The UnRisk Agenda 2011

The financial crisis has raised many questions. In the frame of principle discussions, like do market imperfections arise out of imperfect or asymmetric information, or alike, many of them have a deeply technical nature: Is quant finance about predictive modeling, or is it explanation by inverting real market dynamics? Shall we model more or less, use simpler or even more complex models? Is VaR misleading? Can we assess the quality of a model, without looking into its implementation?  How does the coverage and quality of market data influence the quality of deal decisions?  How will new ultra-fast processors and their mobile-focused versions influence quant finance? 


It is our strong believe that - even knowing that individual behavior will always differ from that "predicted" by models - widely accessible computational knowledge is indispensable for making good trading decisions (by humans or by computers).

Our contributions and what projects our core team will take on in 2011.

Andreas Binder
MAKE THE BOX EVEN WHITER - As head of the UnRisk makers I have pushed to package know-how and launched the UnRisk Academy.
One of my major tasks in 2011 will be to write (together with Michael Aichinger) a book on advanced numerical methods for quantitative finance.
We observed that the numerical algorithms for valuation and risk analysis of financial instruments, which are actually used in financial institutions, still contain quite poor methods like trees (one might call us lumberjacks). Following the strategy of our Academy, we proposed to write a sound overview on different numerical schemes relevant for finance, their pros and cons and equip the book with a wide range of interactive examples available for the readers enabling them to see risk and traps in the deal dynamics.
It will be published by Wiley and appear in early 2012.
From our point of view, the main target group will be junior quants considering a career in finance who want to get a solid introduction to different methods, their advantages and limitations and a good range of examples from the practice.

Herbert Exner
PREPARING FOR THE SaaS/WEB REVOLUTION - I think about UnRisk's strategic initiatives in 2011, they are all about highest performance computing made accessible by Web front-ends or apps. 
We are at the point that our UnRisk FACTORY has shown in many bank installations that high performance computing and transaction software can go where its users go. 
How can we offer the UnRisk solutions and know-how-to to quant developers? 
To realize that vision, UnRisk does work on 3 fronts.
First, migrate the already ultra-fast algorithms to new processors and do things that cannot be done without.
Second, extend the UnRisk/Mathematica front-end for literal programming to a vast variety of new usages.
Third, provide web-enabling tools that allow for the deployment of even the largest scale UnRisk application as interactive web site

Michael Schwaiger
EMPOWER QUANT DEVELOPERS - As UnRisk product manager I still program in UnRisk/Mathematica and C++.
Along with the crisis VaR has been discredited because it too often has been taken as one value for complex decisions.
We decided to extend our already comprehensive VaR solution and make it available for quant developers. Release date of the VaR module based on UnRisk-Q: Q2 2011. 
It will calculate portfolio VaRs,  individual VaRs, the VaR contribution of each instrument and VaRs derived by screening across risk factors. Perfect for developing all kind of high level risk analytic tasks. 
It  will cover all financial instruments which may be priced by UnRisk-Q. The calculation of the VaR may be performed under all implemented pricing models. The package will calculate VaRs across a vast variety of underlying risk factors including interest rate curves, equity prices, fx rates, credit spreads,  Black76 cap volatilities, Black76 swaption volatilities, equity yields (continuous dividends), inflation swap curves, inflation indices, equity volatilities,  fx volatilities, interest rate / interest rate correlations and interest rate / fx rate correlations. The package will offer Historical VaR, Parametric VaR and Monte Carlo VaR methodologies. 
For significant speed-up it will apply principal component analysis, where possible and offer built in parallelism, or parallel programming. 


Not surprisingly,  we will update our UnRisk FACTORY (to version 3.0 a big new release in summer 2011)

Michael Aichinger
MAKE THE BOX FAST AS LIGHTNING - 2011 will be the year where GPU computing looses nerd status and transforms to a  valuable tool to speed up computational tasks in financial environments - why?
The latest versions of GPU cards allow to perform computations in double precision - indispensable in the field of numerics. We at UnRisk, early adopters of this future technology, will take GPU enhanced products of the second generation to financial markets early 2011. 
In a first step we will concentrate on equity linked products and complex models. It is known that straight forward parallelization (Monte Carlo) of the pricing routines is of good nature for GPU speed-up. Model calibration is more challenging, but it is the core for in-time decision making, especially when promoting the very new concept of frequent recalibration. 
Beyond this our users will have the flexibility to use the power of the GPU on different levels of parallelization granularity. If you do, say, across-model scenario runs with different timing of each valuation step an intelligent combination of fine and coarse grain parallelization becomes vital.
We are passionate about generic technologies, but when a speed-up of 10,000 is required you need to know the wires.
Consequently, we cooperate with the leading hardware makers ensuring that our customers get access to the latest HPC platforms enabling much-faster-time-to-insight.


Sascha Kratky
REMOVE THE IRRATIONAL FEAR OF NEW ARCHITECTURES - My responsibility as senior software engineer is to design UnRisk for technology platforms with future.
2010 saw the beginning of the end of the Wintel area. With Microsoft adding support for ARM processors in Windows 8  and many hardware manufacturers moving away from Windows and choosing Android as an operating system platform for mobile computing devices the computing platform landscape will change considerably in the coming decade. One of our major tasks in 2011 with be the overhauling of the UnRisk engines code base to make it compatible with target operating systems other than Windows and to support modern processor architectures like ARM or Atom. 
The UnRisk engine code base consists now of around 700 000 lines of C++ code and an additional 200 000 lines of Mathematica code. 
The first outcome of this refactoring effort will be a Linux version of the UnRisk engines which will work seamlessly with all Linux platforms supported by Mathematica 8.
And clearly, we will continue our spade-work providing tools for integrating UnRisk and web services (webUnRisk).