Setup > Analysis

Engine Setup and Risk Analysis

The Risk Analysis and Engine Setup module is The interface to define industry standard risk methodologies, assumptions and other engine settings and acontrols. more ...




Setting Up your Analysis:

Select the methodology you want to run:
You can choose between the following Methodologies

Once you select a new methodology, the setup screen should change and display all the fields that are available for the selected methodology and permission level.

Choose your Horizon(s)


The quality and depth of results always depend on the time and effort put into designing your analysis, the coverage and quality of the stochastic data supplied and the trade granularity!

Select your Type of Exposure

Exposure depends on methodology:

Monte-Carlo And Historical Simulation methodologies. They allow both Market and Credit Risk Analysis.
Correlated Delta Gamma (Parametric)  is a scenario based market risk factor methodology. It does not age positions and therefore is only appropriate for market risk factor measurement.

Risk Exposure Types:

The most important types are Market Risk, Credit Risk and Market and Credit Risk

The type of risk exposure selected will obviously affects largely the kind of results obtained.

If you select Market only, Credit Exposures, Credit Losses and Country Risk will not be available as they are rely heavily of Credit Risk Exposure type.

For example, if you need to measure your current and projected Country Risk, you must activate a  Risk Exposure that includes Credit Risk.

This is necessary since your exposure to country risk is measured through the receivable credit exposure of all accounts belonging to parties with whom you have open transactions that cannot be compensated locally. This means that these trades are affected both by close-out netting, collateral and other credit related factors.

Once you select the Exposure type,  you can proceed to select the different types of Analytics you want to perform.

Market Exposure Active



Setting Description
Var [VaR] Value-at-Risk is the Mean Zero Value-at-Risk, also known as Absolute VaR.
VaR is computed by multiplying the portfolio's standardized volatility by the VaR percentile confidence multiplier.
DEaR DEaR or Daily Earnings-at-Risk. Also known as Relative VaR or Expected Mean VaR..
Expected Mean VaR takes the average position as the mean which will be subtracted from the position's volatility and then multiplied by the percentile confidence multiplier.
VaR as Percent: Produces results relative to Time 0 Market Value. this makes results easier to compare.
Results are computed by dividing Risk results by the NPV of the Trade, Position or Portfolio(s).
Multiple Confidences Multiple Confidence generate Value-At-Risk results for multiple Standard confidence bands: 
Market Risk is produced with  99%, 95% and 90% Confidence Levels
Credit Risk   is produced with 99%, 98% and 97.5 % Confidence Levels.
Note: Multiple Confidences are not produced in the Dynamic/JavaScript Report since you can select confidence multipliers dynamically
Moments Of Distribution Displays the first four moments of your portfolio(s) and its constituents i.e. Mean, Volatility, Skew and Kurtosis .
Advanced user often select this report to assess the quality of return of the distribution of their portfolio and positions. I.e. test Normal Distribution, test Student-T / fat-tail distribution (i.e. even or odd moments at 0, etc.  This measure is often combined with histogram buckets and Quantiles to get additional insight as to way the position behaves as it evolves throughout time.

Credit Exposure Settings


Setting Description
Credit Exposure Credit Exposure Tail Probability as well as individual buckets can be defined for each Account or Tag in the Exposure Section of the Credit Module
The Credit Exposure Bucket Tails are set to 5% by default, which is equivalent to a 95% confidence level. By definition Potential Future Exposure is computed with a 95% confidence.
The credit Exposure Tail Probability can be set to any valid number between 0 and 99.999%
Loss Given Default The loss setting enables loss given default computation. 
 Loss computation assumes a proper Default probability curve or Transition Matrix has been set .
If you provide a Transition Matrix you can select to perform Migration or Transition Convertion.
If you choose to perform migration, parties ratings will migrate until they reach default or the end of the simulation horizon.
If you choose Transition Conversion,  the Transition Matrix will be converted to a credit default curve and thus default and no-default (survival) are measured.
If you have not defined one yourself, the engine will take the default credit curve or transition matrix defined in the credit module Note you cannot run Credit Losses when performing Correlated Delta-Gamma. Historical Simulation is not really recommended, since Credit analysis requires a high number of samples.
The engine does provide the facility to compute Risk Factor simulation with Historical Returns and then perform a series of Random draws to simulate defaults. This is the mixed mode simulation.
Credit Draws Per Market Sample The Credit Draw field defines the number of random draws that will be performed for each Market Risk Run. For example, if you define 10, the engine will draw ten random samples to simulate Bankruptcy for each Market Simulation.
Detail Losses Due To default Provides detailed statistics of Defaults Loss Computation: the number of default states as well as the amount computed as the loss. Each individual default event, including average, peak and individual loss amount and senior counterparty status can be requested.,
Pay Receive By Rating Pay Receive by Rating displays the overall payable and receivable positions per rating rank at inception (at time 0).  If migration is active in the analysis and a one (for all the time steps) or multiple )one for each time step) transition matrices is/are specified, then payable and receivable amounts as well as a separate received / paid interest (to/from cash-accounts) tables are displayed for each simulation horizon and rating rank defined.
Recovery Control You can choose between No recovery (recovery rates are disabled altogether, this means you assume you will never recoup your loss during bankruptcy) Mean Recovery, where you assume the mean recovery rate is constant and and Stochastic Recovery where Recovery Rates are simulated stochastically by taking into account the recovery volatility. See.

Country Risk - Credit Dependant

Setting Description
Country Risks Both Country Exposures and Country Default Losses depend on Credit Exposure. Credit Loss Requires proper credit curve and Rating System. Country Exposure requires country revaluation//devaluation and limit settings in the country section of the credit module.

Valuation Settings

Setting Description
Credit-Curve-Interpolation Continuous, Discrete and None: If you select none, your must provide at least as many credit curve points & periods as your simulation horizons. see
Ageing This feature is active by default except for Correlated Delta-Gamma. If. you disable position ageing, the engine will assume positions do not mature as they evolve over simulation horizons.
Par at Inception Interest Rate Swaps, Bond Forwards and other related fixed income instruments begin at Par when the simulation kicks off (i.e. at time 0). If this feature is disabled, the engine computes the market fair value.
Revalue Collateral On by default. This flags activates Collateral Revaluation at each simulation horizon. If this feature is disabled, the engine assumes the same collateral value that was computed at time 0.
Mean Reversion This flag activates mean reversion calculation in the stochastic simulation of interest rate vertices of yield curves.

Correlated Delta-Gamma Specific
Setting Description
DV/DP Type This fields sets the type of the Parametric Simulation dv/DP change of Price. Either Relative or Absolute. By default the size is assumed to be Relative.
This fields sets the size of the Correlated Delta-Gamma Parametric Simulation dv/DP change of Price.
By default the size is assumed to be 0.01 i.e. 1%. For each Position in the Portfolio and For each risk factor that affects a positions, the engine computes the Market Fait Value, then shifts the risk factor up by 0.5% and then re-computes the new fair-value, it then  bumps the risk factor down by 0.5 % and then re-computes the new fair-value.
The engine then takes ;the average of the difference between the two changes in order to obtain the price change for a given risk factor change. Then it proceeds to the next risk factor(s) and this for each and every position in the portfolio(s).
The approach performs a Taylor expansion of 2nd degree (Gamma) on each Position's Risk Factors. This approach is also called the effective approach as opposed to the Risk-Metrics approach which only takes the theoretical delta (Price adjusted volatility) of each underlying risk factor.
Diversification Benefit: Diversification Benefit is calculated automatically for certain reports. You can however decide to disable this feature if you find it irrelevant.  Diversification Benefit is computed as the difference between aggregated risk with and without correlation.

Liquidity Risk

Setting Description
Liquidity Risk Liquidity Risk is defined at the position level in the liquidity field of the position screen, at a group level in the liquidity field of the tag exposure screen. Liquidity Risk takes into account position ageing and cost of carrying the positions.

Advanced Topics


Analysis Realm and DataSource

Analysis or Setup Data is sinked (written) to or sourced (read) from the Analysis DataSource.

The Analysis DataSource belongs to the Analysis realm.
The Analysis realm belongs to the Input Realm when data is sourced (read) or the output Realm when sinked (written).
If no specific output realm is defined by the user, the output realm
is assumed to be the input realm.

Thus, If the engine cannot read data from the Analysis DataSource, it will seek Data in the (Default) Input Realm.

When data is sunk (written) If no specific Analysis DataSource is specified the application will lookup the Output DataSource. If no Output DataSource has been defined, the engine will use the (Default) Input Realm.

Note: If no input realm has been defined, data will be sinked and sourced
as a file from the local directory according to the default layout specified (XML/RML, Name-Value-Pairs or CSV Data).

The Analysis Realm is akin to the analysis namespace when providing XML/RML payload data.

To get a detailed explanation of each Realm and DataSource fallback mechanism please refer to the User Manual.


Example of DataSource Realm Injection control:

Note: You can define your own entity names and field order by overriding all or parts of the namespace schema(s).

XML/RML Layout:

The engine picks up the ANALYSIS DataSource either as an xml File, or Payload through a Socket or Host connector accordingly:

Entity Based Layout
<analysis>
<methodology>MONTECARLO</methodology>
<businessDate>2012-01-01</businessDate>
<pricingDate>2012-11-31</pricingDate>
</analysis>


 Attribute Based Layout:
<analysis methodology="MONTECARLO" businessDate="2012-01-01"   pricingDate="2012-12-31" .... />

Note: The engine switches automatically between entity based and attribute based layouts.


Name-Value-Pair Layout:

The engine picks up the ANALYSIS DataSource either as a File, Socket, Host or Shared Memory accordingly:

Names are separated from Values with a Space, Tab, Line-feed or carriage return.
You can define your own separators with the SEPARATOR KEYWORD: i.e. SEPARATOR "=").
Names can be defined uppercase, lowercase or a mixture of lower-upper case characters.

Example of NVP ANALYSIS File or Shared Memory (The Shared memory Offset Segment must be named "ANALYSIS". for more information see  the user manual.:

METHODOLOGY MONTECARLO
businessdate 2012-01-01
PricingDate 2012-12-31
		........


Column Based Layout: Comma or TAB Separated Values
The engine picks up the ANALYSIS DataSource either as a Database Table, a File, Socket, Host or Spreadsheet accordingly:

Comma-Separated Values or Column Based Layout depend on order if no entity name is specified.

As a User you can override the order and entity names through the DataSource Schema Control (Database) or by including in the first line   of the file or spreadsheet a comment line (i.e. preceded with // or ! or *) with the order of the Entities as they appear in your file.
By Default, columns are assumed separated with a "," or "\t" (TAB).

Care must be taken not to blend separator and decimal separator which depend on predefined computer  LOCALE settings.

To address this issue you can override these settings with the
SEPARATOR and DECIMALSEPARATOR Keywords. 
i.e.
// SEPARATOR ","
// DECIMALSEPARATOR "."


Example of CSV or Column Based Layout DataSource named analysis.csv 
// Methodology, BusinessDate,PricingDate, ......, 
// SEPARATOR "," (Default Value)
// DECIMALSEPARATOR "." (Default Value)
MONTECARLO,2012-01-01, 2012-12-31
		.......



More information on Schema, DataSource and Realm Control is provided in the Realm section of the user manual.




©2012 RiskServers SA. All rights reserved.