Engineering proteins to have desired properties by mutating amino acids at specific sites is commonplace. Such engineered proteins must be stable to function. Experimental methods used to determine stability at throughputs required to scan the protein sequence space thoroughly are laborious. To this end, many machine learning based methods have been developed to predict thermodynamic stability changes upon mutation. These methods have been evaluated for symmetric consistency by testing with hypothetical reverse mutations. In this work, we propose transitive data augmentation, evaluating transitive consistency, and a new machine learning based method, first of its kind, that incorporates both symmetric and transitive properties into the architecture. Our method, called SCONES, is an interpretable neural network that estimates a residue's contributions towards protein stability dG in its local structural environment. The difference between independently predicted contributions of the reference and mutant residues in a missense mutation is reported as dG. We show that this self-consistent machine learning architecture is immune to many common biases in datasets, relies less on data than existing methods, and is robust to overfitting.