DukasCopy

The files are available at http://freeserv.dukascopy.com/exp/. Also www.dukascopy.com/swiss/english/data_feed/csv_data_export/.

But you can't start the downloading of all the files and go and do other things - you have to export the data you want one-by-one. Scroll about 4/5 of the way down to get to the currencies available. They have |f| next to them. You can choose any format you like, but make sure you choose "Windows" in the "End line" field.

The good part is that they have data going back to 1986 for some currencies.

DukasCopy provides indicative data, which has no spread. During backtesting, make sure you add on a few pips in the "Commission" field to simulate the spread, otherwise your backtesting results will be overly optimistic.

The Currency Secrets website has covered the issue of indicative data in some depth, so I point you to their pages:

• Beware Of indicative data
• Is indicative data really *that* bad?

p.s. I think their conclusion is that as long as you add the spread, it should be ok.

I guess it's always better to use the prices from the broker you are going to use, if possible.

Once they are finished downloading, follow the rest of the instructions to load text files. DukasCopy historical prices are already in price bar format though. So you don't need to do the steps to create price bars, unless you want to create other price bars not provided by DukasCopy.
 

The File format and Dates in the file(s) fields on the File Loader window have standard settings based on where you got the files from. (You could also just look in the file).

By "file formats", I mean how the price data is arranged in the text file. As an example, here's how Gain Capital price data looks like:

277248713,EUR/USD,2006-11-05 17:00:02,1.271600,1.271900,D

Comparing that with one from an Oanda text file, you can see they are completely different:

13/02/04 08:36:34,1.280600,1.280800
 

Gain Capital

 
Older files up to and including 2009/05 May/Week3

Order of fields - ID, Currency, Data, Bid, Ask, "D"
Delimiter - Comma
Ignore first row - No
Date format - yyyy-MM-dd HH:mm:ss
Timezone - New York (GMT-5/-4)
 

Files from 2009/05 May/Week4 up to and including 2009/11 November/Week3

Order of fields - ID, Currency, Data, Bid, Ask, "D"
Delimiter - Comma
Ignore first row - Yes
Date format - yyyy-MM-dd HH:mm:ss
Timezone - New York (GMT-5/-4)
 

Newer files from 2009/11 November/Week4 to present

Order of fields - ID, "D", Currency, Data, Bid, Ask
Delimiter - Comma
Ignore first row - Yes
Date format - yyyy-MM-dd HH:mm:ss
Timezone - New York (GMT-5/-4)
 

Oanda

Order of fields - Date, Bid, Ask
Delimiter - depends on what you requested from Oanda. XML files cannot be used.
Ignore first row - check the file
Date format - dd/MM/yy HH:mm:ss (note only 2-digit year)
Timezone - Greenwich Mean Time (GMT)
 

DukasCopy

Order of fields - Date, Time, Volume, Open, Close, Min, Max
Delimiter - depends on what you requested from DukasCopy
Ignore first row - check the file
Date format - can't remember - you should check the file
Timezone - Greenwich Mean Time (GMT)

Something to be aware of with these files is that there is no spread included. Therefore ensure you use a few pips of commission when backtesting, to try and mimick the spread. If you fail to do so, you're backtesting results will be overly fantastic.
 

ATM price history files (*.tsphd)

Order of fields - ATM Price History File (*.tsphd)
Delimiter - choose any
Date format - choose any
Timezone - choose any

The File Loader has some mandatory fields that aren't actually mandatory in this case. Just choose any value to get past the mandatory field check.

As per this wiki entry, the price history files have very specific names, e.g. fxcm_23_1d.tsphd. The ones being used by ATM (i.e. the ones which correspond to Data Manager entries) cannot be renamed at all. But if you are wanting to merge a price history file with another, most likely it means you got one from somewhere or someone else. This file you can rename, if you are going to use the File Loader window to merge it. But make sure you keep the file prefix intact. That is, add another dot and type after it, e.g. fxcm_23_1d.temp.tsphd or fxcm_23_1d.copy.tsphd.
 

Text files exported by ATM v3+

Order of fields:

• for files containing tick prices - Date, Utc, FIP ID, Symbol, Bid, Ask, Volume
• for files containing price bars - Date, Utc, FIP ID, Symbol, Bid OHLC, Ask OHLC, Volume

Delimiter - depends on what the setting was when the file was exported.
Ignore first row - check the file
Date format - yyyy-MM-dd HH:mm:ss
Timezone - choose any

As the date and Utc offset is written into these files, the timezone field is not actually required. So just choose any value to get past the mandatory field check.
 

Text files exported by ATM pre-v3

Order of fields:

• for files containing tick prices - Date, Currency, Bid, Ask
• for files containing price bars - Date, Currency, Bid OHLC, Ask OHLC

Delimiter - depends on what the setting was when the file was exported.
Ignore first row - check the file
Date format - yyyy-MM-dd HH:mm:ss
Timezone - Greenwich Mean Time (GMT)
 

Other

ATM supports just about any date format, and any 1-character delimiter. So if you have a text file from somewhere else, all you have to do is somehow get the order of fields to match one of the available options.

If that's beyond your technical prowess, or you know of another popular file format that you think we should support, you can add a post to the feature request forum.