In this article, we'll try to propose a modeling of the Moroccan stock market performance. To do this, the MASI, aggregate index representing Casablanca stock exchange, will be modeled from the method "ARIMA - BOX-JENKINS ', then by a GARCH model. This note is endeavoring to first present a brief overview of the theoretical framework of both models: ARIMA and ARCH, then an analysis of serial MASI, then searching of the ARIMA model most appropriate for the MASI(identification and validation of models to reproduce the series), and finally the choice of the most suitable GARCH model based on statistical criteria. The data available are 494 daily values of the evolution of the MASI, from 21/12/2009 to 18/12/2012. Software used are SPSS 17 and Eviews 6.