dkh said:
I use estibot all the time for exactly what it's supposed to do: Compile a bunch of very useful statistics in a single go, then give a ballpark estimation based on many "difficult to peg" variables.
Thanks for this. The above sentence sums it up better than I could...
Carlton said:
Josh - I was experimenting with estibot and got a strange result. Curious to see if you could shed light on this. NewYorkProperty.com (separated as in: New York Property.com) yielded appraisal of $3600. NewYorkProperty.net however was "reg fee". Thanks for input.
That's a good question but very hard for me to answer. The short answer is that it's due to an imperfect algorithm. The long answer is that I'd have to try to explain everything...
The medium-length answer is that there is no set % of the dotcom value assigned to the other tld's. Well in fact there is, but it's a baseline and it changes depending on the domain itself. After setting the baseline value, the domain name is analyzed "to death" as much as is it possible for my little piece of code (of about 2,000 lines) to do so. The valuation is done differently for different tld's and different domains. Also there are dozens of error reduction routines that may smell a rat and change the value completely, and even the error checking routines may make ...errors....well, you get my meaning. It's complicated. The whole algorithm is "fuzzy", i.e. it's not x * y + z = value. My first version was like that but it just didn't cut it. Now the value changes numerous times as the algorithm proceeds and finds new properties in the domain.
Some domains are immensely stronger as .com compared to other tlds. Other domains are more suitable for other extensions as well. You may see percentages of .net = 20% x .com, and on the other hand, as here, .net = 0.3% x .com.
The reason this is so is because it's the solution that gives the best overall results, when I analyze large numbers of results statistically. Of course, it means that for some domains, you get an underestimation of non-.com tld value, and for others an overestimation.
It's pretty easy to program a very rough appraisal guide for pure keyword domains, but it's difficult to make a universal algo for keyword domains as well as for brandables, LLL's, CCC's, and LLLL's and so on. I have attempted to do the latter in order to avoid having to make differential algorithms for each domain subtype. EstiBot already has differential "sub-algorithms" for different domain types, but to make a whole new algo for each domain type, well it would be more accurate, but it's a near-impossible task for one man. I mean, think CCC, LLL, LLLL, L-L-L, LL, NL, NLL, N-N-N, GEO, keyword, real estate, business, law, adult, brandables, plus different rules for each type in each TLD (see 3-character price guide and that's just for starters).
I can't even think of starting to do that, especially now that I have my hands full keeping the damn thing running because a lot of API services are blocking my calls due to the increased usage, the API bills are growing....
As for NewYorkProperty.com (nice domain by the way), I think the .com seems to be an underestimation and there are a couple of reasons that come to mind. Lack of sales data from domains with similar characteristics, also less favorable keyword profile than for instance, LondonProperty.com, which is appraised at $36,000.
EstiBot algorithm is heavily based on the NameBio database and also my own database of past known domain sales. The very special way that the database is utilized is the "secret heart" of the program.
Looking at similar domains in the past sales database - there are not that many CityProperty.tld sales in the database. NYProperty.com was sold for $5,600 and MinneapolisProperty.com for $440 (although that was 3 years ago). Given this data the estimate of $3,600 seems reasonable, but I think it's underestimated.
I'm sure that didn't answer your question, but I hope it was of some interest
Cheers-
Esa
