This paper examines the dynamics of owner-occupied housing prices both at the level of the individual dwelling and in aggregate. Using a unique data set, a model of individual dwelling prices is estimated that represents features of housing markets more faithfully than competing models. Statistical tests strongly reject the hypothesis that individual housing prices follow a random walk in favor of the alternative hypoth- esis that housing prices are mean reverting. This result also holds in aggregate, and provides an explanation for the \inertia" reported in housing return series. The paper then demonstrates that real and excess returns are forecastable. Finally, it considers empirically the extent to which the transactions costs associated with home ownership preclude protable speculation in owner-occupied housing markets.