Download MetaTrader 5

GMT-Offset of Backtest-(Tick-) data from Dukascopy.. confusing??

To add comments, please log in or register
Carl Schreiber
7567
Carl Schreiber  

Hi

I read that the Tickdata of Dukascopy should have pure GMT-time so any time offset has to be zero.

Fine,that's what I thought, as soon as I watch the H1-chart of the weekends the DST is switched. In Europe: last Sunday in October (here Oct. 30, 2011) and last Sunday in March (here Mar. 25, 2012).

Sometimes it is 20:00 - should be 21:00 GMT. Sometimes the last Friday-bar opens at 20:00 sometimes at 21:00 - all GMT, NO timeshift!

Can anybody explain this to me?

Please look as well at the numbers of h1-bars in the various Sundays and the the (open) time of the last bar on the Fridays. Sometimes you see 2 bars: 22:00 & 23:00 sometimes 3 bars 21:00, 22:00 & 23:00.

Does anybody knows what to do? How can I be sure to catch correctly 09:00 am GMT on Monday..


Additionally I found because of this code:

Print("CurrentTime: ",sDoW(TimeCurrent()),". ",TimeToStr(TimeCurrent()),"  prvDay:  ",sDoW(iTime( NULL, PERIOD_D1, 1)),". ",TimeToStr(iTime( NULL, PERIOD_D1, 1)));

I get: CurrentTime: Th. 2012.03.22 00:00 prvDay: Tu. 2012.03.20 21:00

Why the hell I don't get: prvDay: We. 2012.03.21 00:00 ????

Thanks a lot in advance!

gooly



here again a 2-bars-Sunday:

Ian Venner
2584
Ian Venner  

Did you look at the downloaded files to see if they contained ticks for those 2 bar Sundays etc ?

Carl Schreiber
7567
Carl Schreiber  

I can't read the original bi5-data of Dukascopy, but the file structure seems to geive a hint:

The folders are: ..PAIR\YYYY\MM\DD\HHh_tick.bi5 for every day, if no data the file is just empty.

Mar. 31: 21h-ticks.bi5, 22h-ticks.bi5, 23h-ticks.bi5.

But the time should already have been switched at Sunday 03:00 am.

Can anybody here open to read the bi5-files to see how the time stamp of the ticks is written?

Ian Flanagan
549
Ian Flanagan  

I don't have the data any more but I remember it was csv. Maybe SQ Tick Data, Tick Story or direct from the dukascopy website

Edit: yes, I found the SQ tick data output folder and there's both bi5 and csv files... can't exactly remember but I guess both came from that software

Carl Schreiber
7567
Carl Schreiber  

I now have seen the Dukascopy data.

The problem seems to be there as Dukaskopy has 3 1h-bars on the Sunday the DST-switch should have been taken place many hours earlier at 3:00 am - it should olny be 2 bars in this particular Sunday night - am I right?

Tjipke de Vries
6404
Tjipke de Vries  

... maybe this is the explanation ..... (check it out)

It is not only brokertime what is depending number of bars at sunday

the world has different timezones....

some timezones are using DST some not

at "sunday" first market going to be open according to me New Zealand

new zealand timechange

Daylight Saving Time DST ended on zondag, 7 april 2013, 03:00 local daylight time.
DST started on zondag, 29 september 2013, 02:00 local standard time.
See time changes/daylight saving time in other years

the market is going to be open at a local time new zealand

that makes if your broker is using GMT it will have a difference in number hourly candels at summer new zealand comparing winter new zealand

Carl Schreiber
7567
Carl Schreiber  

hmm - no,

see here: http://www.forexmarkethours.com/GMT_hours/02/:

Forex market welcomes traders 24 hours a day.
Forex market opens on Sunday 5 pm EST (10:00 pm GMT),
closes on Friday 5 pm EST             (10:00 pm GMT). 

What a broker does is s.th. different and of course the local markets open acc. to their local DST.

But if a Data provider claims my data are GMT-stamped the last Friday h1-bar must open at 21:00 and the first Sunday-bar at 22:00 through out the year!

How shall I find the correct time e.g. London opens and closes if the backtest-data don't have the correct time stamp?

Ian Flanagan
549
Ian Flanagan  
gooly:

hmm - no,

see here: http://www.forexmarkethours.com/GMT_hours/02/:

What a broker does is s.th. different and of course the local markets open acc. to their local DST.

But if a Data provider claims my data are GMT-stamped the last Friday h1-bar must open at 21:00 and the first Sunday-bar at 22:00 through out the year!



unless the opening and closing times are based on the New York market hours.. which are dependant on US DST ... don't know, just saying
whroeder1
16336
whroeder1  
gooly: >Sometimes it is 20:00 - should be 21:00 GMT. Sometimes the last Friday-bar opens at 20:00 sometimes at 21:00 - all GMT, NO timeshift!

False. Those are GMT times when the market opened. Market opens 6pm Sunday New York time. 2100 = EDT 2000 = EST and closes 4 PM Friday local time. An individual broker may open later (close earlier) but not open before 6 pm.

Incorrect. All those sites are wrong half the time. They try to convert local time to GMT with fails during DST.
Alain Verleyen
Moderator
33626
Alain Verleyen  
gooly:

Hi

I read that the Tickdata of Dukascopy should have pure GMT-time so any time offset has to be zero.

Fine,that's what I thought, as soon as I watch the H1-chart of the weekends the DST is switched. In Europe: last Sunday in October (here Oct. 30, 2011) and last Sunday in March (here Mar. 25, 2012).

Sometimes it is 20:00 - should be 21:00 GMT. Sometimes the last Friday-bar opens at 20:00 sometimes at 21:00 - all GMT, NO timeshift!

Can anybody explain this to me?

Please look as well at the numbers of h1-bars in the various Sundays and the the (open) time of the last bar on the Fridays. Sometimes you see 2 bars: 22:00 & 23:00 sometimes 3 bars 21:00, 22:00 & 23:00.

Does anybody knows what to do? How can I be sure to catch correctly 09:00 am GMT on Monday..


Additionally I found because of this code:

I get: CurrentTime: Th. 2012.03.22 00:00 prvDay: Tu. 2012.03.20 21:00

Why the hell I don't get: prvDay: We. 2012.03.21 00:00 ????

Thanks a lot in advance!

gooly



here again a 2-bars-Sunday:

Why are all the Volume equals 1 in your screenshot ?

Ian Flanagan
549
Ian Flanagan  

Well, I'm pretty sure I'm right but no one ever listens to me. Check if the changes from 10 start/finish to 9 start/finish corresponds with these dates http://www.timeanddate.com/worldclock/timezone.html?n=179

Most likely the GMT timestamp is perfectly fine

12
To add comments, please log in or register