Qlink/Qcharts inconsistent

B

BA_Trader

I recently started to play around with qlink and excel and I noticed
that the realtime data it shows does not match what is shown
on qcharts...

for example looking at the 5 minute ES bars -- for each bar there
is at least one value (OHLC) off by a tick and the qlink volume is
always lower (but not by a constant amount)

what gives?? anyone else notice this or know what is going on?

JT
 
Could be an inability of Excel on your machine to process the message pump that's involved in feeding it DDE messages efficiently. Needless to say, the Excel environment wasn't designed for realtime processing.
 
I can see how that would be an issue for the current bar -- but
the old 'frozen' bars are off... e.g. if I clear everything out and
then load in the most receent 10 bars the first 9 bars are not
consistent w/qhcarts (all 10 are off -- but you see my point)

Since the 9 bars are absolutely static data I wouldn't think excel
is the problem.

JT
 
For all interested...

I figured out what was going on with this.

If you are looking at 5 minute or faster bars each qcharts server
is going to give you slighltly different bars. I don't know why
and this kinda stinks.

to prove it for yourself -- bring up a chart of ES during the
day (not sure if this effect happens after hours) -- show the 5
minute bars... now do CTRL-ALT-N and pick a new server. Keep
your eyes glued on the volume bars. They will change slightly
when you establish a new connection.

The fix for qlink -- you have to make sure qcharts and qlink are
connected to the same server. Qlink will show the address of the
server it picked and you can use CTRL-ALT-N to force you qcharts
to the same.

Disturbing.

JT
 
I was where you are years ago, I feel your pain and dismay. Quote com's data and servers and whatnot is all just so .... I can't really find a word for it because they caused me a lot of pain, confusion and headaches.... Can't collect for that in small claims court so I will just ......LET IT GO. Man that feels good, no more quote com for me.

Max
 
It's because the data stream from the exchange isn't time stamped. So as each data server receives the raw tick feed, they each time stamp each tick as they each receive them and then assemble the bars based on those time stamps.

There can therefore be slight differences based on communication channel latency - e.g., some ticks arrive during the :59th second on one server and in the :00th second on another server and therefore get aggregated in different one minute bars.

The smaller the time bar granularity the more chances of potential slight differences between servers.

They could potentially have the raw datafeed channel through a single timestamp server, but then you've got possible single point of failure issues and then there's the issue of dealing with 3 different server farms in disperate geographies. Choice between reducing single points of failure and a potential for minor differences in bars.

Except maybe for Globex (possibly Nasdaq too), there's also a margin of error in the delivery of the ticks from the exchanges (e.g., the tick that comes down the pipe right now could actually be for a trade that occurred N seconds ago).
 
Just saw this... excellent. Thanks ArchAngel.

Quote from ArchAngel:

It's because the data stream from the exchange isn't time stamped. So as each data server receives the raw tick feed, they each time stamp each tick as they each receive them and then assemble the bars based on those time stamps.

There can therefore be slight differences based on communication channel latency - e.g., some ticks arrive during the :59th second on one server and in the :00th second on another server and therefore get aggregated in different one minute bars.

The smaller the time bar granularity the more chances of potential slight differences between servers.

They could potentially have the raw datafeed channel through a single timestamp server, but then you've got possible single point of failure issues and then there's the issue of dealing with 3 different server farms in disperate geographies. Choice between reducing single points of failure and a potential for minor differences in bars.

Except maybe for Globex (possibly Nasdaq too), there's also a margin of error in the delivery of the ticks from the exchanges (e.g., the tick that comes down the pipe right now could actually be for a trade that occurred N seconds ago).
More...
 
I know people who used Qcharts in the past and they had nothing but grief: inconstistent data/no data/bad data, you name it. From my experience, eSignal is the best feed, I plug the feed into SWP http://www.crontech.com/swp for charting and backtesting and no probs whatsoever.
 
You must log in or register to reply here.
Back
Top