SAS Products

SAS High-Performance Risk 3.7: Release Notes

November 2016

New features:
  • Parametric (Delta-Normal) risk measures. Their use assumes a normal distribution of risk factors. The user supplies risk factor history or covariance or correlation matrix as input. The user can provide position Deltas or valuation methods for SAS High-Performance Risk to compute position Deltas.
  • Foreign exchange conversion of user-supplied values (TASK=LOAD_VALUES).
  • User-defined risk measures can now access risk measures from previous horizons. This can be useful for user-defined counterparty credit risk metrics such as effective expected exposure.
  • Data set input for SUBCUBES option to join risk cubes.
  • Mitigation methods enabled when risk cube is user-supplied values (TASK=LOAD_VALUES).
  • Support for cash flow scenarios during both stress testing and simulation.
  • New data input called VALUEDATA to supply variables at the position or counterparty level for pricing or scoring methods. This data can be updated for stress testing.
  • User can now update computed variable methods and PostVar methods when joining risk cubes or querying risk cubes.
  • Option to specify output variables as cumulative. SAS High-Performance Risk sums variables across horizons to produce cumulative results at query time.
  • The risk factor surface function SURFACE_COORDINATES in SAS Risk Dimensions is now in SAS High-Performance Risk.
  • Risk factor information measures analysis and UI plot is now available when a risk cube is distributed by position.
  • User-supplied horizon labels can be provided to display in the user interface.
  • New option to show or hide mark-to-market by default in the UI.
  • User-supplied labels for the states of a scenario simulation now display in the UI.

Performance:

  • Distribution of risk metric calculations to grid nodes that had previously been performed only on the head grid node.
  • Option to precompute and store select risk measures at the position (loan, trade) level in order to reduce the size of the risk cube.
  • Option to not store risk factor states with the risk cube. If the project has many risk factors and/or many market states, then this data set can be large on the nodes of the grid. This option enables the user to not keep this risk factor data.
  • Option to store select statistics for the position level in addition to base cell values. This option enables the user to compute statistics ad hoc at the base cell level and higher. It also enables the user to retrieve stored statistics at the position level. This can significantly reduce cube size.
  • Option to store a cube that distributes pricing work by market state as though the work were distributed by position. There are advantages to distributing results by position when performing ad-hoc analysis on risk cubes. There can be advantages to distributing pricing by market state when creating a risk cube. This option enables the user to leverage both types of pricing optimally.
  • Option to transpose the scenario results data set. This data set is structured and written in an optimized way so that it is smaller and faster.
  • Option to subset the horizons returned when writing AllPrice, Scen, ScenTrans, SimStat, and SimVal data sets.

Fixes:

  • Show date format instead of SAS date constant in portfolio details view in the UI.
  • Side-by-side comparison no longer requires risk cubes to have the same number of market states.

Post-release Hotfixes:

  • Exporting portfolio details from the user interface only sent 1,000 rows even when the preference setting was above 1,000. User-set preference is now used and the maximum is 10,000 rows.
  • GET_RANKED_VALUES_TOP was fixed so that it retrieves the correct portfolio-level statistics when the risk cube is created with PRICEBY=POSITIONS. This can be used in user-defined risk measures.
  • A threading bug caused missing values when using SCENARIO_CF under certain conditions. This bug has been fixed.
  • A bug encountered when using SAS High-Performance Risk with SAS Event Stream Processing in solo mode has been fixed. You must use release 3.1 when you stream data from SAS Event Stream Processing.