Analytic Services - MultiHelp

The risk engine’s analytics are accessed via web services. These web services are:

Service	Description
MTM	Trade and portfolio valuation (mark to market or MTM) at current rates.
Decomposition	Decomposition of daily theoretical P&L into the following components; rate movements, theta, maturing trades, cash flows, position changes and new trade valuations. [AggregateDecomposition rolls up results from a list of decomposition calculations].
HedgeEffectiveness	Measure the effectiveness of hedges against long term debt using dollar offset or regression analysis.
LoanImpairment	Calculate provision using historic month end receivables and days past due.
MarketVAR	Historic simulation or Monte Carlo value at risk (VaR) and expected shortfall (ES). Can also be used to calculate initial margin. [AggregateVAR rolls up results from a list of MarketVAR calculations].
TheoreticalPL	Measure theoretical P&L between two dates for VaR or FRTB-IMA model validation.
MarketScenario	Stress and scenario testing. [AggregateScenario rolls up results from a list of MarketScenario calculations].
ExpectedShortfall	FRTB IMA market risk calculation of expected shortfall (ES). Includes calibration to identify the historic period of highest ES volatility. [AggregateExpectedShortfall rolls up results from a list of ExpectedShortfall calculations].
InternalDRC	Merton style model utilizing 10 million path Monte Carlo to calculate FRTB-IMA default risk capital.
PLAttributionMetrics	Calculate P&L attribution metrics as specified in FRTB-IMA.
MarketStandardPreparation	Calculate FRTB sensitivities based on "bump and shift" stress testing. Also, extract data needed for SBA default risk (FRTB-DRC) and residual risk add-on (RRAO).
MarketStandard	Run the FRTB SBA model aggregation using the prescribed risk weights and correlations.
SACCR	Calculate credit exposure for a portfolio using the SACCR specification.
BACva	FRTB basic CVA capital calculation.
SACvaSensitivity	Calculating FRTB CVA sensitivities based on "bump and shift" stress testing and full CVA calculation for each sensitivity.
SACvaSensitivityHedge	Calculate CVA sensitivity hedges based on "bump and shift" stress testing.
SACva	FRTB standardised CVA capital calculation (similar to MarketStandard).
SIMMPreparation	Calculate ISDA SIMM sensitivities based on "bump and shift" stress testing.
SIMM	Calculate initial margin using the ISDA SIMM specification.
CreditExposure	PCE, EPE, CVA using Monte Carlo, including netting rules and collateral.
CreditExposureStress	Changes in PCE due to initial rate shocks. Used to calculate CVA sensitivities as input for CVAVaR.
CreditScenario	Stress testing initial shocks to credit portfolios. Calculates implied changes in collateral.
CurveGeneration	Re-use of instrument definitions and pricing for curve bootstrapping.
ParameterEstimation	Generate stochastic process parameters for curves from historic data. Parameters include volatility, long term average and mean reversion and are used in Monte Carlo simulation.
CorrelationEstimation	Generate inter and intra curve correlations using historic data for use in Monte Carlo simulation.
RateAnalysis	Analyse historic rate series and check for missing, stale and erroneous rates.

The inputs to these web services use a common class hierarchy. The building blocks of this class hierarchy (assumptions, rates and curve evolution parameters, trades etc) are utilized in a consistent fashion across all of the analytic web services. We commonly refer to this as the risk engine API.

These web service inputs fully describe a calculation, so it is possible to run the risk engine as a “stateless appliance”. Due to the volume of input data (eg historic rate data required for market VaR) it is normal to only supply the trade data and assumptions, and have the risk engine look in its database for the rate data. Where trade volumes are high, the trades can also be cached in the risk engine and calls to the web services can simply list the trade references. This is particularly valuable for products such as swaps whose cashflow representation can be large, and therefore slow to transmit across a web service. The Numerix workflow server makes use of this feature.

The analytic web services return the results to the caller as would be expected. For example a call to MarketVaR returns the VaR results across the specified attributions. A call to the credit exposure web service returns a time profile of credit risk, plus other requested statistics like CVA and EPE.

All of the web services can be run synchronously or asynchronously (as non-blocking calls). Data compression is also supported on all of the web services.