||After the failure of traditional financial models in 2007/08, the agent-based approaches in Finance have attracted attention of both academicians as well as practitioners. Recently, number of projects propose the courageous attempt to complement or even alternate current mainstream policy making approaches e.g. on the level of central banks. Although empirical estimation is an inevitable part of the modelling cycle, not many examples exist on structural estimation of financial agent-based models. Moreover, looking ten years back in literature, we neither see any general consensus on the methodology, not any conclusive results. This situation is caused by two crucial complications. First, the highly nonlinear nature and complexity of these systems prohibits researchers of using classical estimation methods. Second, the high number of degrees of freedom, parameters, and possible model settings require huge computational capacity for the analysis. This project aims to propose a general methodological framework for empirical estimation of financial agent-based models based on the new method of simulated maximum likelihood (Kristensen and Shin, 2012). We will extensively test this estimation methodology on particular financial agent-based models, but we presuppose the methodology will appear even more general and useful e.g. for macroeconomic agent-based models in the future.