MR-specific configuration

Scopes

The scope describes which data is loaded and unloaded. Each request to the Data Load Controller contains a scope. In the default implementation, the as-of date scope is turned into a directory and all the files in that directory are loaded. When unloading data, the scope is turned into a RemoveWhere clause to be run on the datastore. There are three different scopes in the default implementation.

Scope	Description
Configuration Scope	This scope is used for configuration files, including the parameters which are fetched by the parameters API. Data in this scope typically has a “start date” rather than an as-of date. The files in this scope are found in the “configuration” directory. This scope has no parameters.
Historical Scope	This scope is used for historical summary data. This data is not at the trade level and does not include recent dates. Instead it is at the aggregate level (for example, book and legal entity) and contains a long history. The files in this scope are found in the “historical” directory. This scope has no parameters.
As-Of Date Scope	This scope is for the daily trade-level files. The files in this scope are found in directories whose names match the requested date. This scope is parameterized by as-of date. For example: `"scope":{ "AsOfDate": "2018-09-27" }`

Topic Aliases

Each input file is loaded with a topic (see Input file formats). Topic aliases are used to group those topics. The topic aliases are defined in the application.yaml file. Atoti Market Risk contains the following topic aliases:

Alias	Description	Topic
DynamicSensiPillars	Dynamic pillars for Tenor, Maturity and Moneyness	DynamicTenors, DynamicMaturities, DynamicMoneyness
SensiPillars	DynamicSensiPillars	DynamicSensiPillars
DailySensiConfig	The whole configuration for the sensitivities based on an as-of date scope	AllMarketData, RiskFactorsCatalogue, SensiLadders
SensiConfig	DailySensiConfig & SensiPillars	DailySensiConfig, SensiPillars
Sensitivities	Sensitivity metrics	Delta, Gamma, Vega, Volga, Theta, Vanna, Correlation, CrossGamma
AllSensi	Data for one day in the sensitivity cube	DailySensiConfig, Sensitivities
ParentChild	Parent-child stores, that is, with a tree structure	BookParentChild, LegalEntityParentChild, CounterpartyParentChild, Counterparties
Attributes	Attributes related to the underlying, currency, and scenario	TradeAttributes, Scenarios, Countries, MarketShifts
SignOff	Feed the cube level adjustments store used for sign-off	CubeLevelAdjustments
AllConfiguration	The entire static configuration	SensiPillars
AllAttributes	The whole configuration: DailySensiConfig, ParentChild, and Attributes	DailySensiConfig, ParentChild, Attributes
AllFacts	Main data (sensi, var, pnl) for all the cubes	TradePnLs, PnL, AllSensi, BaseStore, PnLBaseStore, SensiBaseStore
AllMarketData	All the market data stores needed to compute the PnL explan	SpotMarketData, CubeMarketData, FxRateMarketData, CurveMarketData, SurfaceMarketData, CorrelationMarketData, SplitRatioMarketData, DividendMarketData
All	Everything	AllAttributes, AllFacts

Configuration Properties

The following Spring properties can be set:

Property	Default Value	Description
`mr.data-load.initial-business-dates`	(null)	If set, the set of as-of dates to load at start-up.
`mr.data-load.initial-load-source`	(null)	If multiple sources are defined, this property must be set to choose the right one for the initial load.
`mr.data-load.initial-load-batch-size`	(null)	Number of files per batch when loading the per-date topics during the initial load. When unset or not positive, all files for a topic are loaded in a single request. See How does the batched initial load work.
`dlc.*`		This section of the properties contains all the DLC-specific configuration. Please see the Data Connectors documentation for more details.

The aliases are defined under the dlc.aliases and can be customized.
The input sources are defined under the dlc.local.sources and dlc.csv.azure.sources and can be customized.
The default topics used by Atoti Market Risk are defined by Java beans and not by the DLC configuration. The used file patterns are defined in their own section outside the DLC property set.
The default local csv source setup is provided by the file application-dlc-local-csv.yaml accessible with the Spring profile dlc-local-csv.
The default Azure csv source setup is provided by the file application-dlc-azure-csv.yaml accessible with the Spring profile dlc-azure-csv.

We recommend using the DLC section dlc.csv-topics to add topics to Atoti Market Risk.

Initial Load

On startup, the following topic aliases are loaded (without parameters):

AllConfiguration

Additionally, a list of as-of dates is determined as follows:

If the configuration parameter mr.data-load.initial-business-dates is set, the as-of dates listed are used. For example, in the application.yaml file:

mr:
  data-load:
    initial-business-dates:2018-09-28,2018-09-27

Otherwise, the top level of the data directory is scanned for directories whose names are dates. These dates are used for the as-of dates in the initial load. (This matches the previous start-up behavior).

For this set of as-of dates, the following topics are loaded with the “AsOfDate” field set in the scope:

AllAttributes
AllFacts

Additionally, if there is a next business date available in the data directory, the following stores are loaded for the next day. This is mandatory for the correct calculation of the PNL Explain especially in a distributed environment. For instance, if mr.data-load.initial-business-dates=2018-09-27 then 2018-09-28 will be loaded for:

MarketData
FXRates

How does the batched initial load work?

Setting mr.data-load.initial-load-batch-size to a positive value makes the initial load list the files available for each date and load them in batches of at most that many files, instead of one request per topic. This bounds the size of a single load request on sources holding many files. Listing is performed by a DateFileLister, implemented for the local CSV and AWS S3 sources only. With the dlc-azure-csv profile and a positive batch size, no file is listed and no daily data is loaded, so leave the property unset on Azure deployments. Listed paths are relative to the date-substituted source root and carry no date directory: for a source root of data/#{AsOfDate}, the file data/2025-01-10/Delta.csv is listed as Delta.csv. Topic file patterns are matched against those names, so the shipped defaults (glob:**Foo*.csv) match unchanged, as do a glob: pattern whose leading component is a wildcard segment or / (glob:**/Delta*.csv, glob:*/Delta*.csv) and a regex: pattern written against either the bare or the rooted name (regex:Delta.*\.csv, regex:.*/Delta.*\.csv). What cannot match a bare name is a glob: pattern carrying a separator anywhere else (glob:**/sensitivities/Delta*.csv): it selects nothing and the topic is skipped with a No files matched pattern log line.

REST Service

HTTP POST URL: /mr-application/connectors/rest/dlc/v2/execute Body (sample):

{
  "operation": "LOAD",
  "topics": [ "TradePnLs", "TradeAttributes" ],
  "scope": {
      "AsOfDate": "2023-09-28"
    }
}

Summary/trend data

Depending on whether the summary cubes are activated (Var-ES Summary cube, Sensi Summary cube, PnL Summary cube), the corresponding topics will load a different file into a different store:

property key	value	topic	file pattern	base store	Cube
cubes.enable.var-summary	true	BaseStore	*VaR-ES Cube.csv	BaseStore	Var-ES Summary cube
cubes.enable.var-summary	false	BaseStore	*SummaryVaR.csv	TradePnLs	Var-ES cube
cubes.enable.pnl-summary	true	PnLBaseStore	*PLCube.csv	PnLBaseStore	PnL Summary cube
cubes.enable.pnl-summary	false	PnLBaseStore	*SummaryPL.csv	PnL	PnL cube
cubes.enable.sensi-summary	true	SensiBaseStore	*Sensitivity Cube.csv	SensiBaseStore	Sensi Summary cube
cubes.enable.sensi-summary	false	SensiBaseStore	*SummarySensitivity.csv	TradeSensitivities	Sensi cube

​Scopes

​Topic Aliases

​Configuration Properties

​Initial Load

​How does the batched initial load work?

​REST Service

​Summary/trend data