Spin-orbital coupling predicts whether a material class will exhibit unconventional superconductivity with low error rate.

A data-based predictor for unconventional superconductivity is constructed, which has three ingredients: a layered crystal structure, stable local magnetic moments and an intermediate spin-orbital (not spin-orbit) coupling. This classifier is sufficiently specific to clearly distinguish cuprate and iron-based superconductors from most other materials. Multiple DFT calculations allow for higher computational accuracy. Our results suggest that Sr2FeO4, Sr2CoO4, Sr2VO4, BaCo2As2 and K2CoF4 may have the necessary physics to exhibit unconventional superconductivity or other spin-orbital effects.