This paper contributes to the debate on the liquidity of real estate investment in the context of financialisation. Using microdata built from tax registers, we analyse the geography of rental housing purchases by private individuals from three French cities. We develop a modelling approach in order to better understand the respective roles of space and market characteristics in determining buy-to-let investment flows. Considering the distribution of the data and our objective of integrating both intra-and intercity housing investments in a single model, we use an adaptive zoning approach. This approach allows high spatial resolution where interactions are strong to be kept, and the aggregation of more distant, less populated areas. We demonstrate that geographical proximity is highly determinant in explaining flows of buy-to-let investments from private individuals. We also uncover striking facts related to the geography of rental investments, such as the convergence of investments from rich suburbs toward the centre of agglomerations and preferential flows from the Paris region to southern and coastal cities. Finally, we find that investors tend to buy in upmarket areas and in places that are more expensive than their market of residence. Our results indicate that geographical proximity and safety of investments are key factors in housing wealth accumulation by private individuals.