My question is specifically referencing the file-based grid transformation methods that are now available in ArcGIS. I thought this grid-file method was exactly the same method that NGS's NCAT used when moving from NAD 83 to NAD 83 HARN, however I am not getting the same outputs. They are very close and seem to always been in the correct direction, but ArcMap's outputs fall short when compared to my outputs from NCAT. Either way, these state-specific methods (at least in my state) appear to be far more accurate than any others that are listed in the transformation drop-down menu.
Since this doesn't require concatenating transformations at all, why would Arc's outputs be different than NCAT's?