Without looking at the particular chart you are using, I would say GBZ is right on the money. Your discrepancy will be accountable to a number of sources, but the projection's angular distortion at the points you are measuring is likely to be the biggest contributor (other significant factors could be, depending on your particular example, your measuring a rhumb line track and comparing against a great circle track, measurement artifacts such as chart or ruler deformation, etc.)
In any case, as you correctly surmise, it is a mostly academic issue. At least for me, I couldn't hold a track to 5 degrees to save my life
