As portfolios get larger and more complex, the processing of pricing and risking of instruments moves away from the desktop to the server room. Quantitative models written for excel and other desktop applications are being deployed to grids or as services in a distributed architecture. While the mathematics may remain the same, there are distinct challenges in the provision of service oriented, efficient, accurate, robust versions of local analytics libraries. In particular, the nature of these libraries is that they frequently change and as a result, deployment and testing frameworks are key to delivery of an industrialised, flexible analytics service.

One of the advantages of the distribution of libraries as services is that it allows the massively parallel processing of calibration, validation and back testing scripts. The real challenge comes in the provision of data and the efficient distribution and aggregation of results.

We support the construction of large scale distributed analytics environments through our grid and data caching services. Our understanding of the business of trading, quantitative analytics and parallel computational mathematics allows us to add value by providing processes and frameworks that make delivery quicker and processing faster.

For instance, we have delivered recent projects including:

• code optimisation through exploration of compiler technologies and parallelisation schemes for a European bank
• migration to 64 bit operating system for a tier two securities company
• reviewing the analytics framework and recommending improvements for an energy company
• building on an analytics framework for integrating models into Murex via the Flex framework for a Scandinavian bank
• quantitative development services for a European bank
• consulting on the provision of an enterprise analytics architecture for a tier two global bank