Data differences - What can you rely on?

Sources of data to import to the program
Message
Author
pthomas82
Posts: 60
Joined: Tue Aug 10, 2010 8:39 pm

Data differences - What can you rely on?

#1 Postby pthomas82 » Mon Oct 21, 2013 8:58 am

Hi Guys,

So my old thread: http://forextester.com/forum/viewtopic.php?t=3577&postdays=0&postorder=asc&start=15 I decided the best way to test was to use multiple sets of data.

I have:

Dukascopy 1 min data (my broker)
Alpari Uk
FXDD
Forexrite.

I have built a system which works in one set of data (dukascopy) , but can not confirm in alpari / fxdd or forexrite.

I am at a stage with this program that I dont think it is of any value. Even if you build something that works, there is no way to test if it actually works unless you port it into another software to confirm.

I have built many systems for forextester (maybe > 200) and I am really doubting if it works. The tester is only as good as the data, why does the data differ so greatly? 3 of the 4 data was acquired from your servers (paid subsciption).

If it is the types of system im building (generally with 20-30 pip stops and on 15min - 1hr timeframes). What should I be building - daily systems?

Are you having the same issues?

Have you built / varified and live traded a system with this?

Should I just give up and realise that I have wasted my money and countless hours of my time?

pthomas82
Posts: 60
Joined: Tue Aug 10, 2010 8:39 pm

#2 Postby pthomas82 » Mon Oct 21, 2013 6:56 pm

Ok - so some more homework.

Dataused - Alpari (tick) / FXDD (tick) / Forexrite (1m) / Dukascopy (1m)

The system used is a simple moving average cross - 50ma / 200ma with a 20 pip take profit and 20 pip stoploss.

Simple stuff right.

On each set of data, over the same period 2011 full year. the results are different.

I'll load into dukascopy's backtester shortly and update their result.
Attachments
datacheck.dll
(97 KiB) Downloaded 282 times
datacheck.dpr
(2.93 KiB) Downloaded 276 times
FXDD.jpg
FXDD.jpg (457.86 KiB) Viewed 6605 times
forexrite.jpg
forexrite.jpg (451.92 KiB) Viewed 6605 times
dukascopy.jpg
dukascopy.jpg (428.38 KiB) Viewed 6605 times
Alpari.jpg
Alpari.jpg (449.06 KiB) Viewed 6605 times

FX Helper
Posts: 1155
Joined: Mon Apr 01, 2013 3:55 am

#3 Postby FX Helper » Tue Oct 22, 2013 9:44 am

Hello,

Forex Tester just emulates trading on data which it has. If historical data are different then the results will be different too. Each broker has his own data so data from different brokers always will be a bit different.

If you want to find the reason of different results then please check each trade which strategy makes in "slow motion" so you will see why your strategy does particular actions.

readyyy
Posts: 14
Joined: Sun Sep 04, 2011 6:36 am

#4 Postby readyyy » Sat Nov 02, 2013 6:41 am

You seen to have a similar issue that i had.

The thing is, you should never use 1 Minute data without Tick data as this produces unreliable results. To produce reliable results, you should always use 1 Minute Data that includes Tick data.

Here's the reason: Let's assume a 1 EURUSD Minute bar ranges from 1.3000 to 1.3009. In your strategy, you have a sell market order to be executed at 1.3004.

In this case, using 1 Minute Data, your order will not be executed at 1.3004, but at 1.3009 because it is the maximum of the bar. This can lead to an "optimized" scenario in which the results look better than they really are.

If you would have used tick data however, the order would have been executed exactly at 1.3004, which produces a much more realistic result.

User avatar
Coretta
Posts: 10
Joined: Tue Oct 22, 2013 12:51 am
Location: China
Contact:

#5 Postby Coretta » Tue Nov 26, 2013 8:22 pm

Here you can certainly learn a lot


Return to “Data sources”

Who is online

Users browsing this forum: Baidu [Spider] and 2 guests