Only if you talk about the simulated features of the simulated world, rather than compare the "simulated world as seen by its inhabitants" with the simulating world.

We don't have dragons on earth, but I can simulate dragons.

In the sense that this simulation exist in our world, you are right that the simulating world will then always be "richer" because it contains the simulation.

But if I could enter the simulated world, I could ride dragons. I can't ride dragons in "our" world, so in that sense it is clear that we can simulate things that do not have a concrete existence in our own world, and I to me at least that would make the simulated world "richer" in that respect by making things possible in the simulation that requires you to be in the simulation for it to be possible.

Similarly, we can clearly simulate something with more detail - e.g. we could simulate a world where our elementary particles can be subdivided endlessly, if we choose to. In the simulating world this would "just" be a simulation, but in the simulated world it would be that worlds reality.

There is even no reason why, with sufficient resources and time dilation, it would not be possible for the simulating world to simulate a world equivalent to the simulating world, so it could well be turtles all the way down.

I believe it is not correct. You can put new features that do not exist in the physical world in a simulation. For example, you can double the number of quarks in a proton as long as you define a mathematically consistent interaction to allow so.

Then you created a richer simulated world.

This is not true for infinite sets where a subset can be equivalent to the whole.

Could you possibly explain/reason why this must be, without using "by definition"? Many people in this thread agree with you on this, but I don't understand it (see my other comment using a video game analogy).

Is the richness you describe in your comment implicitly constrained to that which exists physically perhaps?