Conventional real estate portfolio construction seeks to diversify specific risks through allocating assets across property types and geographical locations. Empirical observations have suggested that assets of different qualities and securities of income have experienced rather different recovery paths after the 2007 credit crisis, bringing forward potential benefits by adding two more dimensions to the portfolio diversification process. Ample research has been devoted to segmenting the UK real estate market, thereby revealing the most efficient groupings for risk reduction, and hence also for optimal portfolio construction. This paper builds on past work, focusing on three market segments defined both according to their types and geographical locations, which essentially control for the two traditional dimensions. From this premise, this paper explores the diversification benefit in allocating asset across qualities and securities of income within each segment. Correlation analysis and efficient frontier construction will be used as primary tools to investigate the risk reduction potential. Acknowledging the non-normality in the distribution of real estate returns, this study employees the Mean-Absolute Deviation (MAD) model to span the efficient frontiers in each segment, based on these two dimensions of asset characteristics. Results from this work shall bridge the top-down and bottom- up portfolio construction strategies, by recognising the merit of considering asset specific features in the macro allocation stage.