Secondary RNA structures play an important role in transposition, in particular, in RNA recognition by transposon proteins. Previously, we found a conserved structure at the 3′-end of human transposons and proposed a hypothesis about the role of this structure in transposition. Although there is no similarity at the sequence level, the conserved position of this structure points to the fact that structural properties occur that are under positive natural selection. In this paper, the physical and geometric properties of stem-loop structures at the 3′-end of human transposons are identified and compared with properties of the structures of other genome regions. Each stem-loop structure was characterized by a set of ten characteristics: the Gibbs free energy, enthalpy, entropy, hydrophilicity, Shift, Slide, Rise, Tilt, Roll, and Twist. A model has been built using machine-learning methods, which recognizes stem-loop structures according to their physical and geometric characteristics with 94% accuracy. The most important parameters in the recognition model are hydrophilicity, enthalpy, Rise, and Twist. These properties of transposon structure are thought to be under positive natural selection.