Abstract
The empirical performance of cointegration and copula asset allocation techniques are compared against that of the market. Multivariate copula structures are used to derive index-tracking portfolios which are then compared with that of portfolios constructed using cointegration techniques. The results suggest that modelling the long-term relationships between stocks by means of the cointegration approach do not consistently lead to portfolios that outperform the benchmark. Using a short-term asset allocation approach, such as the copula-simulation approach, lead to portfolios that perform at least as well as the cointegration portfolios.