If you modeled the bolts as part of the base, then they were centered in the base hole. Allowing them to move in the holes would give you some more deflection. If the clearances were the same in the base and the cap, then I would expect the deflection of the cap to double, so you are still off by a factor of 2.5. How much of the base did you model? It could be that you put your fixed surfaces too close to the hinge area and you aren't capturing enough of the deflection in the base. If the bolts are cantilevered instead of being supported on both ends, then they would probably rotate significantly and cause a lot more deflection of the cap.
What are the stresses where the bolts contact the mating surfaces? You may be getting localized yielding in either of the pieces.
Refining the mesh may make a difference. The classic way to do an analysis is to start with a crude model and then refine the mesh until your answers converge to acceptable accuracy. If you just did a coarse mesh and never checked a finer model, then you need to take that next step.