La quasi marche aléatoire

The intention of this article is to take advantage of recent discoveries in econometrics (ARCH processes) to reformulate classical tests about the efficiency of financial market. It is also intended to improve the quality of stock index forecasts. Classical tests about the efficiency are realized from random walk. The change occurs here, considering the innovation of the random walk not like a white noise but like an ARCH process whose characteristics are such that they respect conditions given by Granger and Morgenstern (1970) to test the market's efficiency. This model, the Quasi Random Walk, allows a less restrictive approach to the notion of efficiency. The other advantage of the Quasi Random Walk is that it will help to refine the forecast, owing to confidence interval of forecast which aren't constant anymore. An empirical study based on five financial indexes (Germany, the United States, France, Great-Britain and Japan) shows a bigger tolerance of the Quasi Random Walk in tests of financial market's efficiency and a better approach to forecast.