How about new, minimalistic trading platform... Open source of course

all agreed!

Unfortunately I can't open up my codebase due to various issues... long story...

Quote from Steven.Davis:

I have had only minor dealings with the open-source community, but I think that there are several factors I see which separate successful (actually made something useful) from unsuccessful (little return on investment) open source projects.

1. Business Plan. An open source project needs a business plan just like anything else into which one wants to put time and money. Not wanting to pay programmers is not a business plan. Saying that you are going to build a rudimentary e-commerce engine, and the contributors will be consultants who make their money making websites by customizing the engine. The product would then grow by these consultants checking-in the things that their customers are paying for.

2. Target Functionality. Everyone trades differently. Even clients whose requirements are nearly identical at the outset become rapidly incompatible. Software can either be very expensive or very inexpensive depending upon whether there is a common problem which can be simultaneously solved. The Financial Software arena is thick with products. Why will people really want your version?

3. All encompassing. By wanting a complete package rather than a few reusable pieces, you are limited your developer appeal to those that want very much the same package. Simultaneously your massively increase the effort and the coordination. For example, could you get want you need by having an R/Matlab connector for TradeLink/OpenQuant? Many developers would want to include that in their diverse project, and efforts to enhance the R interface can be only loosely tied to efforts to enhance Matlab interface. It could for the nucleus of a group.

4. Fragmentation. On SourceForge.net alone, there are 690 projects based on the keyword "Trading". Of these only 16 have had 100 downloads in the last week. Of those half, have nothing to do with the Financial Markets. That means that only 2% of the open source project has user interest at that threshold. As far as developer interest:
Code: 
    Dev Status
    * Inactive          12 = abandoned
    * Mature             7 = successful
    * Production/Stable 66 = active
    * Beta              89 = trying
    * Alpha             75 = nothing
    * Pre-Alpha        119 = really nothing
    * Planning         167 = no one cared

5. Culture. Look at the way TradeLink is. It has a bunch of modules closely based off of broker APIs. This is the core value proposition. How much work? There there is TA-Lib, which is an adopted open-source project which slimmed-down. Most open-source things that I have seen with some pull have been extensions/splinters of some successful open-source project. The programmers need to be motivated. Usually because, they think the technology and idea are intrinsically really cool. Not because some users want it for free.

If you really want to start-over, you could look for someone else's commercial venture. Buy the defunct assets, turn it open-source, and at least there would be a core competence upon which to grow an online community.

Best of Luck.
More...
 
So, what did you decide ?
Are you going to start working on this project ? If so, I think it's better to release it early even if incomplete so that you recruit help as early as possible.
 
Still undecded. I cant open my commercial software because of various reasons. The options are:

a) hire one developer to work on it
b) buy someone elses code open it up and develop with comunity
c) buy code + hire

anyone willing to offer his code? Price will depend on maturity and robustness... Of course, it has to meet those citeria from the begining of the thread.

Will post decision shortly

Quote from softdown:

So, what did you decide ?
Are you going to start working on this project ? If so, I think it's better to release it early even if incomplete so that you recruit help as early as possible.
More...
 
Puzzled by the anti-GUI. A properly implemented GUI doesn't load the core trading platform. Neither does backtesting or forward testing code as it sits there - not affecting live trading code - unless called.

As various people want various things, you need to either accommodate a lot of what people want or have the interfaces for people to implement their own add-ons.

I played with jsystemtrader a lot, but there were a number of errors and some odd structure, making it difficult to modify to work correctly. Managed to implement a variety of custom technical indicators. Major flaw was no trade until the current candle was over. A lot can happen in part of a minute. Need something that updates indicators for TA on tick data, or at least the filtered tick data from IB, and allows trades based on that update.

jbooktrader is an example of something that doesn't offer enough. By just looking at the book, they leave the TA price people behind. Their guide doesn't match the distribution either. So do I modify that to provide the TA stuff I trade with?

I like the idea of being able to forward test the actual algo code against the code that will actually be trading the algo live. Otherwise you're testing: the testing/opto code, the testing algo, the live algo and the live algo driver. If they're already integrated...

Having a core trade platform sounds great, but why not have a GUI sitting on the side for adjusting things like position sizes on the fly, and provide simple things like: enter long/short, increase position, decease, reverse, exit. Trivial to add, and attracts more user base. Particularly if one can java code in their own buttons.
 
You must log in or register to reply here.
Back
Top