QuartetFS: Real-time VaR
Posted: 3 June 2014
Our client needs regarding VaR
Our client is a major regional bank active in capital markets, asset management, retail banking and insurance. In line with the Basel II requirements, the bank has to comply with the Value at Risk (VaR) framework. Its vision for VaR however goes beyond the calculation of a single number or a tick-box exercise to comply with regulatory requirements. Instead, the bank has developed several functional enhancements, working alongside Quartet FS, to bring VaR into the front office and facilitate true collaboration between traders and risk managers. This unique approach to VaR has helped to transform the business and its management of risk.
Helping our client required overcoming extremely challenging technical aspects. Very large amounts of information had to be stored while maintaining sub-second query response times. Indeed the non-linearity of VaR calculations imposes to store all the P&L simulations for VaR, and not only VaR as measured for each trade.
Our client had already implemented a custom-made system based upon a market leading OLAP analysis server solution. The classic cube was built during night batches with end of day risk data. However, our client found that it wasn’t practical due to the very nature of VaR calculations which require non-linear aggregation of vectors. Indeed, unlike for P&L positions which simply add up, the VaR of a trade does not add up to that of another trade or portfolio of trades. Classic OLAP used for VaR involves developing highly complex queries that are extremely hard to maintain. In addition, execution of the queries can be very slow which is frustrating for users.
With such slow response times, it was impossible to envisage a cube that could be updated with real-time trades since the system would block all user interaction while changes were being applied. Such functionalities were required by the business, but could not be achieved with the bank’s existing solution.
As we engaged in discussions with our client, we realised that the bank had an even more detailed and complex set of requirements; none of which seemed feasible with the incumbent solution.
These requirements included:
- The need for VaR data to be sourced from two or more separate trading systems
- The requirement to update the target systems intraday to refl ect latest trades and provide risk exposures in real-time
- The ability for users to confi gure temporary portfolios for VaR analysis on a subset of trades
- Ensuring that simulations made by one user do not impact others