There is not yet an agreement on notions of what complexity means. Each of the definitions captures some aspects, but all have both theoretical and practical limitations and have so far rarely been useful for characterizing any real‐world system. In this paper, I develop, first, a novel conceptual framework, providing a characterization, an appreciation of complexity, which offers several advantages. It allows a meaningful comparison between different formulations of complexity and aids the identification of deeper insights into its possible causes. Second, I exemplify the usefulness of the complexity concept I propose by showing that modern financial systems can be characterized as such. Third, in that case, banks, as a part of the financial system, are confronted with dynamic complexity, which necessitates structure‐oriented (rather than data‐driven) explanatory models for risk management practices. However, this call stands in marked contrast to commonly used risk models that are not sufficiently explanatory. Copyright © 2016 John Wiley & Sons, Ltd.