If you are actively making a mouse in world space you should do it all in one state.
I approached it with the idea of 1. Get the mouse position on the screen, 2. Convert that to world space, 3. Raycast from the camera to that point in world space, 4. Store the point it hits, 5. Set the position of the 3d mouse object there.
That's all done in one state every frame. That was my approach.