About Bloomberg
Bloomberg provides a platform for market data, news, analytics, and messaging to financial services firms and other organisations with an interest in markets. There are some interesting developments on the horizon but, for the time being, for investment banks and hedge funds it remains the standard platform for such.

Their API is public, and they provide implementations for C++, C, Python, Java, and .Net amongst others.

Interesting for D?
The financial services space might be an interesting domain for D because it's an economically important area where performance, safety and productivity all matter, and none of the existing solutions (mostly C, C++, Java, Python) are particularly satisfactory. At the same time, people in finance do not tend to be in the business of taking big technology risks with a product not yet widely seen as industry standard because the costs of messing up may be especially high (and messing up for creative reasons is much less acceptable than if you follow the herd).

Perhaps in the face of new emerging realities regarding data set sizes, memory speed, and single core processing power that creates opportunity for more entrepreneurial participants to use the best tool for the job. I certainly hope so.

People often want to see results quickly, so even if it's not really much work to do the port, there might be a big difference in the chance someone might consider trying (and then learning to appreciate) D.

Beyond a mature numpy/pandas equivalent, the other missing piece is an analytical library for derivative pricing and risk management. QuantLib is the main open source project - it is written in C++ but has D SWIG wrappers. (There is the Java Opengamma too, but I am not so familiar with it). SWIG is not great though, and I don't think the wrappers cover all the functionality. (There was a project to port QuantLib to D but I am not sure it ever got started, and it is now defunct).

It is super appealing to be able to rapidly prototype and explore ideas about largeish data sets and have the results be fast, knowing also that the code one has written can serve as the basis for something used in production (thanks to strong typing, elegant language design, etc). So I think there is a chance for D to take off within finance for some uses.

Project status
So for my own purposes - but I hope they might benefit others - I have ported the C headers and C examples to D. It's pre-alpha status - everything compiles, but I need to start up a Windows machine with Bloomberg to get it linking and get any bugs out of the port.

I thought of holding back posting till they were perfect, but decided to just post what I have in case anyone else should have an interest. It may be that the C++ headers create a more appealing interface, but I do not know the current status of C++ vs D interface as the dlang.org docs are stale. For the time being the examples are just an ugly direct translation, with lots of casts and not much attention paid to beauty and elegance. But it's something, and a start.



Reply via email to