When preview resolution != native resolution, the camera rescales to show more or less information - this is configured using the overrides. If you're using fit visible, it will compress the native resolution image to display inside the preview resolution. I.e. in your image its displaying more data horizontally, which inevitably means that there is data beyond x < 0. This allows you to not have to reposition everything in your game depending on resolution, but simply design for one resolution and display more or less on others.
If you want to get exact positions on screen, you can either use tk2dCamera.ScreenExtents to get the extents after scaling, or use camera.ScreenToWorldPoint to get a world position for a point on screen.