The applying of radical polymerization techniques is an important economic and theoretical task of chemical technology. Hence, the elaboration of the radical polymerization processes has both theoretical and practical significance. The ability of the CORAL software to build up robust models for transfer constants in radical polymerization is demonstrated. The Monte Carlo technique is applied in the CORAL software as the principle of development of the models. For molecular fragments represented by simplified molecular input-line entry system (SMILES), correlation weights were selected that maximize the correlation coefficient between the logarithm of the chain transfer constants in radical polymerization log (1/Cs) and the descriptor calculated from the correlation weights for the training set (n = 25). This correlation was preserved for the external validation set in three optimization samples by the Monte Carlo method.