The prime housing market is an area of research in which there has been limited attention given in recent academic studies. In this paper we try to identify drivers behind the formation of prime residential property submarkets. We argue that as well as relying on prior knowledge and assumptions of property agents and market participants, the submarkets can and, to an extent, should be derived empirically. As an example we compare our empirically defined submarkets with boundaries defined by experts and analyse differences. We recognise the differing motivations between defining a submarket through prior knowledge versus empirical means and try and evaluate the benefits of combining both methods. London is one of the largest prime residential property markets with a significant amount of foreign ownership, and thus it was chosen as a pilot for our prime submarket research. Drivers for prime residential real estate differ from the ones for the rest of the residential markets. We compile the appropriate dataset for London markets by using Knight Frank’s own transaction data supplemented with information from Land Registry in addition to relevant socio-economic and economic data sources. Spatial and locational effects are accounted for both in the methodology and variables within the models. In this study we utilise a combination of hedonic modelling, principal component analysis and cluster analysis. We will build upon previous studies into defining housing submarkets by assessing the suitability of incorporating local spatial econometric principals into the methodology.