The study of small experimental communities has been key to the development of ecology as a discipline. Yet, for most ecological communities, the number of experiments required to build, model, or analyze the system vastly exceeds what is feasible, rendering these communities experimentally intractable. To address this challenge, we present a statistical approach that uses the results of a limited number of experiments to attempt to infer the outcomes (coexistence and species abundances) of all other possible species assemblages. Using three well-studied experimental systems, we show that this method predicts with high accuracy the results of unobserved experiments. These findings suggest a scalable method for building and exploring large experimental systems, facilitating efforts to study the ecology of diverse natural communities.