Applying Omega

摘要

We begin with a very simple requirement: the need to compare and choose among a range of funds, moving forward to an optimal allocation among them. The data, from UK equity income funds and their benchmark, are daily returns: some 1,300 data-points over the period 1999-2004, including three erroneous mispricings. The funds and the benchmark are strongly correlated, with typical values around 0.92. The time series are highly and statistically significantly negatively auto-correlated. There is strong evidence of clustering of volatility and co-dependence. The datasets all fail standard normality tests and exhibit asymmetry of returns. None is time homogenous - if subsamples of a series are compared, it is possible to reject the hypothesis that these subsamples are drawn from the same distribution. They are sufficiently similar that they may come from a common distribution.