What is the most challenging part while coding for HFT? To what extent we can achieve high performance through C++( any particular approach)?Looking forward for some valuable comments from the experts.
Writing low latency, scalable, flexible, and bug free code in a short amount of time. You can’t have everything so you always have to make some trade offs, and better engineers make smaller number of sacrifices.
I agree. If scalability is achieved, low latency can be addressed. I am just wondering that given “a short amount of time” and expecting a bug free code, what about testing? How testing is addressed in such a short time?
Well certainly what people need most is combination of low latencies and highly parallel system which runs on reasonably costing hardware…
But I think it’s quite obvious.
About C++, generally by using language without any other regards you won’t achieve anything by using any language, you need proper methodologies and skills and then suitable language. Typically for truly low latency applications it’s C and ASM combination, this discipline simply needs true developers not some Java/Python/whatever monkeys. So if you want to achieve something in this it’s not really about language but more about knowledge and skills towards makink applications fast and efficient. If I may suggest found the proper platform and libraries first, evaluate their “time waste factor” and then decide. If you simply asking what should you know for a job. Well more and more customers realizing they won’t found people they need amongst typical commercial programmers so they are looking towards real time specialists and people who have something to do with FPGA and usually with multiple language knowledge while knowledge how things works and ability to code really on the lowest levels (assembly plus profiling optimization). That is if you want to do more than Java/.NET/whatever clicking monkey.
Low latency can be achieved with not only good programming but an also knowlege of what asm the C translates into and how that can be optimized with minimal OS interupts. These ASM programmers that have turned into C programmers understand exactly what the C translates into and if not can be tweeked at the asm level.
That’s what I’ve meant. But being able to do actually something at that low level is sort of prerequisite and of course if I mean C and ASM I mean that the man know how things work because you couldn’t be really good using these languages in real world without actually knowing how it works by what I mean what C translates to and what ASM instructions are better to do this or that… Just as example … how many people don’t use even just MMX or SSE2 instructions because they don’t know and I’ve seen achieving tens of times faster results with them during few days. That is the difference which can be between coding monkey and real programmer… knowing how things work and being able to use it. to purchasing power parity as a means for valuing currencies? Unfortunately this way of evaluating currencies seems to be a lost art. Today FX markets seem to be driven by technical analysis and news flow. Our…
This is a short term approach; on a macro perspective, I believe that the USD is going to further appreciate ag. the Euro. The Euro is still in a mess due to the debt problem, and certainty will go through 2012 with the coming elections. The Euro base rate will drop further. There seems to be positive indicators on the US front while the next step for the US rates is up. While the focus is on Europe, the Euro will depreciate and further down. Once the Euro debt is resolved, the attention of the markets may move to the US debt which is gigantic. The US has however a trump card which energy – now North Dakota produces as much oil as Ecuador and there are deepwater exploration in the gulf of Mexico + the shale gas coming on line. The production of energy in the US will need to be monitored.
I now post my TRADING ALERTS
into my personal FACEBOOK ACCOUNT
. Don't worry as I don't post stupid cat videos or what I eat!