© 2016, © Emerald Group Publishing Limited. Purpose: This paper aims to investigate how a prospective buyer’s optimal home-size purchase can be determined by means of a stochastic-dominance (SD) analysis of the historical data of Hong Kong. Design/methodology/approach: By means of SD analysis, the paper uses monthly property yields in Hong Kong over a 15-year period to illustrate how buyers of different risk preference may optimize their home-size purchase. Findings: Regardless of whether the buyer eschews risk, embraces risk or is indifferent to it, in any adjacent pairing of five well-defined housing classes, the smaller class provides the optimal purchase. In addition, risk-averters focusing on total yield would prefer to invest in the smallest and second-smallest classes than in the largest class. Research limitations/implications: As the smaller class provides the optimal purchase, the smallest class affords the buyer the optimal purchase over all classes in this important housing market – at least where rental yields are of primary concern. Practical implications: The findings suggest that in the Hong Kong housing market, long-term investors may be better off purchasing smaller homes. For other type of investors, it depends on their risk preference. Originality/value: There is a very small body of empirical literature on housing investment, especially if the focus is on the optimal home-size purchase.