The traditional approach to portfolio construction has been based around the concept that there are performance differences across regions and functionally between different types of real estate. The result of which is that portfolios are described and compared on the basis of their portfolio weighting to different regions or sectors. For a number of years now IPD in the UK, and more recently in other countries, has produced a portfolio attribution model of real estate assets based upon a functional regional segregation of the market. Whilst such a method of segregation feels intrinsically correct, and there is some evidence to support it, relatively little of the variance in returns between individual assets is explained by such methods. The principal problem with segmentation of real estate markets is that property is by its very nature heterogeneous and any one segment lumps together radically different investments. This paper aims to compare the segmentation used by IPD with a totally different approach, which does not explicitly differentiate between location or sector specific factors. Instead we intend to test whether a portfolio could be more efficiently constructed on the basis of a number of different factors such as vacancy rates, lease lengths, development activity and lot size amongst others.