Weber & Partner was invited by a big automobile manufacture
in Germany to support the user acceptance testing for a Infinity based risk
management system. To use the Infinity system it was first necessary to
configure the whole system including counterparties, internal trading structures
and risk reports for the user. Most of the data was technical available
in the Infinity database but was not made available for the user. The testing
task was to explain and resolve differences in the pricing of a front-office
and the risk management system. To achieve this to validate the pricing
functionality of the two systems, to identify different interest and volatility
curve interpretations and to do the necessary reconfiguration, clarify and
document errors in the pricing functionality of the systems. Part of the
user acceptance testing was the validation of the Value-at-Risk calculation.
All the testing and essential parts of the system configuration was done
using Mathematica and our Excel and database links. The system was later
reviewed by the auditors and is now in daily production.
| top |
A major European bank wanted to check the consistency of
their internal market risk model against a full Monte Carlo method by applying
both methods to a realistic highly non-linear portfolio consisting of caps,
floors and swaptions. To achieve this Weber & Partner set up the Monte-Carlo-method,
implemented the banks internal market risk model outside of the usual work
flow and validated the valuation routines and market parameter settings.
The complete report was finished after fifteen working days. The whole analysis
was done using Mathematica and our database extensions. The valuation routines
for caps, floors and swaptions could have been easily written in Mathematica
but to further simplify the whole task validated valuation routines from
a third party were linked to Mathematica with MathLink. The results were
later used from the bank for discussion with the banking authorities.
| top |
Weber & Partner was asked to implement a zero curve
estimation routine and was provided with some notes on a proprietary curve
estimation procedure. Weber & Partner completed the specification and
implemented the estimation routine in a first version for daily use within
seven days. In the regular practical use of the estimation routine it became
obvious that some assumptions for the algorithm do not hold and extensions
were necessary. The estimation routine was implemented in Mathematica using
an object oriented framework. The framework eases the extension of the application
and also simplifies full or partial reimplementations of the application.
| top |
An international operating client was using an EXCEL based
Visual Basic application which had reached the technical limits of this
environment. Weber & Partner was asked to do a reimplementation in Mathematica
so that the application can grow with the business needs of the client and
new risk analysis can easily be implemented. The Mathematica version is
now also used in the London branch of the client.
| top |
Using Mathematica Weber & Partner
specified and prototyped an important conceptual extension to the current
counterparty limit system for a major european investment bank. With the
help of Mathematica it was possible to calculate the effect of the new method
for the current limit system on the real portfolio of the bank using all
accessable data and also to evaluate alternative approaches. To reach this
goal it was necessary to analyse and understand the data in 6 completely
different trading and backoffice systems, to match the data of these systems,
to model 20 different derivative products and to duplicate the current limit
approach outside of the existing system.
Weber & Partner was able to present calculated results for the new approach within 4 weeks. Dr. Weber said: "What the bank wanted from us was a very detailed business concept and a complete specification of the necessary interfaces to implement the new approach. The systems we had to deal with were mostly new to us. Whoever works in this industry knows that documentation on such systems is hardly available in this volatile environment. Data quality is mostly poor when you are working on a new report but you do not necessarily know where the data problems hide. What complicated matters in this project even further was the situation that we were only able to work on data which changed every day. So we had to apply a fully programmable approach to do our analysing.
Using Mathematica we were able to scan and synchronise the large amount of data in the different systems and put the analytics on top of that. What really helped to identify data problems was the ability to do calculations with the data and to use the graphical power of Mathematica to quickly see data problems. Being able to achieve and document our results in Mathematica simplified the task to a great degree. The ability to write the calculation very compact in a mathematical notation within Mathematica allowed it to easily exchange the retrieved results betweeen the two persons who worked on this project. I do not know another tool with which we would have been able to produce these results so fast."
| top |