Home | Topics | Software Engineering

Quantitative Libraries for Financial Predictive Analytics

Printer-friendly versionPDF version

Asymptotix' support for the language R (and S+ where I started and seems cool to mention alongside R these days, mystifyingly but probably the subject of another blog) is prima facie, just read our website. We have a particular pleasure in a good relationship with REvolution Analytics, with whom we have been beside since the start and who have driven the R standard with great success over the last couple of years. We have a strong partnership with TIBCO Spotfire and we understand S+. But what if R or S+ just doesn't fit, just is not compatible with the application architecture into which you have to develop a quantitatively analytic application. It happens sometimes, the application architecture is a context, a constrained environment in which the developer has to exist and develop. Talk to young developers these days and they will generally say why didn't the Solution Architect THINK before he handed me this prison cell of constraint, why can't I just do this in R?

iona motif asymptotixBut (in this scenario) the application architecture component of the overall Solution Design constrains the developer to use C++ or JAVA or even some feature, some extension of the mega-vendor BI products. In this scenario I am considering, it often seems that R doesn't fit, the standard languages for application development are set in stone, they are standards, there is a vendor platform for BI. The application architecture is a defined component of some overall Solution Architecture which includes the Data Architecture and the physical infrastructure.

We have entered the world of the "mashup" or "widget". Treating a quantitatively analytic application like a mashup or widget is in my view totally insane. (Asymptotix saw it go all too badly wrong at a very large client in London). But in this all too common scenario the "Enterprise Architect" at the top is chasing a nirvana which he can only envisage since he generally has no understanding of what it takes to make quantitative applications do the job they are required to do, the iterative aspect of quantitative analytics is completely missed. It is into this weakness that some customers are prey to vendors with all encompassing solutions. These almost never work.

Confusing BI and Predictive Analytics, i.e. Rear View Mirror Arithmetic with Extrapolative Prediction is a common error made by those with little or no quantitative background, all too common these days. Anyway! What do you do, as a developer if you find yourself in this position? Do you reinvent the wheel? Obviously not, but you need to know what kind of wheels are out there which you can use and there are a few.

Before we go on to a review of the quantitative libraries which you can plug almost any development environment into, "call" as we used to say, let's be clear on a methodological process for a quantitative developer: You should be aware as a quantitative developer that it is not the case that the business analyst should specify the detail of the quantitative analytic required. No! The business analyst does not need to specify to the last superscript the mathematics of the quantitative analytic. You should both agree on roughly what it is you are trying to do and then in general you will find references either on this website or on my analytic bridge blog as to what the correct implementation of the quantitative analytic you require needs to be, that is if you are undertaking quantitative risk management or financial predictive analytics (you wouldn't be reading this if you weren't). Don't specify it yourself and don't let anyone in your team be daft enough to think that they have eureka'd up a new and better way of quantitative analytics, its highly unlikely and a waste of your time which can only lead to problems. We have wide experience in this area and Asymptotix can help you too!

Just as we are laying out the standards for quantitative libraries, use the standard meth odologies too; from published academic and quasi academic papers. The quality of the work is getting better and better since "le Crise" (it has to); generally published work is becoming more and more applied, less theoretical and thus more accessible, indeed on my summer holidays I had the privilege to read an advanced copy of a soon to be published paper on stress testing which will be the blueprint to define the methodology, I think this year and next but more of that later.

We are going to provide a list below of quantitative libraries which Asymptotix have used or just perused, which we think are useful or interesting. As usual there are three or four front runners, whether commercial or open source which are the standards as libraries for quantitative analytic applications. Let us just say something of these for introduction. In general you will find that the package solution vendors who sell specific niche solutions of VaR calculation or Credit Risk quantification or Asset Pricing are using these libraries as integral components of their application. Of course some of these vendors do not! They may have Managing Directors who think nothing of waking at 2am just to tinker with a piece of C code to make a brute force algorithm run just that bit marginally faster but these cases are unusual, strange even!?! (You know who you are!).

nag asymptotixSo, about quantitative libraries for Financial Predictive Analytics then. The daddy of them all is probably NAG (Numerical Algorithms Group), their C library just went to version 9 this year, NAG supports Python, .NET, VB, C++, anything you like, you can even use R under Windows (the way we do), NAG really does it for you, below we have included an additional reference paper from NAG which Asymptotix used for a client recently about integrating NAG in a JAVA framework.

statpro asymptotixThe other one we really like is QUANTLIB, which is a free/open-source library for quantitative finance. A few companies have committed significant resources to the development of this library, notably StatPro, a leading international risk-management provider, where the QuantLib project was born. StatPro per-se do not get enough recognition in our view for their contribution to "Open Source" concepts in this regard and its not a drum they bang at all (very cool); StatPro focus on Asset Management vertically but their solutions are at least equally applicable in any complex pricing challenge e.g. liquidity risk, asset pricing (credit risk/structured products) should you wish to gain marginal advantage by automating your IP.

fincad asymptotixFINCAD is another major player, they used to be very focused on supplying very clever libraries for very clever EXCEL users but they have developed a lot since then, as they say themselves perfectly; "FINCAD provides a range of financial analytics pricing and risk management solutions to satisfy a variety of end-user needs. The right solution for you will depend on a number of factors." You can make FINCAD do the heavy lifting for SAP or similar Ledger packages, it can be done.

quantifi asymptotixFinally in this overview another very clever company which started recently as a supplier of libraries but has now developed into a more broad based product company, that is QUANTIFI; who are really focused on the credit derivatives and structured products vertical niche but as they say their tools support state of the art pricing algorithms and what you use on the buy side you should be clearly using on the sell side, is that not one of the lessons of this crisis? QUANTIFI is really clever stuff and there is no reason why you couldn't integrate the appropriate tool into a Financial Predictive Analytic of any type.

Enough! Of my favorite quantitative libraries, we hope you have found this useful, given the Asymptotix partnership with Intel can we just say a word about the Intel Summary Statistics Library, which we find very interesting, quite where one might use it baffles us but we are certain that it will be functionally excellent as is everything Intel does. The recent acquisition by Rogue Wave of the Visual Numerics operation, who own the IMSL library has entailed that we lost track of what was happening there but certainly these products are worth your attention under the circumstances we have posited above.

Here below is Asymptotix table of various available quantitative libraries for Financial Predictive Analytics.


URL References















Intel® Summary Statistics Library




Visual Numerics


[Rogue Wave]


NMath Stats




Extreme Optimization


Boost Accumulators





Open Math


JAVA scientific library










For more information

Contactasymptotix JAM sept 10

John Morrison
CIO, Asymptotix
Tel: +32 475 63 2991



Short URL
Asymptotix on Twitter

Are the key legislative pillars such as Basel II & III, UCITS IV and Solvency II forcing you to re-examine how you identify, measure and manage risk and capital?

Asymptotix work closely with our partners to help clients develop a more proactive, systematic and integrated approach to governance and risk management to deliver proper value.

Asymptotix can offer the support you need to deliver on time. Read more...

Is the goal of your website to sell services or products, educate, or collect data?

A positive customer experience is vital to conversion, no matter what your conversion goals may be. Our designers and developers will create a positive experience to maximize your conversions and deliver the optimal return on your investment. We strive to find the perfect balance between the web site’s design and functionality.

Asymptotix implements interactive solutions for European companies. From corporate websites to social communities, our clients will tell you an investment in building a scalable online experience will deliver long-term tangible benefits.

Based in Luxembourg we can help you all over Europe. Our multi-lingual team can work with projects and speak your language! Read more...