Residential real estate is a key component of household wealth. While we have some insight into the aggregate returns of this asset class---thanks to real estate price indexes---we have little understanding of the characteristics of investment in individual properties. This paper outlines and applies a methodology for estimating and examining the variation in risk and return for unique homes. We use large data sets of home prices and rents for Sydney, Australia, from 2002-14, to estimate flexible spline hedonic models which incorporate spatial and characteristics smoothing. Using these models we estimate total returns---the sum of capital gains and rental yield---for a large sample of properties for each time period. This enables use to consider the risk and return of investment in individual parcels of residential real estate and explore the benefits of diversification. As a reference point, we contrast housing with investment in equities. We find that there is dispersion in returns and their volatility---though significantly less than shares---and that this is tied to certain home characteristics. However, when we investigate the benefits of holding a diversified portfolio of homes over a single home we find they are small compared with the corresponding case for equities.