Résumé en anglais

With hedgefunds, managers develop risk management models that mainly aim to play on the effect of de correlation.In order to achieve this goal,companies use the correlation coefficient as an indicator for measuring dependencies existing between(i)the various hedge funds strategies and share index returns and(ii)hedge funds strategies against each other.Otherwise, copulas are a statistic tool to model the dependence in a realistic and less restrictive way,taking better account of the stylized facts in finance.This paper is a practical implementation of the copulas theory to model dependence between differen the hedgefund strategies and share index returns and between these strategies in relation to each other on a "normal" period and a period during which the market trend is downward. Our approach based on copulas allows us to determine the bivariate VaR level curves and to study extremal dependence between hedgefunds strategies and share index returns through the use of some tail dependence measures which can be made into useful portfolio management tools.