We present the preliminary stages of an agent-based model of the housing market in the Washington, DC metropolitan area (1.8M households & 1.2M houses) during the recent housing bubble and collapse. We are attempting to create an empirically grounded model using data from various sources, where the some data sources are used as inputs to the model and others as target outputs for the model. Once the model has been properly calibrated, we will use the model to adjudicate the various theories cited as the cause of the U.S. housing bubble. This talk will provide a brief review of the data collected so far, a description of the model, current model results, and preliminary counterfactual experiments.