The displayed radiation pattern is not a physical object in a meaningful spatial relation to the elements of the antenna, it's a graph of data. Consider the meaning of the (polar) coordinates of each:
- For the antenna model, $(r, \phi, \theta)$ are coordinates of the antenna structure elements.
- For the pattern graph, $(\phi, \theta)$ specify a direction away from the antenna, and $r$ is a gain value, not a length.
Remember, when you are modeling an antenna like this you are considering the far field results — the pattern is telling you the radiated power that would pass through a given point infinitely far away from the physical antenna structure.
To make spatial sense of the pattern, if you must, you can think of it like your antenna model is “shrunk to a point”, made infinitely small. And the point it's shrunk to might as well be the coordinate origin, not the center of your entered model, because that's simple and well-defined, and there's no particular guarantee that your antenna model is symmetric to have a well-defined center.