There is a growing body of research that seeks to understand the mechanisms that drive and maintain neurodevelopmental differences, without recourse to conventional diagnostic categories. So called ‘transdiagnostic’ studies change multiple design features, relative to a more conventional case-control design. Different sampling frames, approaches to assessment and analytic frameworks are allowing researchers to explore neurocognitive mechanisms across hitherto accepted diagnostic boundaries. Sensitive measurement, bridging levels of analysis, and capturing developmental change all pose significant challenges to transdiagnostic researchers. However, methods borrowed from elsewhere, allowing researchers to model individual-level data rather than the group average, and a new focus on developing circuits rather than singular brain regions, are bringing fresh mechanistic insights. We argue that a transdiagnostic approach needs to do more than just improve the characterisation of neurodivergent young people; in future it must also incorporate the developmental dynamics that capture how and when neurodevelopmental differences emerge. This crucial step will require methodological innovation. We introduce approaches that are just emerging within the field, which may allow us to do this, including Generative Network Modelling, Impulse Response Analysis, Network Control Theory and Normative Modelling.