Pages: [1] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Dante Fachis
|
Posted - 2011.08.10 02:39:00 -
[1]
hello all.
I gotta say, I've read tons of stuff about the market and I still just don't really get it.
last I read it, the way EVE calculates stuff for the market history tab is still a bit of a mystery? I remember something about how it represents only completed transactions, and that it excludes outliers (prices too high or too low)
I'm looking at the market history table, yesterday's entry for a given item. It says the lowest price was 20,000, the highest price was 30,000, and the average was 25,000. (I made up those numbers, but they are in the ballpark for what I'm talking about).
Well -- I know for a fact that the highest price for yesterday was *not* 30,000, it was in fact 30,500, because I have transaction entries in my wallet with yesterday's date for that value. So how come the highest price for yesterday isn't the price I got for said item?
It doesn't seem like my price is so far above the reported max to be classified as an "outlier".
If someone could point me at some (additional) reading material, I'd much appreciate it. |
Devai Starchild
Caldari
|
Posted - 2011.08.10 04:18:00 -
[2]
It's possible either it:
A - required a minimum amount to sell before it counts that in the market history
B - requires the order to be completed or calculates the median price of that order if you adjusted it.
Did you complete your order? Did you change it ever? |
Akita T
Caldari Navy Volunteer Task Force
|
Posted - 2011.08.10 04:51:00 -
[3]
Edited by: Akita T on 10/08/2011 04:52:13
We have no idea how EVE calculates what exactly is "an outlier".
All we know for sure that SOME trades are indeed considered outliers and are trimmed from the market statistics. Also, we know that at least some of that trimming happens at downtime for the previous day (i.e. 11 hours after the "day" ended), while some of it happens at least on a roughly hourly basis as the day goes by (probably even more often, but it's not quite fixed either).
There have been theories, there have been some small scale tests, but the algorithm is probably rather complex so it would be fairly difficult to discern from the limited data available to any individual anyway.
P.S. Placed orders are indeed irrelevant. Only actually processed trades (pair of matching orders) matter for the market stats//graphs. _
Akita T USEFUL EVE LINKS collection |
Dante Fachis
|
Posted - 2011.08.10 17:36:00 -
[4]
Thanks for the replies.
I might have modified the order, but before any of it moved -- and when it moved, the whole thing moved, for the price that was higher than the reported daily max.
also, I forgot to ask/confirm: the market data is just for that region, right? if it shows 10,000 units of widget X sold on day Y, that's in that specific region. Right?
I mistrust black boxes. Is the idea then to just use the market history data as a rough outline of how a given item sells? |
Claire Voyant
|
Posted - 2011.08.10 17:59:00 -
[5]
Well, it could be a really stupid algorithm to remove the outliers. Where statisticians would remove trades more than so many standard deviations from the average, it wouldn't surprise me if CCP just did something like removing the top and bottom x% of volume or x% of trades.
|
Zagam
|
Posted - 2011.08.10 18:14:00 -
[6]
Originally by: Claire Voyant Well, it could be a really stupid algorithm to remove the outliers. Where statisticians would remove trades more than so many standard deviations from the average, it wouldn't surprise me if CCP just did something like removing the top and bottom x% of volume or x% of trades.
It isn't that difficult to create a few lines of code to figure out SD and SE, and then removing those that are statistically significant. Consider the physics and mathematical calculations needed just to play EVE. SD and SE is cake in comparison. ---------.oOo.---------- Chaos, Madness, and Destruction. My work here is done. |
Claire Voyant
|
Posted - 2011.08.10 20:07:00 -
[7]
Originally by: Zagam
Originally by: Claire Voyant Well, it could be a really stupid algorithm to remove the outliers. Where statisticians would remove trades more than so many standard deviations from the average, it wouldn't surprise me if CCP just did something like removing the top and bottom x% of volume or x% of trades.
It isn't that difficult to create a few lines of code to figure out SD and SE, and then removing those that are statistically significant. Consider the physics and mathematical calculations needed just to play EVE. SD and SE is cake in comparison.
I can't stop laughing at this. If they actually used physics that might be true, but what passes for physics in eve is pretty straight forward first level linear approximation. No gravity, momentum, or acceleration required.
|
Abdiel Kavash
Caldari Paladin Order Fidelas Constans
|
Posted - 2011.08.10 20:19:00 -
[8]
Edited by: Abdiel Kavash on 10/08/2011 20:19:05 I seem to remember CCP actually saying they are intentionally screwing up the data so that players won't datamine and abuse it (and/or place orders to intentionally affect the data). ---
|
|
|
|
Pages: [1] :: one page |
First page | Previous page | Next page | Last page |