In February 2019 a workshop was convened at the Banbury Centre at Cold Spring Harbor, NY. The purpose of the meeting was to identify key developments required in the practice and infrastructure of computational psychiatry research to accelerate its ability to address real world clinical problems in the near future. This report provides a summary of the conclusions of the meeting. At its core are suggestions to improve the measurement properties of computational assays through a rapid, iterative process that leverages coordinated waves of online and clinical testing, followed by deployment of the assays in innovative study designs to address clinically relevant questions. We particularly focus on theory-driven tasks but, where possible, the potential of data-driven approaches is also highlighted. Finally, the report suggests that for the promise of computational psychiatry to be realized, the research environment must be developed to encourage large-scale, collaborative, interdisciplinary consortia.