Writing credit models for a hedge fund
We were hired by a large London-based hedge fund to help in writing models for valuing credit derivatives.
The client was a large London-based hedge fund (AUM approximately $5bn). They monitored and traded in many markets, especially credit. The client’s credit research desk was primarily tasked with identifying and analyzing trading opportunities for the fund traders. The research desk required a set of models which:
(a) Integrated into the fund’s existing systems architecture.
(b) Had the ability and speed to calculate in real-time.
(c) Would cover the most liquid credit derivative instruments seen in the market.
(d) Would provide an initial level of sanity checking against market data – to differentiate between genuine potential opportunities and incorrect pricing data.
We initially built a price filtering system. We would assign a confidence in a price based off its distance from previously quoted prices, the liquidity of the instrument, who the quoting party was, and other qualitative factors. There was an element of self-learning within the system. Only if there was enough confidence in the price would it then be used for opportunity spotting.
Then we wrote a suite of models to value the set of credit derivatives traded by the client. These used the standard market-consensus mathematical models for these instruments.
We then extended the models to allow checks for arbitrage across different instruments e.g. given the price of derivative type A, what should the price of derivative type B be?
Finally, we built a scanning system which allowed for opportunities to be spotted and analyzed by the research desk.
Technically, the solution was implemented in C# within the client’s development environment, with pricing data being subscribed to and results published on an internal TIBCO network. The front end was initially in Microsoft Excel but was later replaced by an in-house developed customized front-end.