Open source in quant finance

[Andy Nguyen]

A top down look at how the open source trend is gaining momentum among big players.

In the financial world, the biggest news was surely Oracle's $7.4 billion bid to acquire Sun Microsystems, a move that could make Oracle the world's largest open-source technology provider and hand over control of projects that include Open Solaris, Open Office and MySQL. Elsewhere, Bloomberg announced a data distribution system, called, simply, The Platform, which will be licensed as an open-source software initiative, and built on servers running on Linux. Some form of open-source software is now almost ubiquitous among capital markets firms, according to many industry observers.

http://www.watersonline.com/public/showPage.html?page=872979&goback=.nvr_90917_1

 

[quantstart]

I'm still worried about the future of MySQL. In time, for web projects (which I work on quite a lot), it may be prudent to switch to PostgreSQL.

 

[DominiConnor]

As someone who helped the guy who was writing the Waters article I am upbeat about OSS, but see the Oracle acquisition of Sun as a backward step for OSS and for users of s/w in general.

If you've ever done business with Oracle, you will know that they didn't get to this position by helping people, and I expect that they see the damage their acquisition has done to MySQL as a good thing.

The platform may be a good thing, when I actually get some details about it, I might comment further, but a full commitment to an open platform is such a radical change for BB that I need to have personally kicked the wheels before I believe it will drive away.

However any industrial strength platform needs people to look after it, and with usage costs are typically the main factor in how much you end up paying for a system. Often this means there is very little difference, so you must choose on grounds of functionality, reliability and flexibility, and a term often missing from the discussion, whether your team has anyone who can drive it.

 

[Daniel Duffy]

I have a general question: under which circumstances would we forego open/non-proprietary software and choose a proprietary software solution?

 

[quantstart]

One issue with choosing proprietry software over open source software is that there are generally more people around to employ who are versed in proprietry tech. The prevalence of Visual Studio means that there are far more C# devs around than Python devs, even though the languages (approximately) cover the same ground.

 

[DominiConnor]

Daniel, I assume you mean a rational basis for that move, as opposed to politics ?
Some OSS systems are dropped after mergers or because of the desire to have "one unified system"
Some proprietary systems look good on the resume, and I've seen that be a choice in systems.

Global support is one set of reasons. You may have N people in your office that can drive the beast, but in remote sites there might not be any obvious resource, so a vendor can be the way forward.

Ease of use.
Too many OSS systems are as user friendly as a cornered rat. I have sworn a bloody oath that the next OSS oaf who says "all you have to do is read the source code" will die at my hand.

For a personal project I've been using Gimp, and never ever has the documentation been been at the same release level as the code. In fact I skipped one release because I couldn't get the help to work at all. It has no bloody idea how to manage it's windows, (capital W or small w), and navigating between pains is like using a student Windows project from the 1980s.
It's mean time between failure is about three hours, with about 5% of the images I create, it refuses to save them, and I have to do awkward things with the MS tools to save my work.
If there was any alternative I'd buy the proprietary solution.

More relevant is QuantLib, which still isn't documented in any real sense.


Some OSS systems seem purposefully set up to be hard to use on Windows, which if it's the platform I need for a task, can be enough to choose something else.

Vertical market functionality.
OSS tends to be stuff that people want to write, as opposed to what others want. A lot of s/w is boring bollock like accounting, international languages, etc.
MS SQL has truly vast facilities for sucking data in from many different sources, for instance.

Another reason as above is that the person who knows how to drive it is not there any more. For widely used bits of s/w that doesn't matter, but can hurt if it's a rare skill.

Also some prop s/w goes faster and crashes less.

 

[Daniel Duffy]

Daniel, I assume you mean a rational basis for that move, as opposed to politics ?

Dominic,
Yes, the technical issues which impact the 'half-life' of a product. For example, if one writes a solver in a proprietary language now (lot of effort) knowing that more standard solutions are not too far away (let's say one year), what is the optimal approach?
For example, I consider C++ to be 'open'.

 

[quantstart]

I agree that FOSS software can sometimes have horrendous documentation, but there are projects which do possess incredible docs. These are predominently web projects though where the software and the communication processes between developers are tightly coupled. In finance, the culture is not one of collaboration between organisations.

In terms of adoption, a flashy website can add trust for the enterprise sector. Then again, look at Apache's: Welcome! - The Apache HTTP Server Project