obeying anomalous scaling.

Rif. P. Dai Pra