M. Senneret, G. Perrin, L. Jaffrès, Y. Malevergne, P. Abry
Study presented by our team in the framework of the first “Agora de la Gestion Financière”, created at the initiative of the AFG, which was held in Paris on February 10, 2016.
The theoretical and empirical results of this study allow us to conclude that it is necessary to use robust correlation estimation methods to avoid an undesirable increase in portfolio risk and to significantly improve its Sharpe ratio, for example in the context of variance minimization strategies.
We have implemented different estimation methods in the context of euro equity portfolios (244 stocks of the EuroStoxx 600) over a period of 15 years. The results are analyzed in terms of ex-post volatilities, Sharpe ratio, turnover rate and dispersion indicator (equivalent number of stocks). The differences between the direct estimation of the covariance matrix and its inverse (precision matrix) according to the classical Markowitz method, or the direct estimation of the inverse matrix, are also considered. Finally the simulated portfolios are either “long-short” or “long-only”.