In recent years big financial institutions are interested in creating and maintaining property valuation models. The main objective is to use reliable historical data in order to be able to forecast the price of a new property in a comprehensible manner and provide some indication for the uncertainty around this forecast. In this paper we develop an automatic valuation model for property valuation using a large database of historical prices from Greece. The Greek property market is an inefficient, non- homogeneous market, still at its infancy governed by lack of information. As a result modelling the Greek real estate market is a very challenging problem. The available data cover a big range of properties across time and include the financial crisis period in Greece which led to tremendous changes in the dynamics of the real estate market. We formulate and compare linear and non-linear models based on regression, hedonic equations and artificial neural networks. The forecasting ability of each method is evaluated out-of-sample. Special care is given on measuring the success of the forecasts but also to identify the property characteristics that lead to large forecasting errors. Finally, by examining the strengths and the performance of each method we apply a combined forecasting rule to improve performance. Our results indicate that the proposed methodology constitutes an accurate tool for property valuation in non- homogeneous, newly developed markets.