In this paper we study optimal portfolio allocation of real estate assets using IPD transaction-based data. Currently, appraisal-based indices represent a standard tool for the analysis of property returns in portfolio decisions. However, appraisal-based indices present a number of shortcomings: they may understate volatility, lag turning points and be affected by client influence. In contrast, transaction-based indices should be a more authentic measure of market returns. Therefore we tackle the natural question of exploring portfolio choices on the basis of such market-based indexes. In particular, we perform a number of quantitative exercises: (1) We investigate the value of a de-smoothing parameter that could re-align appraisal-based to transaction-based indices (2) We compare optimal portfolio allocations based on both indices and the relative performance (3) We investigate whether portfolio allocations based on transaction based indices can improve the efficiency of the real estate market and the implications for market returns