IB FIX CTCI interface experience

Can't comment on further private issues. Are you in HK as well?

 

My tax is based in HK, but been living in kamakura for the past 1 year.

 

btw, which language do you mainly program in IB? Aware of any other simpler API (for equities) than IB's for one to get started?

 

I use the C# api for IB. One should not start with which api is simpler but which broker or dataset suits your needs. The IB api is not my choice, it's a badly written api but if I want to use IB I have no choice than to deal with it.

 

Hi, thanks for the post.

1- Is Interactive Brokers directly listed in Equinix list of providers of their Fabric or they are located in CH4 and a cross-connect has to be coordinated directly with IBKR and Equinix?

2- You said 10micro seconds. This is crazy good. What sort of connection did you test this with? and also what action was this measure for specifically? (placing order, streaming data,...)

Thanks,

 

Hi Luo8, thanks for the informative post. I see you didn't get much far here as no one seems to have the experience.

Were you able to setup or test your FIX solution whether over a dedicated line to IBKR or over VPN or internet? If yes, what were you speed findings and other interesting things pertaining to FIX?

Thanks,

 

Hi Luo8 did you off the ground with this, and if so what was the total cost? Also, I am guessing that using FIX and IB is a necessary evil here? As there are number of other brokers who could connect you natively to these markets using the exchange protocol and data feed, NewEdge (SocGen), UBS, Nomura, etc... but I am not sure what their capital requirements are, Appreciate if anyone can comment on this as well.

 

Has anyone actually seen the FIX interface to be significantly faster than TWSAPI?

In my experience with the TWSAPI, orders can take a very long time to get routed to the exchange, hundreds of times as long as my ping to the IB server. I strongly believe that ping/latency is not the limitation when using TWSAPI. Instead it must be slow risk checking, or some kind of throttling. Question is, does using FIX make any real difference here, other than reducing latency by a few ms?

 

I thought that if you want to colo, you must use FIX. And FIX is not available otherwise.

 

Yes, this is true for let's say your own build bridge (EMS/OMS) and most broker provided frontends. But systems, such as Orc and GL Trade that I used many years ago used to have their own native connection/gateway written in the exchange native protocol, should be added that some exchanges only used FIX protocol anyway as the gateway interface. But most have their own protocols for order routing and data, Nasdaq for example is the old OMX technology now a days and uses ITCH and OUCH protocols. These native protocols are a magnitude faster than FIX, moreover the amount of data that you need for HFT if you are market making for example on a handful of instruments is not even comparable between the native data solutions directly from the exchange with FIX, even with the FAST compression of FIX you would still not be able to make any sensible and proper HFT strategy out of. I gave up on the race some 11 years ago and moved to optimal execution strategies instead of HFT. Although I never really considered hiring people and building my own stuff, I jut moved over to strategies where I know I don't get to have the first picks, but being guaranteed to trade within my offset.