The approach I took when I ran into the same problem years ago was to experimentally determine the required correction factor (e.g., 0.714) to apply to the movement numbers. I found that I was able to get reliable distances by making the mouse moves in a series of small movements (e.g., 10).
To determine the correction factor, I used a sketch that makes a fixed mouse movement and then running it in combination with a program that shows the mouse pointer coordinates.