Knowledge about the appropriate indicators to point out the best components in a catalytic process is a basic prerequisite for obtaining insights into optimized reactions as, for example, in the chemical vapour deposition method, which enables the growth of carbon nanotubes. In this work, we report a density functional theory study of 13-atom transition-metal nanoclusters interacting with (5,0) zigzag and (3,3) armchair carbon nanotube fragments, considering all transition-metal species from the periodic table as possible candidates for the chemical vapour deposition method. The icosahedral configuration was found to be a good model to simulate the seed of nucleation in the case of the short carbon nanotube fragments that are initially formed during the growth process. From full geometric optimizations, without any constraints, we found that the energetic and structural nanocluster properties change as a function of the occupation of the bonding and anti-bonding d-states. The center of gravity of the occupied d-states for nanoclusters is found to be a good indicator to reveal the best candidates for the interaction with the carbon nanotubes, namely, Sc-Cu, Y-Nb, Pd, Lu, Hf, and Pt. The interaction between all transition-metal nanoclusters with both armchair and zigzag segments is favorable in terms of the adhesion energy, where the adhesion is larger for systems with smaller occupation of the d-states. The bond strength is more pronounced for systems with zigzag fragments than those with armchair fragments, which is confirmed by the smaller average bond length between the metal atoms of the nanocluster and the C atoms of the zigzag segment. Our prediction about the best 13-atom transition-metal candidates is reinforced by the linear relationship between the adhesion energy and the center of gravity of the occupied d-states. Thus, the adhesion energy presents increased intensity for the interaction between carbon nanotube fragments and nanoclusters in relation to the smaller occupation of the d-states. Consequently, our model is able to provide a good descriptor for indicating the best 13-atom transition-metal candidates in the chemical vapour deposition process.