I am currently working on strong-coupling linked-cluster expansions for various t-J models as well as the Kondo lattice model. The linked-cluster expansion method is based on Rayleigh-Schrödinger perturbation theory. It's home turf is the computation of ground-state properties. Recent developments, however, have extended the method to computing effective Hamiltonians of degenerate excited states, which in turn yield experimentally accessible quantities, such as excitation spectra or spectral weights.
One particularly appealing feature of the linked-cluster method is its numerical exactness in the sense that all quantities are automatically obtained in the thermodynamic limit. This makes it a complementary tool to the more widely known DMRG or exact-diagonalization methods.
Furthermore, unlike quantum Monte Carlo techniques, it does not suffer from any fermion sign problems (the student implementing it may suffer, but the method prevails).