Source: Cotton & Wilkinson, Advanced Inorganic Chemistry, III edition 1972.
Mo neutral atoms have configuration [Kr]4d5s1. By giving away three electrons we could expect both d2 and d3 configurations, [Kr]4d2s1 and [Kr]4d3s0, for the Mo(III) species. We cannot say which one is energetically stabler without considering the ligands and their geometry around the Mo atom. The 3+ oxidation state is not one of the most stable states of Mo, and the few reported complexes are octahedral. Thus the d3 configuration minimizes the repulsion of 4d orbitals' lobes with the ligands better than the d2s1 does, as specified below.
Scientific knowledge of chemistry is based on measured properties of known and real substances, not on school talk&chalk teacher's omniscient speculation, or Mo3+ "naked" and useless ions, isolated in the empty space.
So let's welcome relevant information of really existing chemicals: a yellow ion Mo(H2O)3+6 ion can be obtained, but it is unstable because sensible to oxidation, whereas chlorocomplexes of Mo(III), as MoCl3-, form red precipitates with the biggest alkali metal ions, as in K3MoCl6 which are stable red crystals in absence of humidity.
The magnetic behavior of MoCl3-6 corresponds to an electronic configuration t32g. The magnetic momentum is 3.8 BM (Bohr magnetons) meaning there are three (d) unpaired electrons.
The 2g subscript in t32g means that the three unpaired d electrons occupy the three 4d obitals which haven't lobes directed in the x, y, z directions. This in turn confirms that the six halogen atoms are octahedrically coordinated around the central metal atom.
If you want to learn more about the split of the five d orbitals under the octahedral coordination, search the Crystal Field Theory .
Cotton & Wilkinson report also the species [Mo(NCS)6]3- as being octahedral and d3.
Chemistry having to do with real, coloured substances, is quite more interesting. On the other side a kind of "school chemistry" dealing with only theoretical issues does not exist at all out of school.
