DataSynapse on Amazon EC2 February 12, 2009
Posted by James Webster in : finance, virtualization, development , add a commentDataSynapse have just announced a service and beta program for running their grid computing platform on the Amazon EC2 service (press release here). It is only open to existing DataSynapse customers but those who do sign up will be able to run the DataSynapse engine on as many EC2 instances without incurring licensing costs during the beta period, just the CPU and transfer fees from Amazon.
In my experience DataSynapse is the grid vendor with the greatest market share in the financial services sector. I sometimes wonder why however; deployment of gridlibs can sometimes be flakey and the administration interface leaves a little to be desired (although to be fair I haven’t seen the very latest GUI). Still its good to see the major grid computing vendor supporting Amazon EC2 as a first-class host for their software. The open-source Java grid platform GridGain are allegedly working on deep EC2 integration and it will be interesting to see what they come up with.
A few other interesting and unrelated tid-bits:
- Wolfram Research announced a Home Edition of Mathematica but only available in the US and Canada for some inexplicable reason at present.
- Announced late last year was VMware Studio, an interesting looking tool for building virtual appliances, similar I guess to CohesiveFT’s Elastic Server On-Demand. Given that command line tools are available it would be interesting to think how you might integrate this into a continuous integration build perhaps, Organic Element and Will DeHaan have some similar ideas.
CDS = OSS? February 1, 2009
Posted by James Webster in : finance, development , add a commentJ.P. Morgan has taken the step to donate their CDS pricing code to ISDA, the International Swaps and Derivatives Association (via Finextra, see the ISDA press release).
The motivation around this move is to demonstrate the credit market players’ greater willingness to ensure transparency around the pricing of credit default swaps. Many derivatives houses consider their in-house analytics to be the ’secret sauce’ that makes their pricing and risk management better than the rest of the Street.
This news arrives as the US Congress consider a proposal to ban ‘naked’ CDS trading; that is it would be illegal to buy CDS protection unless you actually hold the bond for which you want protection against the issuer’s default (i.e. the loss of your principal). If this ban did go ahead then CDS trading volumes (and profits) would be dramatically reduced, assuming that legal jurisdictions outside the US follow suit.
I’ll stay out of the economic, social and moral debates about whether eliminating naked CDS trading is a good idea or not. Here are some of the questions I have surrounding the release of this code as open-source software:
- What is the timeframe for making it available? No detail appears on the ISDA press release but if the motives are primarily political then it ought to happen sooner rather than later?
- What sort of open source license will the code be made available under? What terms and conditions surrounding its use (and potentially profiting from it) will come attached?
- Will a Sourceforge or Github-style repository be set up around the codebase or will it simply be a case of downloading a zip of C++ code from an ISDA server?
- Some quantitative analytics libraries are tied to the market data and risk systems that surround them so it will be interesting to see how clean the interfaces are.
- Putting the code to work on an Amazon EC2 grid to stress test a CDS portfolio might be an interesting exercise.
Any discussion of quantitative financial analytics and open-source would be remiss in not mentioning Quantlib; the original (and only?) open-source quant finance library.
Quants in the cloud November 25, 2008
Posted by James Webster in : finance, development , add a commentWithin a week of each other the vendors of industry standard mathematical software packages MATLAB and Mathematica have demonstrated their software running on Amazon EC2. MATLAB has greater dominance in the field of financial mathematics but as a layman in the frequently mind-boggling world of quant finance Mathematica’s approach seems more appealing and visual. Might the embattled banks start to use these packages to offload intensive batch risk calculations to 3rd party compute grids on demand? Or maybe just buy a couple of these?
Of course the future shape of the quant and quantitative finance is under much debate at the moment as the leaders of the world seek to tame the global financial crisis which many blame on quant finance in the first place; Nassim Nicholas Taleb (author of Fooled by Randomness & The Black Swan: The Impact of the Highly Improbable) says Many of You Will be Sued. As a technologist within the field (and assuming things don’t collapse completely!) my bet is that the skills to cultivate will be around exchange connectivity and standards such as FIX and FpML… as the call for greater transparency of the OTC derivatives market becomes ever louder some products might be regulated out of existence and a large portion of the rest will become exchange traded. Why bother with mark-to-model when you can simply mark-to-market?
The Economist has two great level-headed articles about the much maligned credit default swap’s role in the current crisis and how they can still have a valuable if restrained role in the markets moving forward:
Meanwhile major players and central counterparties in the CDS market are already making progress in netting out CDS exposures in an attempt to reduce that much quoted (but misunderstood) outstanding notional CDS value of $62.2 trillion USD.