Assessment of reaction substrate scope is often a qualitative endeavor that provides general indications of substrate sensitivity to a measured reaction outcome. Unfortunately, this field standard typically falls short of enabling the quantitative prediction of new substrates' performance. The disconnection between a reaction's development and the quantitative prediction of new substrates' behavior limits the applicative usefulness of many methodologies. Herein, we present a method by which substrate libraries can be systematically developed to enable quantitative modeling of reaction systems and the prediction of new reaction outcomes. Presented in the context of rhodium-catalyzed asymmetric transfer hydrogenation, these models quantify the molecular features that influence enantioselection and, in so doing, lend mechanistic insight to the modes of asymmetric induction.asymmetric catalysis | free-energy relationships | computational chemistry H uman brains are highly experienced at recognizing patterns in observed data. Organizing information and drawing connections between data enables general conclusions to be made, whether fast or slow, good or bad, or high or low. Although these qualitative assessments are routinely crafted they are subject to biases, causing evaluations to differ from one individual to another (1). The examination of a reaction's substrate scope often takes on a similarly qualitative air (2-5). A substrate scope for a developed synthetic method typically provides an indication of functional group tolerance and general trends in reaction outcomes for sterically and/or electronically varied substrates. This qualitative approach, which lacks quantitation of how substrate features will influence a reaction's outcome, particularly product selectivity, often limits a reaction's application to contexts with high degrees of similarity to the initial scope library. Additionally, it can be difficult to predict, beyond generalities such as poorly versus well-behaved, how a new substrate will perform under the reaction conditions. Addressing this limitation through quantitative prediction of reaction outcomes would significantly affect how one both develops and applies a new synthetic method while simultaneously imparting fundamental mechanistic insight (6).To accomplish this goal, an entirely new approach to examining a reaction's substrate scope is required. Because the ultimate goal is to mathematically predict a broad range of reaction outcomes, an initial library of substrates would need to be carefully designed to represent many of the impactful features influencing the reaction. Specifically, one would need to include systematic variation of steric and electronic features of a given substrate class while also limiting the initial size of the substrate library to make this a practical venture. With this in mind, the tenets of design of experiments (DoE) and regression modeling will need to be exploited, where broadly descriptive models are built from data that systematically sample the experimen...