Hey guys.

Over the past year a good half dozen or more of you have told me that you are writing your own trading platform. Each of you have your own reasons for writing your own platform, but I think the one thing in common with all of you is that you were unhappy with the platforms that already exist in the market.

A lot of you guys are really brilliant programmers, and excellent traders to boot. I've often wanted to try to find a way to get you together to work on a new trading platform in a joint manner. It seems like a lot of wasted resources to do it independently.

So, I finally just decided to start this thread. I am unsure what kind of traction to expect. Obviously, there are a lot of things to consider when developing a new platform. Everything from the language and architecture, to licensing and pricing, to features and support.

I would just like to say that I will offer up any resources I can to try to bring you guys together to create a joint team.

I think to get us started it might be beneficial for those of you have already started writing your own platform to step up and reply with at least the core fundamentals (language, vision, license) of your new platform, then we can step back and take an inventory and see where we stand.

Mike

It would be fantastic, I've always thought to the possibility of a custom platform.

Luke.

Zenfire has sample code for .NET & C++. It compiles and allows you to get ticks into your application. Not a bad starting point. I guess that's one of the first choices is whether to use a managed framework or roll your own.

I had offered to help NetTecture test & develop a GUI (if he would let me) for his proposed platform and of course Zondor was interested as he had met him on-line previously and thought it was in interesting idea. Nothing happened so far... (Maybe this will give him a push.)

Then beta NTs got better and better. By this point we pushed the development of the ladder & persistent indicators, and even got NT talking out our alerts on line and then complex conditions, etc. & stopped worrying as we felt we had a great discretionary platform and were anticipating all the great things with IQ &/or Kinetick.

Now, there has been the problem with historical data and am thinking for back-testing and strat development there needs to be something else. Until NT 8?

there is enough stuff in the air. the problem at the first time is to find a broker api and a good testing army.

what are the goals of "our" framework?

- indepent
- c api for real algorithmic trading
- fast without lags
- forex / future / options
...

to build this we must make a model on a paper/uml.

look at codeplex for example:
CodePlex - Project Directory

i believe sefstrat is a good adress.

Let me preface this by saying I'm an absolute crap programmer, but here are some thoughts and the logic process I've had on a "roll my own" platform over the past few months:
- What do I trade and do I want the same platform and analytics for stocks/futures/options?
- Do I need charting in the same platform?
- Do I need data warehousing and backtesting in the same platform?
- What type of execution latency is accepatble?
- Having decided the instrument(s) I want to trade, what APIs are available?

For now, I've come to the conclusion that Ninja meets my requirements, though it is not ideal. I try to keep my trade logic platform agnostic, and to that end I'll be a Matlab owner and user this year so that I'll have a common trade execution and analytics platform that I can use with some of the widely available commercial trading platforms if I move on from Ninja. Matlab can talk to IB, TT, PATS, RTS, CQG and others though some work is required. If I get good at Matlab and trading programming in general that is also a marketable skill, should the need arise.

The alternative (in my view) is a from-scratch effort to develop a ISO C++ or Java platform on Linux, though that will limit who you can connect to. Again also a (very!) marketable skill.

IMHO creating another .Net multi-asset, multi-broker trading platform is not worth the time. Also weighing in on that thought is my style of risk management and how that risk is calculated, monitored, and adjusted.

In the end a trading platform is just a tool you use to express your opinion about the market.

BTW an idea could be to build one from some good existing one that provides api, lately it seems to take interest T4 platform from CTS, very fast and seems accurate feed too.
Anyway one simple no frill platform, fast and realible, would be ideal.

Luke.


P.s. It seems there's no much interest in this topic.

I think the right people haven't seen the thread yet. I can't remember everyones names is part of the problem -- but a lot of people have told me over the last year they've got code already started, not just an idea but actual work done.

So hopefully they'll find this thread and we'll see if there is any interest in working on a group project. I guess it will depend on their objective.

Mike

SourceForge.net: Download and Develop Open Source Software for Free

Better join an existing active trading program project ?