Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
The point was (from post #1) I know several people writing their own, so I wanted to see if there was interest in sharing. Yes sourceforge is great, but these people that I know writing their own were doing so privately. Now the question is, are they willing to do so publicly with the promise and interest of help from others?
If they would, they would be on sourceforge IMHO ?
I find the idea great. Me as a J2EE and C++ programmer (but not really a good one) can't imagine that it's hard to write a blazing fast trader platform. After all we connect to an easy API of for example IB ...
for the rest it's receiving data and displaying it on screen, making some calculations for the indicators, and some way to allowed self written code interact with the program.....
I don't think that is always the case. A lot of times someone starts a project for their own needs and never goes as far as making it "public" like on sourceforge. So, this thread will either lure them out of the woodwork and make them consider working as a group, or the thread will die a slow and painful death
As some one who initially started out developing their own trading platform, I would caution on jumping on the bandwagon. The domain is so vast with so many different kind of requirements that it is not easy meet them all. To me the biggest challenge was the visualization part; I can write the core code but I have limited visualization expertise.
I think for pure discretionary trading with some trade management, NT7 is great. Yes it has its quirks but it is getting better over the last 6 months. It has all the cool features to do back-testing and validation and also allows you to export your results to do further analysis. You can not back-test on tick by tick data but if you use a single range bar, you get pretty close to it; it is possibly a better representation of what you would face in the real world as you long as you account for the slippage.
Where it lacks is performance and database integration.
There are multiple factors for performance all of which are inherent to the architecture and can not be changed soon.
-> .NET managed code
-> Windows Forms drawing library based on GDI+ which is very CPU intensive
-> Single threaded main loop charting loop.
However the question is where do these limitations matter? They matter when you are doing very fast trades; were the 100s of milliseconds really matter. They also matter in high volume environment where NT might just fall behind in processing speed.
The other big area where NT falls behind is the database integration. You can not request historic data; everything has to be on the charts. In some cases you can always workaround this by building longer and longer charts but clearly this is not the most elegant solution. There are other cases where there is no workaround possible. For example, if you want indicators which run both on 24/7 and RTH data in the same strategy you are out of luck since a chart (or a strategy) can only run one session.
---------------------------------------------------------------------------------------------------------
To summarize:
-> Do not reinvent the wheel: Any custom platform should not be a jack-of-all but focus on some specific need. It should be seen as a supplement to what is already out there, and not a replacement.
-> Database Integration: In the meantime the best investment would be to develop a high-speed system (a set of dll files) which allows you to:
(a) Maintain a database of historic data
(b) Provide APIs to easily access the data in your code without having to load it on the charts.
------------------------------------------------------------------------------------------------------------
The Big Picture Proposal:
I will continue to refine my trading plans and strategies using NT. The database integration is one obvious area which can be done within the NT framework which will make the system a lot more complete.
To address the performance need, when I am ready to go automatic, I will probably write my own C/C++ based program which will be light weight, bare-bones and multi-threaded. If done properly, we will also be able to reuse much of the code we develop for NT in the system, especially the indicators and strategies AND the database access mechanism.
To keep the system light-weight, I will stick to IB/DTN combo.
IB data-feed is great during high volatility environments because its sampled feed ensures a near real-time information. DTN is great 99% of the other times when you want your tick based analysis tools to work.
IB is one of the most cost effective and reliable brokers out there and DTN-IQFeed is one of the most reliable data-feeds out there
I have been threatening to roll my own for a long time. I have even PM-ed many of the people Mike is trying to get together to offer help. I either never got a response or got a "I'll let you know when its ready" type of response. Why is that?
For one, most people that can actually pull this off, are pretty sure they can do it themselves, and can't or don't want to work with others, and want to do it their way (the old NIH - not invented here syndrome). Also, if they can do it, this might be their edge.
Next, if you are going to spend countless hours trying to develop something, why give it away? If you are working as part of a group, this becomes less of an issue.
If you are giving it away, how do you protect it? Do you create a great structure, and let someone else use it to make a paid application?
Next, what are you trying to write? I know I can write the pieces to support what I need. Do I really want to write a bunch of functionality that others want/need? If you want to sell it, then the answer is likely yes.
Sorry, not trying to throw cold water on this effort, just trying to explain why it may not take off.
aviat72 said "Do not reinvent the wheel", could not agree more. If you are doing this to just fix/add to Ninja, then you are really wasting your time for the most part. It has to be something different/better if you really want it to take off.
If you want to build something great, the architecture is the key! You have to start with a great extensible architecture if you don't want to fall into the issues that Ninja, TradeStation, ESignal, etc. all currently have. Anyone starting now has the advantage that they know what everyone else has, and can create a good starting point.
In terms of language, I think you have to go with either C++ or C#. C++ will ultimately be faster and more portable, but don't under-estimate C#.
You need to decide on a platform. Windows is obvious, but what about MAC or Linux? If you go with windows, do you support legacy stuff or just Win7 WPF? Will you be 32/64-bit?
What data feeds and brokers will you support? Zenfire and IQFeed are pretty easy to interface to on windows. If you want to use another platform, you feed/broker needs to support it. Not many support MAC today :-(
Personally, I have been writing my own code as I know my requirements and did not expect to make it available to others (could change). This makes the coding much simpler since I don't need to code things that I don't use (it is architected for it for it). I am using C# and WPF, and would add in some C++ if I had to code some performance sensitive code (not anticipating that).
I would check out tradlink before starting from scratch. Would probably get a lot farther ahead pitching in to help them out than from rolling your own.
Good luck for tradelink, I haven't been able to run most of its pieces.
I agree with Aslan's post, and he's right when he says the archiecture is the key (but I think C++ will be a better choice than C#, and forget about Java ).
).