That’s right. Flash internally uses Twips for screen positions. Twips are 1/20 of a pixel, so anything between e.g. 0 and 1/20 is rounded to one or the other (I’m not sure which). 1/20 of a pixel is too small to see, which is probably the point, but if you use the positions in your own calculations it can introduce errors.
The solution is never use the positions of DisplayObject and similar classes in your own calculations. If your objects use Bitmap instead subclass Bitmap so you can add your own x and y. Or separate your game logic and rendering more cleanly, so the classes describing game logic are independent of the game graphics.
This has the other benefit that it’s much much faster to use your own x and y. The x and y of a DisplayObject and derived classes use accessors which are slow. Better to avoid them except when you actually need to place or move something onscreen.