I'd guess that IQFeed is aimed at people that aren't trading with such critical requirements. NXCore is the replacement for DTN Satellite and it's presented on a whole 'nuther level, as a customer you can talk to the engineer in charge and get things defined right down to the real detail.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
IQ Feed Latency
- Thread starter clearinghouse
- Start date
Hi folks, I thought I would chime in to clarify a few points. Both our NxCore and IQFeed data services are used by traders with stringent requirements. While IQFeed is less expensive, that doesn't mean it is of cheap quality. We purposely keep our marketing to a minimum and do everything we can to keep costs down on this product in order to make it affordable for the broadest set of traders. We don't compromise on things needed to deliver a quality service. I feel it is important to have 800# support for users and for developers to have access to our development staff via RT chat. I personally listen to feedback as I want to ensure I hear what our customers need.
IQFeed and NxCore effectively serve 2 different needs however. NxCore is a full broadcast service (full exchange feeds) while IQFeed is a symbol based service. Although IQFeed was initially used to feed screen based trading systems, the quality and speed of the data has resulted in it being used by everyone from individual traders to large algo and prop firms.
Because it is internet delivered, latency is a tricky variable. In a large majority of cases, latency can be tracked to the reading of the data off the local socket as opposed to latency of delivering the data to the users machine. Because we send more data than other feeds, local apps will need to process the data faster than if you were using something else. Increasing processing efficiency could involve tweaking code within your socket buffers, utilizing the dynamic fieldsets, or simply increasing the CPU on your machine. Because this local processing is where we have seen developers have issues, we are working on providing better statistics to developers within the IQFeed client. This should help them identify when the IQFeed client is queuing data because it isn't being read off the socket fast enough.
Jay Froscheiser
jay.froscheiser@telventdtn.com
IQFeed and NxCore effectively serve 2 different needs however. NxCore is a full broadcast service (full exchange feeds) while IQFeed is a symbol based service. Although IQFeed was initially used to feed screen based trading systems, the quality and speed of the data has resulted in it being used by everyone from individual traders to large algo and prop firms.
Because it is internet delivered, latency is a tricky variable. In a large majority of cases, latency can be tracked to the reading of the data off the local socket as opposed to latency of delivering the data to the users machine. Because we send more data than other feeds, local apps will need to process the data faster than if you were using something else. Increasing processing efficiency could involve tweaking code within your socket buffers, utilizing the dynamic fieldsets, or simply increasing the CPU on your machine. Because this local processing is where we have seen developers have issues, we are working on providing better statistics to developers within the IQFeed client. This should help them identify when the IQFeed client is queuing data because it isn't being read off the socket fast enough.
Jay Froscheiser
jay.froscheiser@telventdtn.com
I had this problem once. I started complaining to IQFeed.
They proved, that I was wrong. The problem was the bus speed of my computer (in general - the channel capacity). I installed a reasonably powerful server and the problem disappeared. I keep track of lag between IQFeed time stamp and mine. When market is inactive lag can go up to 300 sec., but in the middle of the day - it is never bigger than 2 sec.. Average value is close to 1.15 sec.
They proved, that I was wrong. The problem was the bus speed of my computer (in general - the channel capacity). I installed a reasonably powerful server and the problem disappeared. I keep track of lag between IQFeed time stamp and mine. When market is inactive lag can go up to 300 sec., but in the middle of the day - it is never bigger than 2 sec.. Average value is close to 1.15 sec.
Quote from abarzov:
I had this problem once. I started complaining to IQFeed.
They proved, that I was wrong. The problem was the bus speed of my computer (in general - the channel capacity). I installed a reasonably powerful server and the problem disappeared. I keep track of lag between IQFeed time stamp and mine. When market is inactive lag can go up to 300 sec., but in the middle of the day - it is never bigger than 2 sec.. Average value is close to 1.15 sec.
Do you mean seconds or milliseconds?