These forums have been archived and are now read-only.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

EVE economic simulation using Input-Output/Gravity analysis.
Author
Previous Topic Next Topic
Ewo Deveraux
Care Bear Stare IP
#1 - 2012-04-18 08:46:55 UTC
Hey everyone, I am fresh out of planning school and I want to put some of my quantitative skills to good use. Using EVE market data I want to create 1) Input-Output model for the EVE economy, 2) A gravity model of trade. Essentially I want to rig together an IMPLAN model for EVE from scratch. If you aren't familiar with I-O analysis combined with gravity model, it is as close as you get to a total economic simulation. That is deep, a simulation of a simulation.

it can answer questions like

What is EVE's GDP
How far will someone be willing to travel to purchase a good
Is there clustering taking place, where are they, and what type of resources or services could these clusters benefit from
What areas are under-served for certain goods
What economic sectors most boost the local economy
What happens to the economy if you raise tarrifs
What is the cost/benefit ratio of flying through null sec
How much will prices rise in other trading hubs if Jita is shut down for a day

There is lots of potential to profit from this data, although my interest is mostly academic. My goal is to just present results from the study on a blog somewhere and offer market advice to large corporations

The input-output tables are obviously easy to come by, from the data dump. Calculating demand for every item in every sector, well, it will require some database manipulation and automation to run all those regressions. Fortunately, I do have the computer that can handle it. I just need to find a way to automate the process

I was just going to do it for fun to see where it would lead, although if I found people who wanted to help develop this into a web site, that could be interesting. I am very poor with programming. I could use help with MySQL, getting the data queried into manageable batches (I need to run at least two regression analyses for every item in every system...that's a lot of regressions!). And more statistical rigor wouldn't hurt, if you have a PhD in Economics, Statistics, or Geography. I am alright with SPSS, but if you know how to crunch enormous datasets in R, that would be helpful. Finally, if anything comes of it, I would need to publish the data somehow. I am thinking of using Tableau, but I would probably end up WordPressing the results. I am sure someone who was clever could route the daily EVE-Central data dumps to create a real-time glimpse of the EVE Economy, but I would be satisfied if I could just create a PDF report.

Thoughts?
ComradeEngineer
Deep Core Mining Inc.
Caldari State
#2 - 2012-04-18 09:13:51 UTC  |  Edited by: ComradeEngineer
Pointless exercise. Why try to simulate an economy when you can simply work from observational data? The simulation will only tell you what your already know. Hell, I can answer the questions:

How far will someone be willing to travel to purchase a good?

Does not matter. All one needs to know is whether or not your good sells at a rate according to your time preference.

Is there clustering taking place, where are they, and what type of resources or services could these clusters benefit from?

Look at the star map and use relevant filters.

What areas are under-served for certain goods?

Where prices are comparatively higher.

What economic sectors most boost the local economy?

Trading hubs and areas where there would be the factors of production.

What happens to the economy if you raise tarrifs

Raise prices and resources will shift to unhindered sectors of the economy.

What is the cost/benefit ratio of flying through null sec

That is subject to the whims of the individual. The very fact they seem to be doing so would entail they value doing it more than their alternatives.

How much will prices rise in other trading hubs if Jita is shut down for a day

They will raise. How much? The only way to know is to shut it down. Your model will certainly not have the predictive power to determine this.

Economics and mathematical modelling should not mix...Economic laws are only qualitative.
Johnny Frecko
Federal Navy Academy
Gallente Federation
#3 - 2012-04-18 09:29:07 UTC
^ that.

You can use regressions to figure out vague things like the effect of taxes on trade amount.
However you'll need information of various trade rates throughout many different months, and having all gross trade amounts etc.

Because so many things in eve have not changed, It is infact hard to run any sort of regression on economic variables.
if i would run any regression it would involve at least 10-15 variables, and that still would probably only explain something like 70-75% of the deviation we would see in the data.
I don't know if there's enough information logged of older times to run any sort of regression on the variables.
in RL you run regressions on data gathered over 10-20 years(to be accurate).
And still the further you go from the averages, the more un-accurate your result becomes.

How much will prices rise in other trading hubs if Jita is shut down for a day
For a day?
50% of the population wouldn't even know jita was shut down, unless you plan it in advance and tell everyone.
People would just scatter to buy what they REALLY need right now, and others would just wait it out till jita comes back online.
no one would shift all their stocks from their hangers in jita to rens for less than 24 hours, set sell orders, and then cancel them to move them back to jita.
Seriously - no one of importance.
Vera Algaert
Republic University
Minmatar Republic
#4 - 2012-04-18 09:50:58 UTC  |  Edited by: Vera Algaert
Ewo Deveraux wrote:
Thoughts?

make sure you really understand the scope of the problem you try to tackle before investing significant effort into this venture

you have no idea about stocks, you cannot separate production/consumption from reselling, you don't have comprehensive market data (for all items/regions), there are important markets that you will never get decent information on (e.g. public contracts, or intra-alliance markets [ship building programs, reimbursements etc], ...), you have no idea how much raw materials are introduced into the economy or which/how many items are removed from it, there are constantly strong exogen. shocks (such as announcements for upcoming features/changes or in-game politics) that will make people behave in weird ways and mess up any model you come up with, ..

If you take the whole of EVE as one system then you have no way of knowing the inputs, no way of knowing the outputs (that leave the game), no direct way of determining the level of stocks, ... all you can see is a small fraction of the trading going on in EVE.

I wouldn't even know where to start.

.
Caleb Ayrania
TarNec
Invisible Exchequer
#5 - 2012-04-18 09:57:42 UTC
I would say do a test run, and involve MD in your results.

What EVE is in desperate need of is more things like this.

If the ingame economy is going to be taken serious, more serious work like this is going to help..

Sure there might be things that limit or makes it less relevant, or not directly relatable to RL economies and calculations, but its still worth testing to see the resulting answers.