Market Discussions

I see, so to basically while the segmentation of the market in candles is arbitrary, the numbers they enclose are not, and by segmenting those we can get a clearer picture of what the market was doing in that segment of time. As such we can rely on the sequence of those candles to give us a picture of how the market has been evolving over time not only in high/low/avg/daucion(sp), but also in buy vs. sell giving a potentially clearer picture of movement vs. graph, because graph isn't segmented. (although I guess with daucian(sp) channel you could still have the "wick" sort of. But by enclosing that in segments/candles we can remove complexity by way of abstraction.

I guess I should dive into the older thread though, but it's all quite interesting. My current strategy has been to simply find stable fast moving products and set up buy and sell orders that get triggered in the normal weekly movement of the product price. My all time favorite item being DC2's that more often then not have a almost flat price with very little fluctuation. But still a workable margin between buy and sell so that I don't have to adjust my prices all that often. (why yes I'm lazy)

Although for the last 2 weeks I've been writing my own application to be able to rapidly automate analyzing of market items to find hidden gems. Gaps the market hasn't yet picked up on. Luckily being a developer of trade the building of this software was fun and easy. Unfortunately I have little market knowledge so felt a bit lost as to what to do with all this data that I can now so easily collect. Which incidentally is why I stumbled upon this post. I don't think the method you describe lends itself easily for automation, but it's interesting in it's own right.

I'm thinking about computing either the avg base production cost of items to see how much of a "bubble" the market value is vs. production cost, or maybe compute the reprocessing value vs. my own skills and see if I can pick up stuff that's worth more reprocessed then it's market value. (Which I've seen happen once or twice)

The easiest would probably be to pick two market hubs and do a comparative price check and either haul or hire a hauler to make my profit. But meh, that sounds too much like actual work.

Welcome Back VV. I've got at least one corp member poking their head in this subforum now, so will point out this thread and the old one. I was actually preparing to re-post internally what I pulled from your site, when this thread re-surfaced.

After giving up on trying to re-synthesize the OHLC from CCP's history files, I started my own research for alternate TA scanning methods (with emphasis on something that leverages the existing Eve Graph). I may have actually found one, but won't go into detail until I read more and try it a bit.




As already mentioned, some have implemented the algo's involved already - But, the one 'public' product is web-based so there's some trust involved.

EDIT - The methodology I'm exploring would use the existing Eve data (for 'at a glance' scanning once you get used to it), but also requires you view the markets using the client (a dealbreaker for some, maybe most). I've already used a primitive form of this in a couple mid- to long-term trades, and profited from all of these so far.


@ Callean - screenshot please? Are you actually displaying candles, lines etc or just doing the OHLC synthesis?

1) If you have lots of time you may even fetch open and close by carefully parsing EvE Central data's first / last orders of the day. It's approximate though but probably well OK for most actively traded items (> 30 transactions a day).

2) About your question to Callean, go to this URL and then click the "Trends" tab.

@Callean

You know, appetite comes with eating.
What if under Options and Tools you added the last tab data export: Trends?

The EOD (end of day) industry formats are several but this is the most common:


2011-12-23,2829.87,2839.97,2774.16,2794.00,18058104
2011-12-24,2794.00,2838.94,2732.11,2799.87,12132568
2011-12-25,2799.87,2800.00,2737.11,2799.82,10524017

where

First column is YYYY-MM-DD (0 padded ofc if needed)
Second is Open
Third is High
Fourth is Low
Fifth is Close
Sixth is volume (optional for some platforms).

There are also other standards I can give you if you so desire.


I have to say that your website is functional and fast, with no fluffy-sluggish garbage others had.

Since I have enough money to not know what to do with it, If you implement the changes + data export above, I am going to donate you 1B as a sign of appreciation.

Yup, will be done, the only thing that prevents it from being online right now is that the new lib I'm using isn't as fire and forget as google's API was, so the graphs are ugly now, and I need to do some polishing :)

Have you tried something like this free graphics package?

The site here is very slow atm, the search querry was not loading on "enter" and a product is not loading beyond the second tabel heads.

Note: the site keeps giving a loading sign in the top corner though

1) Will see what is going wrong with the exported data format, though I think it might be possible to have a lower open than low due to the way the data is generated (I can certainly use a different way, but that will probably only work for minerals, so I'll have to provide both options).
2) Data for Megacyte might very well stop 30 days ago, I do not control the data is uploaded (or, well, I control it to a lesser extend).
3) It probably crashes (crashes? Or does it simply not show anything) because there isn't any candle data in the last 30 days (as you noted earlier, there is no data for those days).

EnyoG would be impressed

Callean,

I see you have updated the website as I suggested to.

I did a brief consistency check between some of your data samples and mine, taken on those days were data was not clean:


Your gathered and sanitized data





My own gathered and sanitized data



As you can see the differences (besides separator) are on the fourth decimal digit (you should round it to 2 digits imo, that's what EvE prices do), the match is basically perfect.


Notice how now the weekly candle bars graph looks completely clear, it is almost screaming "trade me now!"


Therefore, as I said above, I am going to send you 1 Billion as reward and tangible support for your outstanding job!