Étude du Modèle d'Évaluation par Arbitrage sur le marché des actions suisses

In this paper the application of Arbitrage Pricing Theory (APT) and multifactorial pricing is studied on the Swiss stock market. In order to estimate the factors used in the multifactorial model, it is proposed to use the new method of Independent Component Analysis. This method laying on neural networks allows including information on high-order statistics and dependences between distributions of stock returns. Its use on a twenty years period of swiss stock returns beginning January 1, 1975 allows significative improvement in the multifactorial model tests when comparing with the commonly used Principal Component Analysis.
[ - ]
[ + ]