Stress Test & Scenario analysis



Where should I look at about them?



It often seems that no two people mean the same thing when talking about scenarios and stresses!

Our definition is that a scenario is a single cycle of our compute engine, with a particular data set. In practice users want to see some arbitrary number of scenarios, either a small number (e.g. equities shifted -25%, -10%, -5%, 0, +5%, +10%, +25%)

We have two mechanisms. One is what we call ‘micro scenarios’, which are allow you to selectively modify (e.g. shock) market data in a column set using an expression language. This is in the 1.0.0 release arriving on Monday morning at around 8AM BST. You can get a sneak peek a the docs here. This expression language is something of a work in progress, so expect more features to arrive over the next few months.

The second is to use the R or Excel modules to run larger numbers of scenarios and analyse the results statistically (e.g. distribution). Excel and R allow considerably more sophisticated market data manipulation, including treating curves, surfaces and cubes as vectors/matrices/tensors for bulk operations like skews, parallel shifts, single point bumps, etc. R and Excel also have the facility to run historical scenarios using historical data rather than live data. This is not currently available via the standard HTML5 web UI, although we plan to add it in the future.

The R module will be released as part of the 1.0.0 release on Monday under our usual open source license (APL v2) and will run against an open source version of the system either with or without Bloomberg integration. See the demos included with the R module for some examples.