current UK domain appreciation

[julian]

Hi,

i'm doing a bit of research and i'm interested to hear your thoughts on what you think the current rate of domain appreciation is for UK domains annually.

This is very difficult to quantify as some domains have revenue (type-ins)others don't - yet their worth is still generally increasing over time (at the moment :mrgreen

I'm looking for a baseline appreciation figure. Obviously exceptional domains (loans,co,uk etc) will probably go up more than good to average domains (whats a good average domain...?!) Of course most people see value in a domin others don't so they may feel its tripled in value each year while the next man thinks its only gone up 10%...

So if possile I want you to take a higher level view. If you only bought 1 domain for £6 last year and sold it for £12 last week you could say 50% but this sort of figues imo arnt truly reflective of todays Uk domain market.

anyway your thoughs are welcome.

 

[Edwin]

Here's the problem: the appreciation could be 1000% per year and it would still make regfee domains regfee domains. There has to be a certain minimum quality there in the first place. And as you pointed out, the best domains will appreciate faster.

I'd guess 5-10% for "ok" names (0% for names that are worth nothing - a thousand years from now they'll still be worthless) and 20-30%+ for good and excellent domains. That's for the sustained increase across a large number of names...

 

[anthony]

Julian,

I guess you're simply trying to get a rough idea, right? Not an easy thing to assess, but (MASSIVE 'BORING SH!T' WARNING) you could use what is known as the Standard Deviation (SD) figure in a Gaussian distribution curve. Really! You can't really apply this to every domain registered (you don't know them all), but you can apply it to the domains you know have sold, which after all are the samples which influence the sales pattern (stick with me Julian!)

Here's how it works: Get together as many sales as you can find for say the past 6 years. Break the sales figures down into say £100 steps, and plot them in a distribution. Calculate the Mean & SD from the distribution (not easy if you don't know about these things). +&- 3SD is always 99.7% of the sample group, which means that if, year on year, the SD relative to the Mean gets smaller, then the relative values of the domain names sold is getting higher! Piece of p1ss!

The rest of you can wake up now!

 

[julian]

thanks for your input guys - can you please vote though!!

 

[rob]

It is hard to vote on this one.

This is based on own historical data and increased values. One issue with this is published prices need to be put in context, ie. was the purchase a bargain at the time?

Likewise what type of domain?

Three letters, for example, used to be manually registerable / posted as tat on forums. Accepted they are duff ones, but in 2004 they were reg fee and today they would be £100-200 if not more.

Good three letters changed hands for steadily increasing figures, roughly 25% a year increase however as they are usually £xxx - low xxxx it is not as big in real terms.

CDs.co.uk DVDs.co.uk will only drop in value in the coming years, whereas new tech names would rise.

Lots of similar factors, and as they are domain names not gold it is hard to give a market wide verdict.

I would vote 0% to dissuade any would be domainers reading this to plough their life savings into domains that they know little about!

 

[julian]

come on boys.

put your money where your mouth is!

I hope people arn't frightened to commit to voting?

Don't forget i'm only looking for an rough appreciation figure within our RESELLER market

 

[JMOT]

just need to find a bank willing to lend on the value of a domain portfolio and we could be laughing ;-)

 

[Beasty]

You may find Sedo's figures for 2007 helpful when trying to judge this question.

http://www.sedo.com/press/domainmarketstudy2007.pdf

 

[retired_member26]

Julian,

I guess you're simply trying to get a rough idea, right? Not an easy thing to assess, but (MASSIVE 'BORING SH!T' WARNING) you could use what is known as the Standard Deviation (SD) figure in a Gaussian distribution curve. Really! You can't really apply this to every domain registered (you don't know them all), but you can apply it to the domains you know have sold, which after all are the samples which influence the sales pattern (stick with me Julian!)

Here's how it works: Get together as many sales as you can find for say the past 6 years. Break the sales figures down into say £100 steps, and plot them in a distribution. Calculate the Mean & SD from the distribution (not easy if you don't know about these things). +&- 3SD is always 99.7% of the sample group, which means that if, year on year, the SD relative to the Mean gets smaller, then the relative values of the domain names sold is getting higher! Piece of p1ss!

The rest of you can wake up now!

How do you know the parameter "domain sale price" has a perfect normal distribution? Probability, Stat A, Stat B and stochastic processes are no way boring stuff at all. They are the core of Economics and Engineering. You should calculate Mean, Median, SD and Variance. Probably you will get a skewed normal distribution towards higher end (meaning the big sales end). Besides you should look at the sales data and look for outliers. Outliers mean here crap names which should have sold for 5 bucks but had sold for GBP1000. My proximation is that you will get a mean of GBP2500 and a SD of GBP500.

Actually,If someone has S-plus he can do this stuff in a few hours incase data is ready.

TurNIC

 

[anthony]

How do you know the parameter "domain sale price" has a perfect normal distribution?...
TurNIC

You don't!

But as you know TurNIC, i suggested normalising the distribution for each year simply to conduct a year on year comparison. Even using a Gaussian distribution analysis on a skewed distribution, the corrolation between the two is usually there.
And i would hesitate to remove the 'outliers' as well, they still represent part of the capture data.
I agree that stats are interesting, and very useful indeed. It's such a shame the British Army rely upon the Gaussian to such an extent that they seem to order 10,000 pairs of boots religiously to the results. The outcome: Half a dozen squaddies at the back of that queue end up with badly fitting boots because then never order an extra six pair for each size!

 

[retired_member26]

You don't!

But as you know TurNIC, i suggested normalising the distribution for each year simply to conduct a year on year comparison. Even using a Gaussian distribution analysis on a skewed distribution, the corrolation between the two is usually there.
And i would hesitate to remove the 'outliers' as well, they still represent part of the capture data.
I agree that stats are interesting, and very useful indeed. It's such a shame the British Army rely upon the Gaussian to such an extent that they seem to order 10,000 pairs of boots religiously to the results. The outcome: Half a dozen squaddies at the back of that queue end up with badly fitting boots because then never order an extra six pair for each size!

I would do this for boots:

I try to get the each years' histograms. Try to see what kind of distribution they have. If the histograms give a shape of Standard distribution I will go with it or try to use other distributions like poisson etc.

I, then get the distribution of means or medians of each years data and check whether means or medians of each years' means or medians have a tendency. By taking 95percent confidence interval of the distribution of the means I will make sure I have some margin.

Then I go back to the last years standard distribution and expand each foot group by that number coming from the distribution of the means.

Or you can use elastic material and produce just 10000 mean number shoes

TurNIC

 

[retired_member12]

I would do this for boots:

I try to get the each years' histograms. Try to see what kind of distribution they have. If the histograms give a shape of Standard distribution I will go with it or try to use other distributions like poisson etc.

I, then get the distribution of means or medians of each years data and check whether means or medians of each years' means or medians have a tendency. By taking 95percent confidence interval of the distribution of the means I will make sure I have some margin.

Then I go back to the last years standard distribution and expand each foot group by that number coming from the distribution of the means...

TurNIC

Exactly!!

 

[julian]

any changes in thought given the latest sales

thx

 

[Rich]

I would do this for boots:

I try to get the each years' histograms. Try to see what kind of distribution they have. If the histograms give a shape of Standard distribution I will go with it or try to use other distributions like poisson etc.

I, then get the distribution of means or medians of each years data and check whether means or medians of each years' means or medians have a tendency. By taking 95percent confidence interval of the distribution of the means I will make sure I have some margin.

Then I go back to the last years standard distribution and expand each foot group by that number coming from the distribution of the means.

Or you can use elastic material and produce just 10000 mean number shoes

TurNIC

If i ever come to an Acorn meet remind me not to sit between you two

 

[julian]

Would it be a fair assumption that a ball park figure for general annual domain appeciation would be around 15%.

I could stick my neck out and say thats still a far better return you'd get from a banking dosh etc..?

(of course its still entirely subjective as that figure (15%) is based on the knowledge of the people who voted 50/50 between 10% and 20%?)