The use of quantitative methods within realist methodologies are fairly rare. This is perhaps because a realist understanding of the social word as complex and dynamic (messy but not chaotic) does not sit well with traditional variable-based causal analysis which test specific theoretical assumptions, yet cannot account for interaction, moderation and emergence. In this paper, we explore the ontological challenges and epistemological issues which underpin the development of our complex realist approach to quantitative data analysis. We provide an example of its application to a large case-based time-ordered dataset and the resultant discovery of the deep patterns that underlie what happened to similar and different mentally disordered offenders as a consequence of the implementation of a new policy of custody diversion.