I've been having this trouble for a while and haven't figured it out. I run simple simulations using foundation bolts and compare them to the same simulation where the bolts are replaced by a "fixed" restraint. When I compare the resultant force results of the fixed restraint to the connector force result of the bolted version, the shear load matches well but the axial load never does. It seems that the fixed restraints method matches the hand calculations much better and the bolt method ends up with almost twice the axial load. I attached an example.