NPC shop prices

[Guest]

How is NPC demand determined?

Does demand change continuously as players add or remove items from their shops? This appears not to be the case, since demand conditions seem to persist until the next update.

[Guest]

How is NPC demand determined?

Does demand change continuously as players add or remove items from their shops? This appears not to be the case, since demand conditions seem to persist until the next update.

Demand is based on the number of inhabitants in a country * a item ratio.

For example France has 8835963 inhabitants and the demand for the following items are

Diamond rings 8.836 so 1 out of 1000
Leather jackets 883.596 so 1 out of 10
Wardrobes 4.417.982 1 out of 2

Now every daily update (about 2am server time) the number of items sold in each store is added up and added to supply* and then we get our % of demand fulfilled which lasts until the next update.

*Its actually more complex than that but if you sell the same amount of goods every day then supply will supply, and if everyone were to stop selling for one day supply will only fall by half.

[Guest]

Thanks, but by "demand", I don't mean the number listed on the Supply and Demand screen, but rather the NPC demand curve, the mathematical formula that determines how many items the NPC will buy and at what price, given the player offerings.

[Guest]

You have hit the nail on the head so to speak
Its been a goal of mine to find the 'price to time' formula but so far I have only figgured out the basic formula for the curve and have yet to find a way to tie that into price and Q changes. The problem is that I cant get my hands on enough data to confirm the effect that different S/D and AVG stats have on said curve. In short I need 3 at least recordings on what occurs whan only 1 stat is changed but all others are kept constant for each stat. If we can get those then its fairly easy to work out the full formula.

My current equ for jewerly is: (Note LN/e refers to natural logarithm adn its base)
(e^(4.95608603043 + 0.319992306101 * LN[x])/60/60)
Were x = price

The formula works to within 10 minutes varaition but the problem is it needs to be updated every 24h because as you can see it is a simple snapshot of prices-time for whatever day I took it. If you have any data from your own stores please send it to me
I hope it helps and for the full account of our current attempt to find the forumla go here
H

[Guest]

jesus christ what a complicated formula ... ever heard of simplification?

E^(a+LN[x])/(b/b), where a and b are constants, can be written as

1/b^2 * (E^a * x), where E^a is again a constant and can be written explicitely. (note that E to the power of natural logarythm of x is exactly x)

so what you wrote would be 142x/3600.

which i doubt is very correct. maybe you wrote your equation wrong. next time you try fitting data try some simple model.

[Guest]

You can

[Guest]

OK this is way more complex than i originally thought

For example

In Oman I put diamond rings for sale at exactly the stats price (Q8 7.56142c) and the selling time was 11:22:08

In Saudi Arabia I put Q5 diamond rings for sale for 7.831,40 the stats price is (Q4.45 7.831,40c) and the selling time was 13:55:25.

So clearly the selling time increases with the average stats price (or a background price level)

[Guest]

You forget that the S/D has a big impact so an item may have a selling price of 50k but if demand is 80% filled then it will take far longer to sell I cant say for sure but if you dident check them this I think thats why its changing the time

[Guest]

You forget that the S/D has a big impact so an item may have a selling price of 50k but if demand is 80% filled then it will take far longer to sell I cant say for sure but if you dident check them this I think thats why its changing the time

Another reason I have stores in yellow, constant supply

Oman got 27 supplied 3759 demanded 0.007%

SA got 250 supplied and 10759 demanded 2.3%

So although that will explain a small part of the difference (if we were talking about 80% and 82% then I would put it down to supply) something else (or somethings else) is causing this difference.

[Guest]

Thats still a huge jump from almost 0.01 to 2.3% is 2300% rise but im still a bit stumped, anyone?

[Guest]

E^(a+LN[x])/(b/b)

now as you probably should know, the exponents adding is same as multiplying the powers

so if you have E^(a+LN[x]) thats effectively the same as (E^a) * (E^LN[x])

furthermore, E^LN[x] is x

so E^(a+LN[x]) is equal to x*E^a

as i said, you probably wrote your equation wrong

AND yes, i can simply fit polynome of third order very accurately to your data, just send it to me in the form of {{price1,time1},{price2,time2},...,{priceN,timeN}}

price1,price2,...priceN are the test prices and time1,time2,....,timeN is the time needed to sell full rack with corresponding price

dont use any units please

[Guest]

Sorry I took so long to reply its been full on!

Sure thing just send me your email addy on PM and ill send along my excel sheet, I forgot which tab I took this off though. Is colour coded ect ect so shouldnt be too hard to work out what is what =)

I coppied the formula correctly but it seems that you however havent...
Its e^(a+c*LN[x])
You havent moved the 'c' over (0.32 in this case)
Should change a few things around

I have gone over your new numbers in detail and if you put in the c it should work fine.
=(e^a)*(e^ln[x^c])
=(x^c)*(e^a) (still messy but slightly better)

I still have no idea were you pulled that 142x from but hey

Apart from that good find! I stand to some extent corrected (took alot to say that ) Ill keep looking for a smaller forumla
H

[Guest]

Its e^(a+c*LN[x])

H

ok i forgot about that constant anyway, in exponent, if you have x^(y*z) thats the same as (x^y)^z or (x^z)^y. so if you have E^(c*LN[x]) thats (E^LN[x])^c which is the same as x^c.

one question, kid - which country are you from?

[Guest]

Im from good old down under (Australia)
Yourself?
H