Search
× Search
Tuesday, November 19, 2024

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

Trading The US markets from overseas

photo

 Muhammad A., Independent Day Trader at Equity Day-Trader

 Monday, September 23, 2013

my strategies do not require extremely low levels of latency, but I was wondering how much deterioration, in latency that is, to expect if one were to trade from overseas like from Singapore or Hong Kong or Dubai. I am assuming its going to still be under 1/10th of a second which is tolerable.


Print

4 comments on article "Trading The US markets from overseas"

photo

 Scott Boulette, Algorithmic Trading

 Wednesday, September 25, 2013



@Alex - I can see where you would ask that question; it is perfectly reasonable. My comment to the OP was very generic to avoid complicating the discussion. The short answer is, the devil is in the details, as the saying goes.

My setup involves 4 separate processes spread over 3 - 4 machines. Depending on the configuration, in some cases 1 machine can handle 2 processes: analytics and monitoring.

There is the actual trade server (no logic whatsoever) running co-located in Aurora (CME) and has ping times in the low microseconds. The trade logic server is co-located with the trade server and contains very little logic (primarily reacts to prices in predetermined ways) and uses the analytics server that does the heavy lifting in terms of calculations required to set the parameters for the trade logic server (tick by tick calculations but distributed on a 1 minute schedule).

Last there is the parameter selection/data visualization/monitoring server that always runs wherever I happen to be (I am working on a smart phone/tablet controller but that will be awhile).

A typical scenario is I monitor the trades to determine if anything going on requires changing any parameters (normally not), I transmit any parameter change requests to the analytics server and to the trade logic server which in turn makes trade requests to the actual trade server and reports back to the monitoring server (located in my office).

The beauty for me of distributing the work load this way is I get all the benefits of rich monitoring/data visualization (metrics, graphics, etc.) with zero performance impact on the trade logic. As long as I have more cores than threads on the trade logic server, there is virtually no context switching. My price/trade message loops are single threaded to avoid context switching.

There is the added benefit of I get to feel really cool for making it all so complicated yet still work seamlessly :)


photo

 Alex Krishtop, trader, researcher, consultant in forex and futures

 Wednesday, September 25, 2013



Scott, thanks for clarifications. I agree with every word you say. The only problem that appears regularly is to explain investors why they need to support that "large" number of machines :)


photo

 Scott Boulette, Algorithmic Trading

 Wednesday, September 25, 2013



Alex, indeed that is difficult at times. I have been very lucky in that area.


photo

 Muhammad A., Independent Day Trader at Equity Day-Trader

 Friday, September 27, 2013



thanks

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association