Thank you for your calculation of the almost infinite number of names possible in any TLD
@4pm but I must respectfully disagree with your opinion that the number of domains in a TLD that are for sale should be ignored.
As a domain investor I want to know what the probability of success should I make some choice. Now while all sorts of individual factors govern that, I think it is logical to ask 'When others invested in similar domain names, what was their success?' For example if other domain investors tried investing in 100,000 domain names in a certain niche or TLD and sold 5 that means something rather different than if only 50 tried to sell in that niche and 5 sold.
I follow the monthly real estate board report in my city. They report the number of sales in different categories and they compare that to the number of active listings in that category. By your argument they should not do that, they should count every property in the region, as in essence almost any of them could be for sale if someone came along and offered an outrageous price. But they do it in a way that is meaningful to someone selling real estate. If I am average, for every x listed properties for sale, about y are selling in a month. I can adjust that for whether I think I am above, or below, average and use it as a measure to compare how I am doing as a real estate agent.
I would argue that we should be doing exactly the same. Until a few months ago I used the total registered as the number in denominator, while recognizing a number of these were not for sale. Now however, with Dofo advanced search, we can actually see the number for sale across all of the major domain marketplaces plus many (not all) of the registrar marketplaces. Does it miss names for sale or report some that were left up for sale and are sold? Absolutely. But NameBio also misses many aftermarket sales because they are not reported. By comparing the sales to a measure of those actively for sale we get a reasonably probability, if I am average, of what might be expected.
I might say that there is at least one situation where I would not do it this way. That is where the majority of sales reported on NameBio are registry premium sales. Then your argument that a larger number of possible names, held by registry, should be in the denominator (I guess if you knew for a TLD how many were reserved as premium you could still calculate). For example, in .global most reported sales are registry. However, in .app, the registry do not directly report premium sales and a quick glance suggests few, if any, of the 25 were not aftermarket sales.
I want to stress that the statistical argument, while useful guidance I would argue, does not tell you what is the right decision for any particular name. Some great names will always sell. Many poor names will never sell.
@bmugford made an excellent point that most of the recorded sales in .app are in short names which must be taken into account.
The importance of looking not only at the number sold, but the number for sale, explains how a number of our colleagues on NPs have success in various niches. In total there are far fewer sales in .io or .co or .org than in .com, but there are also far fewer people trying to sell names in those. One should not, in my opinion, ignore the number of domain names that are actively for sale. Fortunately, Dofo now allow us to do it easily!
Bob