It actually should be possible to evaluate the relative performance of two teams without assigning them both the same project. Take two projects, ask both teams

It actually should be possible to evaluate the relative performance of two teams without assigning them both the same project. Take two projects, ask both teams to estimate the time required for both projects. Assign eah project to the team which thought it would be relatively easy, and rate each team against its own estimate, and the other team's estimate. You probably have to do this with more than one project to be confident, but the general idea should be sound.