Page 1 of 2

Stupid question about GAIN data

Posted: Wed Feb 28, 2007 7:22 am
by Toothfish
Hi,
I have downloaded all the data from

GAIN capital data:
http://ratedata.gaincapital.com/

however it seems there is only bid and ask price. There is no traded price.
Am I misunderstanding the forex market?
Also the time indicated would you know which time line ie Greenwich? USA Eastern?
Please forgive my stupidity as I am new to the currency market.
I was going to use this data to play around and design some strategies.
Many thanks in advance for your time

Posted: Wed Feb 28, 2007 8:50 am
by michal.kreslik
Hi, Toothfish,

I'm using the same data in my giant SQL Forex history machine database :)

You see it correctly: there are only Bid and Ask prices. And before 2004/4, there are many cases when the Bids and Asks are screwed up - intermingled, nonsensical values, etc. It takes a man to sift that data out, I can tell you.

The time used is ET (not the Extra Terrestrial, as it might sometimes seem when you look at the data, but Eastern Time).

Michal

Posted: Wed Feb 28, 2007 9:50 am
by Toothfish
Hi Michal,
Thankyou very much for your reply.
So how do you use the data from GAIN?
Isn't it pretty useless without the traded price?
I'm trying to get historical data on the forex to play around with and design some systems but don't want to pay thousands of dollars for data yet. I understand it's not going to be reliable if it's free.
Is there any source you could recommend.
By the way thankyou so much for your site.

Posted: Wed Feb 28, 2007 10:27 am
by michal.kreslik
It is going to be reliable if you spend of couple of months developing the database and an application that makes sure the data is relevant :)

There is about 30% of duplicates, for instance, mainly in the older data. So you just have to make sure you remove those duplicate ticks. But at the same time you have to make sure you don't remove any unique ones. It gets tricky for a couple of reasons (one of them being there's only a second time stamp, so the ticks with the same second and Bid and Ask prices that really happened one after another need to be discerned from those that are duplicate), but it can be done.

I'm simply using the Bid prices for my further calculations. In Forex, there are only Bid and Ask prices. As far as I know, knowing which one of the two happened first doesn't provide you with any extra edge. By the way, I was able to fine tune my SQL database so that the average processing time for one tick in a statistical calculation (when the ticks are getting loaded sequentially) is about 4 microseconds.

I've moved this topic to the datafeed subforum and I left a shadow topic in the original beginners forum.

Michal

Posted: Wed Feb 28, 2007 10:51 am
by Toothfish
Hi Michal,
Thankyou for your beacon of light.
I have just come up with what I think is a cunning although compared to your SQL is pretty dumb.
To separate the date from the the time which if viewed in Excel gets put in the same column as there is no comma separating them
I used wordpad and do a findall and replace blank spaces with commas.
This allowed me to cut out the date, time and bid columns.
From there I have the arduous task of joining the years together.
This looks like it's going to take a few months and I have just cancelled any social activity off my diary.....which wasn't much anyway.
Again thankyou
It looks like I have found a new road and I need to start walking.

Posted: Wed Feb 28, 2007 11:06 am
by michal.kreslik
Toothfish, just come to our pirate meeting this weekend in Tucson and you can have my SQL copied on your harddisk :)

But it's 100GBs big with all the indexes, so it's no good posting it or transmitting it over the internet :)

Michal

Posted: Wed Feb 28, 2007 1:10 pm
by Toothfish
Thankyou very much for your generosity Michal but I'm in Australia. It makes me want to get on a plane as it would be worth the flight.
I appreciate your time answering the questions already.
I'll be okay. Sometimes we have to do our own dirty work.

Posted: Wed Feb 28, 2007 5:49 pm
by geff
Michal, are you sure you can't E-mail a simple 100GB attachment??? (lol)

How's this for a solution: You can now buy a 100GB external drive for under $100 - the same as one month of Trade Station!

Huge data transport has never been so affordable. Ah, the possibilites.

Jeff

Posted: Wed Feb 27, 2008 1:48 am
by Rauss
777л

Posted: Thu Feb 12, 2009 11:25 am
by Toothfish
Hi,
I'm still trying to make sense of the GAIN data. (2 years on) ( I have been trading equities and have just come round to looking at trading forex again)
In each week of data, the time seems to start from 00:00 but goes to 60:00 instead of 24:00hours. Could someone explain this please?
Many thanks

..... After a couple of good night's rest the only answer is data error
I think the data is fine after 2006.