I glossed over this important LinkedIn article posted by a professional who does high-frequency trading development for their career. They have been kind enough to share all the lessons learned in developing low latency trading systems.
As for me, I can completely agree with every single word in this article so it is quite imperative you read it if you plan to go into this career. Also, I am planning to implement all the latest learnings in trading strategy development for foreign exchange, futures, options, and equity. This will all be built around using:
Combined back and NoSQL database with Redis and MongoDB
Capability of Python scripts to download free market data which is pushed into my database clusters
Use WordPress plug-ins to customize front end interfaces with free plug-ins as well as commercial charting plug-ins.
This will lay down the foundation to implement and analyze global macro views with major asset classes in web deployable interfaces. All short-term indicators will be based on fundamental with limited technical analysis confirmation techniques. I’m also planning to include automatic order management system using Interactive Brokers as my preferred broker of choice.
Remember this Tuesday, I will be revealing how to use US Fed data to your advantage when you want to implement your own arbitrage or pair trading system. Yes source code is made available as well!
We are now moving into our second week of our arbitrage/pair trading strategy with our second live lesson starting next Tuesday May 17 at 8 PM Eastern standard time.
This is part of my “Algo Trading Course Series in Python.”
A new member of mine asked about an overview how my Quant ELITE membership works. So…my little brain came up with a neato idea. Why not give an overall tour of this thing?
This was the genesis of the recent page I posted about my upcoming ‘Python Algo Trading course series’. Get more info here: https://quantlabs.net/blog/python-algo-course-series-for-your-indie-automated-trading-business/
This will all start MARCH 15 in exactly 30 days from today! Details will be posted on my Meetup groups.
So let’s get this on but there is still lots of current stuff this membership has which can be accessed right now.
As a result, I will be doing a LIVE tour of this on:
TUESDAY, Feb 16 at 7 PM EST
More details can be at the above link about this upcoming algo Python trading course series! This will be the most comprehensive endeavour I have done in the last 5+ years.
Ok, so where can you get more info on this current Quant ELITE membership? Get it here at the following links: http://quantlabs.net/mkt/quant-elite/
You get access to everything listed here plus more: http://quantlabs.net/academy/courses-2/
Let’s do an entire Q&A for everyone on this.
1. Please join my meeting, February 16, 2016 at 7:00 PM Eastern Standard Time. https://global.gotomeeting.com/join/984345437
2. Use your microphone and speakers (VoIP) – a headset is recommended. Or, call in using your telephone.
Dial +1 (647) 497-9353
Access Code: 984-345-437
Audio PIN: Shown after joining the meeting
So many people want to build these high frequency trading systems but personally I think they’re learning from the wrong people. I am very lucky to have people who have contributed
to my own knowledge to know the right way and a wrong way to develop these trading systems. If you followed me for a while, you know who exactly I’m referring to. If you don’t know
who I’m referring to, let’s just say they’re making more then your family for the next 20 generations will ever make. Fortunately he was able to give me a solid advice and probably the best
advice I’ve ever heard regarding all these programming languages.
I also find it quite preposterous how these old school trading gurus claim they are pumping out well to do traders. To me, that is utter garbage, if you’re confident with your trading audience
or community you would easily allow third-party analysts to to measure their income. It is to showcase who is making bank! if they are, then you would showcase it. Wouldn’t you? Well guess
what? I do that with a service called Quantcast. At my last check, 40% of my trading audience makes over $150,000. Let me stress this: it is verified by Quantcast!
Expert Insider thoughts on low latency programming trading languages like C C++ Java vs Python
A set of tips came in from another expert through my newsletter:
Let me try to categorize things from Low latency point of view.
such systems like HFT application has two parts of it, one part which is latency sensitive and one part which is non latency sensitive.
Component which are latency sensitive are developed majorly in C and C++.
and non latency sensitive component can be developed in any language of choice in which developer feel comfort and have expertise to make the development much faster.
–> Yes this is definitely a big standard, I prefer C as it even simpler and better hook development into a low latency operating system like a custom Linux kernel
The major reason why industry and senior developers in profession prefer to use other language is because Mastering C and C++ is too difficult hence getting an expertise developer in C and C++ is difficult. One things I would like to point if developer has not written a latency sensitive code in C and C++ it will not be latency sensitive application.
People prefer Java because it is too easy to learn and things are faster to develop and maintain as compared to the core part of C and C++. Java is pure Object Oriented language. While Object oriented is just one part of C++.
–> The suggested open Trade Manager Java project could be a decent starting point here. The .NET stuff I have worked is definitely another great starting point
Which is normally low latency developer avoid and use the template generic programming which is much harder to learn.
–> Totally agree, you want your systems to be generic based only with no state as that takes up less processing
Python is never been a main stream language for Low latency side of the system. It will be a poor choice. But yes for the part of the system which is non low latency python might be a smarter choice it will give you much faster development cycle and easily maintainable code base.
–> I really agree here so most should stick with a high performance language like C to reduce jitterness at the OS level, Java does this with the garbage collector. I have provided many video talks from experts on this including from the biggest banks on the planet.
Start categorize your systems.
Also avoid mixing and using language adapters and try to find other alternative solutions.
Good luck sukah, it will take 7 years of C or C++ low latency very technical to get decent at FPGA in an HFT changing space
The video showcased a lot. Wow this is the second one which really highlight the path I am wanting go down.
This sounds like a highly technical architect with FPGA solutions where he sounds like he worked with some banks like HSBC since they were mentioned. Speed is critical to stay in business as there was only 2% average returns in 2014 for Hedge Funds which is nothing! He also mentioned some huge billion dollars per day at risk for only $2k risk vs $70k. Again nothing but that is the era we live in. Also, many funds are going bankrupt because of this.
In terms of technical, he mentioned C is the only way to transcode the assembly as well as the custom VHDL put onto the FPGA. Very interesting how these HFT shops are transitioning out of software solutions. In the end, I am trying to figure out if Matlab Simulink is an efficient way to translate any trading idea into VHDL via the Simulink Coder. I have sent out questions to clarify but something tells me that you will need to hand code. Ugh. I am still looking into it. The question is: Is it worth pursuing these systems via software with PFGA out there? For now, I think it is but don’t spend too much time with it if my trading interest direction radically changes in coming months. Also, it seems that LUA is used as well which is just a reminder why it is the embedded scripting language for Redis which is done in C. It confirms my thinking about this whole process!
This video is kind of funny as the dude of QuantStart asked the first question. I think I saw the look in his eyes when Python was not really part of the equation. I got confirmation from others it ain’t with the same as R. Sorry. Just watch it to see the radical direction this HFT is taking. Oh yeh, it was mentioned that the average length of time in learning this tech stuff is 7 years. I am just fortunate to have some highly technical and other in high places within the industry to help me get there faster!!
As said, I need to start ramping my EFFECTIVE learning in futures and options. That seems to be where the really profit payoffs lie despite all this risk people are not willing to take on.