The backbone of all UnRisk products are the UnRisk Engines, all programmed in C++.
Covers the calibration of models and the valuation of financial instruments. Highly advanced numerical schemes are used to provide fast and robust solutions. A critical step in the valuation process is the parameter identification for the models used for the simulation of the underlying risk factors, as the inverse problem may be ill-posed. Our algorithms targeting and solving these inverse problems have been developed together with some of the leading mathematicians in this fields. For the solution of the forward problem (valuation) UnRisk chooses among different methods including the solution of PDEs with Finite Difference/Finite Element schemes, the solution of SDEs with (Quasi-) Monte-Carlo methods as well as direct integration methods like Adaptive Integration and Fourier based techniques. The UnRisk Valuation Engine covers many models and a vast variety of deal types.
Covers parametric, historic and Monte Carlo Value at Risk calculations for a variety of risk factors. For a deeper understanding of the sources of risk, the VaR Engine allows the calculation of incremental, marginal and contribution VaR from single instruments up to the portfolio level. Besides the VaR other important key ratios, like the expected shortfall, are calculated. Routines for backtesting the results are included. Also scenario analysis and stress testing are part of the engine.
Utilizing our most advanced numerical schemes, UnRisk xVA engine is capable of simulating expsoures and calculating derived quantities like CVA and DVA. All risk factors of a portfolio are simulated together to create market scenarios - all instruments of a portfolio are evaluated under these scenarios to obtain the exposures. In the aggregation process, netting can be applied and the xVA management ratios are calculated taken into account probabilities of default for the counterparty and the institution.
Driven by the regulatory requirements our customers face UnRisk has built a PRIIP engine. This engine covers all tasks from simulating the market data, building the scenarios, calculation of relevant cash flows, valuation of the instruments at different points in time to the calculation of all necessary key ratios to produce the PRIIPS-KIDs.
UnRisk Data Services
Depending on the UnRisk solution different strategies working with data exist. For UnRisk Quant and UnRisk Pricing Engine no database is included in the product but a variety of supported interfaces allow the users to plug in their own data sources.
- Excel workbooks
- CSV files
- User databases via database connectors
- HDF5 files
- XML files
UnRisk Bank and UnRisk Capital Manager come with their own database and their own adaptors. These adaptors are repsonsible for the transformation of the external data into the internal data model. For some of the most widely used data providers UnRisk offers preimplemented solutions (for example Bloomberg).
The UnRisk FACTORY keeps all data in a database which can be accessed directly via SQL views or indirectly by means of a SOAP based web service.
The following figure given an overwiew about the UnRisk FACTORY architecture:
Once the UnRisk FACTORY is deployed, it consists of the following components:
- a database which stores all the data related to financial risk management (instrument data, market data, computed risk measures, accounting and meta data)
- a web server which implements the web front end which can be accessed by users through a web browser
- an adapter process which allows for importing data from foreign systems
- a service process which controls the computational grid
- a computational grid which performs risk measure computations in parallel using a MapReduce based parallelization strategy
- a web-service which allows for accessing data stored in the UnRisk FACTORY database programmatically
- a monitor application to administer all UnRisk FACTORY related processes