Interactive brokers data

Questions about MultiCharts and user contributed studies.
aljafp
Posts: 184
Joined: 28 Oct 2005
Been thanked: 1 time

Interactive brokers data

Postby aljafp » 08 Jun 2009

hi guys,
Anyone using IB's realtime data with this option "Use Server Timestamps" checked ?

Is it better than leaving it uncheck and using the PC clock ?

Appreciate if anyone can share their experiences.

thanks

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Postby TJ » 08 Jun 2009

don't touch the MultiCharts default setting.


p.s.
IB does not stream time stamp on its real time data.
most broker/dataproviders don't.

p.s.
The time stamp is added by MultiCharts using the clock in your computer.

When you press the Refresh key, you will notice that the new chart looks different from your pre-refreshed chart.
That's because when you refresh, you are getting "historical" data from IB.
The historical data is time stamped.

Spaceant
Posts: 254
Joined: 30 May 2009
Has thanked: 1 time
Been thanked: 3 times

Postby Spaceant » 09 Jun 2009

Hi there,


Just share some experiences:

1) TJ is right that IB doesn't stream timestamp with realtime data. To my understanding, IB does provide timestamp for historical data.

2) I have tested my old desktop and notebook. One day, I found that the data are difference in terms of timestamp. I finally know that my desktop's internal clock runs faster and faster while my notebook's internal clock runs slower and slower. To synchronize the clock, I now use a software to do the synchronization when computer starts and every one hour, with the atomic clock. Well, my two computers can't run the time in phase, but they are getting better.

Sa

aljafp
Posts: 184
Joined: 28 Oct 2005
Been thanked: 1 time

Postby aljafp » 09 Jun 2009

guys,
thanks for all your inputs.

Yeah i know its impt to have a time sync software running.

The reason i am asking about "server timestamps" is that i find that the opening tick for the start of the day are sometimes way off and i am looking for a way to improve that.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 17 Jun 2009

Hi guys,

Just to clarify a couple of points:

The option to use TWS bar timestamp must be checked for TWS versions 873, 874, 875 (because the same minute data can be time-stamped differently in different versions). For all other versions of TWS, this option must be disabled.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 18 Jun 2009

Qcharts had terrible feed delays with quotecom data (3 to 5 seconds standard and very often up to 3 and 10 minutes at any time without any warning). I often wondered how chart delays would effect my decisions (obviously a 3 minute delay means you stop trading). My bad Qcharts quotecom experiences is why I will only use IB data. I figure it is about as fast as I can get. However, I often wonder if IB data has some delays and wish MC had a solution to this curiosity.

I emailed the IB technical people over a year ago and they say that they do provide an exchange based time stamp with their real time feed. If this is in fact correct then even if MC does not use it for the actual charts due to programming complexity issues, if MC could at least have the trader selectable option to test it (when it is not blank and not corrupt) to detect if the difference between the "exchange time stamp" and the "trade arrival time" is too great then MC could alert the trader that the data has an excessive delay. It would also be useful if the trader had the option to turn on a log to get the min, max and average delays to get a feel for how they may want to set their alarm. Maybe the news would be good and the trader finds they never have an excessive delay. On the other hand maybe the news is not that great and in time some traders may even figure out ways to adjust their studies based upon this log if they could get at this delay info somehow.

An after thought. If there are stamping issues (of any complexity level at the exchange - including inverted sequence issues) that causes problems with charting then these issues could be reported with 100% disclosure in the log without effecting the charts. I am curious and would love to know of these issues and when they occured during the trading day.

janus
Posts: 835
Joined: 25 May 2009
Has thanked: 63 times
Been thanked: 104 times

Postby janus » 18 Jun 2009

I emailed the IB technical people over a year ago and they say that they do provide an exchange based time stamp with their real time feed. If this is in fact correct then even if MC does not use it for the actual charts due to programming complexity issues, ...
You make some very good points. If indeed IB does timestamp their real-time date, then we should use it to avoid inconsistencies when re-loading a chart and uses historical data. I'd always assumed IB didn't. However, even if they do, we would need to make sure the timestamps of the real-time feed corresponds precisely with those of the historical ones. Any discrepancies should be updated with the historical ones in case IB for some reason changed them.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 18 Jun 2009

I know Quotecom use to go back and correct their data nightly but that was partially because they were stamping from the servers. It would be interesting to know why an exchange based stamp would have to be changed other than rolling the trade back and actually removing the trade. Otherwise only price would change. Maybe the vendors pick up these corrections later. However none of this would effect an alarm system or log. An alarm system could be programmed to take an average then give an alert (dumping anything out of the ordinary).

Maybe I am wrong here but I don't think so. Here is something to consider if getting every correction is really that important. If IB does do corrections and they do it at midnight in a batch run (I think that is actually when they do it) then if you start MC at 6am in the morning with an old database so you get lots of backfill (supposedly corrected) then the data between midnight and 6am is going to be live and not corrected and saved to the database when you close MC down. The only way to keep your personal MC database 100% up to date with IB corrections is to download nightly immediately after their fix run then shut MC down and save that database for the next night when you would do the same thing again (every single night or once a week if you value your sleep). I personally don't think it is going to make that much difference in backtesting unless that data is really out of whack. I think a good system will trade well if tested with some bad data. After all isn't it going to be trading live with some bad data. It is probably better to backtest with some bad data and just leave it alone. However I do remember really strange bars when I was trading stocks. The Emini does not seem to have this.

I am pretty sure part of the IB nightly shutdown is to consolidate their tick data into special data for daily bars to reduce size. That of course would partially explain the 12 midnight time of the batch run.

janus
Posts: 835
Joined: 25 May 2009
Has thanked: 63 times
Been thanked: 104 times

Postby janus » 18 Jun 2009

I think we are both getting too fussy but there's no harm in striving for more accurate data. I'm actually more interested in making sure there;'s no bad/spiky data. I'm not sure what IB does to clean up their data. In any case I'm developing sophisticated techniques to filter such bad spikes in real-time, such as comparing trade with bid and ask of the current and other symbols (via global variables). I hate the thought of a bad trade due to a 200 point spike in say the DOW. Maybe I'm too fussy but I like my sleep.

brodnicki steven
Posts: 407
Joined: 01 Jan 2008
Been thanked: 3 times

Postby brodnicki steven » 18 Jun 2009

I think we are both getting too fussy but there's no harm in striving for more accurate data. I'm actually more interested in making sure there;'s no bad/spiky data. I'm not sure what IB does to clean up their data. In any case I'm developing sophisticated techniques to filter such bad spikes in real-time, such as comparing trade with bid and ask of the current and other symbols (via global variables). I hate the thought of a bad trade due to a 200 point spike in say the DOW. Maybe I'm too fussy but I like my sleep.
** I'd love some type of real-time tick filter, I've even suggested it a few times. Just look at MSFT data and you'll see many spikes with IB data. No mechanical system could use that and I get tired of editing the data. Some provision for a custom tick filter would really be nice. (hint hint)

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 19 Jun 2009


In any case I'm developing sophisticated techniques to filter such bad spikes in real-time, such as comparing trade with bid and ask of the current and other symbols (via global variables).
Gee janus, you just reminded me that I did that way back in 1998 when I was trading stocks. It has been a long time since I was interested in stocks. I completely forgot that I did that. I had no bid and ask so I was just using price.


Return to “MultiCharts”