I have flown several models both in the real world and on FMS. I'll give one example. I built and flew the SPAD Debonair for many, many flights before it bit the dust. I also have flown the FMS version. My model was more lively than FMS in both acceleration and roll rate. I physically upped the thrust factor to compensate as well as reduce the roll inertia.
I believe the modeling in FMS solves the six degrees of freedom and does it well. These equations need the physical (weight and moments of inertia) and a complete description of the aerodynamics and propulsion system. The tough job is to figure out what they are and then input them into FMS. Lift and drag can be estimated, but where do you get the roll dampening or the aero cross coupling? All the flight simulation programs have the same problem ... good data input gives realistic flight characteristics, otherwise garbage in/garbage out.