We describe a method–used, among others, by financial-software firm StatPro–to perform portfolio risk analysis based on a two-tier client/server approach. The risk server computes the numerical simulations of single-asset prices for a wide universe of investable instruments. The risk clients, using the server outcome, compute portfolio cash risk scenarios, stress-test simulations, and bid/ask liquidity spreads. We focus on the risk-client implementation and describe the details needed to compute the different scenario types for a portfolio of heterogeneous assets. It is also shown how it is necessary to treat differently bond-like instruments that always have a positive quote, futures that are settled on a margin account, and swap-like contracts that may have a positive or a negative net-present value. Finally we show how the computation of daily simulations can be used to estimate risk for longer time horizon, even when the financial instrument considered have special bounding constraints.
Keywords: historical scenarios, client-server architecture, full-repricing, StatPro, market risk, risk, risk simulation, pricing functions, bonds, futures, swaps, derivatives, stress tests, liquidity risk, risk exposure, risk scaling, risk horizon
Pdf file: portfolio-simulated-scenarios.pdfportfolio-simulated-scenarios