"What are the odds" depends to a large degree on "what sort of stars are you looking at". If it's completely unfiltered, or just filtered to scoopables OBAFGKM, then your odds of finding any of those valuable planets drops quite significantly.
@Factabulous 's method of assuming you've found an ELW, then what are the odds of the other things appearing, is a good one.
But here too, you can increase your odds of finding good stuff by being selective about your target stars. Hotter stars have wider Goldilocks zones, meaning it's more likely a randomly generated planet will land in the zone and have a chance of being turned into a valuable planet.
I've surveyed over 5000 systems for my ED7K Project so far, equally divided among the seven most common star types (BAFGKML) and across four sectors in different parts of the galaxy. Looking at just the 2400 A, F and G stars, I have found 44 Earth-likes and 62 Ammonia worlds. On this basis, the pure-random-chance probability of one of those ELW systems also having an ammonia world would be 1 in 38, so statistically, I ought to have found a couple by now.
Here's the only one I have actually found, and it also happens to have a terraformable waterworld in it too (which, as Facta noted, is far more probable than sharing with an Ammonia World):
Thraikoo LQ-X d1-1. The star is F class, and all the worlds in question orbit that one solo star. Planet 4 is the ELW, planet 3 is the terraformable WW, and planet 8 is the Ammonia World. Planet 2 is a terraformable HMC. Here's a screenshot of the system I took earlier: