to 125trueptMonique PONTIERtruept to 125trueptU.M.R. CNRS C 5583 to 125trueptL.S.P. to 125trueptUniversité Paul Sabatier to 125truept118 route de Narbonne to 125truept31 062 TOULOUSE cedex 04 to 125trueptFRANCE to 125trueptpontier@cict.fr
Summary Students's works about modelling and fitting are exposed. A first try to fit the Black-Scholes model is done with estimations of trend and volatility, test of normality, estimations of Hölder exponent in case of a driving fractionnal Brownian motion. Otherwise, the stationnarity of the return logarithm is studied and non-parametrix estimation of the autocovariance function is tried. Finally, in the Diffusion-Jumps mixed model, an algorithm is done to detect jumps in a financial series and tests are done for the mixing of two (or more) Gaussian laws hypothesis. Data are these of ``Bourse de Paris'' (Lyonnaise des eaux, AGF, Peugeot) of the year 1997 and French stocks index (1802-1993).