Developing materials with outstanding performance for sorption thermal energy storage (STES) is vital in utilizing renewable energy. Metal–organic frameworks (MOFs) and covalent organic frameworks (COFs) have attracted much interest for application in STES due to their excellent adsorption properties, including large capacities and stepwise adsorption isotherms. However, the energy density (Qed), an essential property to look at when choosing a suitable material for STES, is still elusive due to the different composition methods in the experiment. This work evaluated and compared the material-based Qed’s of MOFs and COFs for STES via grand canonical Monte Carlo simulations. It was demonstrated that most MOFs exhibited larger Qed than COFs since MOFs tend to have high loading during the charging process. Nevertheless, it was found that one COF exhibited the highest Qed ascribed to the low density and complete desorption during the discharging process, which suggested that COFs can possess excellent performance as long as they achieve sufficient capacity during the charging process. Moreover, the structure–property relationship indicated that large pore volume, relatively small density, suitable carbon atom ratio, and isotropic 3D cage were favorable for large-Qed structures. The successful implementation of data mining and machine learning algorithms paves the way for rational design and speeds up the assessment of the Qed of nanoporous materials.