These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Market Discussions

 
  • Topic is locked indefinitely.
 

The first semi-successful result of my historical market analyzer.
Author
Previous Topic Next Topic
Sciencegeek deathdealer
Echelon Research
Goonswarm Federation
#1 - 2011-09-12 21:18:13 UTC  |  Edited by: Sciencegeek deathdealer
http://www.geeksserver.com/tritgraphfromgeek.jpg

Be warned, it is massive.

It isn't really that impressive, all I did was graph tritanium in the 4 regions, and an average. BUT! I've laid the basic framework for my program, which will allow you market buffs to comprehensively analyze historical data provided by the recent data dump. I have a lot of functionality planned, but am always looking for suggestions. I'm hoping it will be web based, but I need to see how my server holds up (its an overpowered private server on a commercial connection).

ANYWAY! I was just excited and figured I'd share. The file is so big because I wanted to preserve accuracy, and well, it is a year of data. :D I didn't use any smoothing algorithms, nor time based averages. (IE 5 day average)

Edit:
Smaller pic:
http://www.geeksserver.com/tritpic2.png

Still looking for more requests... (also fishing to see if anyone will be interested in this before I spend a lot of time to make it nice and pretty...)
Sciencegeek deathdealer
Echelon Research
Goonswarm Federation
#2 - 2011-09-13 00:02:06 UTC
Wow, not even a troll. This must be even more stupid/useless than I thought. Welp, back into my cave then.
BornToDieAnotherDay
Tarazed Technology
#3 - 2011-09-13 01:05:12 UTC
It's cool, but I don't really see what can be gained from this that we can't get by looking at the ingame history in a few differing regions?
Sciencegeek deathdealer
Echelon Research
Goonswarm Federation
#4 - 2011-09-13 01:08:54 UTC
Its just a bit easier, and it has built in analytics. But that's exactly the type of thing I'm wondering, is it needed...
Cyniac
Federal Navy Academy
Gallente Federation
#5 - 2011-09-13 11:18:42 UTC
I think you need to look closer on how you calculate the 15 day moving average.

It's not normal that the 15 day moving average is higher than all of your data points in some parts of the graph - especially where Trit is overally raising in value.


I suspect that you took the average of that day and the next 14 days as your moving average (which would explain why in a rising market the moving average seems to be above market values) when you need to take the week before and the week after.


It's fun just have to figure what to do with it.
Postitute
Center for Advanced Studies
Gallente Federation
#6 - 2011-09-13 18:20:58 UTC
Hey Sciencegeek, thanks for publishign your efforts. I do agree with you that having combined graphs like this does make things a lot easier to read. I've also done some analysis on the data and will have some to add (perhaps in a different thread as I don't want to hijack your work) and I'll have some comments and requests for you later tonight when I get a chance.

Thanks
TornSoul
BIG
#7 - 2011-09-13 20:01:21 UTC
Why so large...

a year is 365 days... so 365 pixels + padding for axes etc would give enough accuracy.
It can't get any more accurate...


Current size is simply impractical - 'cause if one zooms out - the lines get all wonky :-)

You might go 365*3 - Just to make it look nicer - or even 5*365... But no more..

Just my 0.02 ISK
Postitute
Center for Advanced Studies
Gallente Federation
#8 - 2011-09-14 03:27:53 UTC  |  Edited by: Postitute
So I think this graph supports the idea that (at least in Heimatar and Domain if I'm remembering region IDs correctly) Jita/The Forge provide a more expensive source of low-end minerals consistently. On the flip side, I'm pretty sure that high-ends, imported into Jita from 0.0, are cheaper on average than what can be sourced outside of Jita.

You asked for some suggestions/requests for research - here's an interesting area perhaps:

Something I've always been curious about is the timeframe and extent to which an advanced product, such as a Hulk, is affected by movements in moon mineral price if all other inputs are roughly constant or are somehow controled. Thinking about it, perhaps a raw moon mineral (such as technitium), may or may not provide the best independent variable as it may be reacted up a few times before it gets to the market in a form that reflects the current supply/price most accurately - perhaps someone else could weigh in there, or you could just make a guess after looking at the graphs.

Trying to give a little more specific example, how long does it take the 5-day average price of a Hulk to increase to a price level that reflects a significant, and sustained increase in an underlying building material (such as technitium), relative to the proportion of that material input in the Hulk.