The Green's function of the wave equation is a spherically expanding wave. In physical terms, this means that for an infinitesimally small disturbance (think for example of a small droplet falling into the water), you will get a spherically expanding wave.
When you throw an arbitrarily large stone in the water, you can think of it as a bunch of infinitesimally small stones added together. The resulting disturbance will then be the sum of all the spherically symmetric disturbances that are caused by all of the points on the stone's surface.
Of course, you are right in remarking that it is strange that the overall wave pattern still looks rather spherical. If we add a lot of spherically expanding waves together from the surface of the rock, wouldn't we expect the resulting wave to be as rigged as our original stone? There are two reasons why this isn't the case
First, the expanding spherical waves quickly form a smoother expanding spherical wave as they interfere with each other. The second reason (I think) is because of the adhesiveness of the water. When the stone falls in the pond the water sticks to its surface, causing the stone to look smoother as it is covered by a layer of water. This causes the resulting wave pattern to look smoother as well.
You can experiment yourself and you will probably see a clear distinction between spherical and elliptical stones.