Risk System*

Business Problems:

The client is a multi-billion dollar hedge fund. The client has a risk management system that was developed a few years ago that computes risk index for both equity and fixed-income portfolios. The system calculates 30+ risk measurements including all common Greeks. The fixed income instruments include both traditional and exotic and structured products.

The system is in a high performance programming language. Because of the shortage of programming resources at the early stage of the business, the system was built rather too quickly without a well thought through architecture. The system was developed for windows and later was ported to Linux. The system has been loyally producing rightful numbers and was sufficient when the business was small but gradually got unmanageable when the business grew to handle new strategies, new products, new currencies, and new regions.

Solutions and Achievements:

We re-architected the system using a rich combination of methodologies such as design patterns, OO concept, config files, and utility deposit. The architecture was made simple, flexible, and extensible to handle new algorithms, new strategies, and new products. The usability of the system was improved dramatically through well-hinted error messages, deep analytic reports, and straightened logic structures.

We started as a small project engagement but had quickly won trust and confidence of the client so that we can take on more challenging work such as re-architecting the system and on-going maintenance such as feature updates and algorithm changes of the system.

Technical details:

  • Fixed income pricing and risk
  • Cover most traditional and exotic products
  • Proprietary risk technology and pricing algorithms
  • Open source standard quant libraries
  • Cover instruments in dozens of countries and currencies
  • Main engine in a high performance language with a web front end
  • Market data from leading data providers
  • Linux environment
  • * This is a hypothetic case study due to strong industry compliance requirements but the real cases are comparable technically and the engagement scenarios are similar too.