Input file formats

Out of the box, the Market Risk Accelerator will work with a predefined file format, CSV. If you produce your data in this format, you can load and use the Accelerator with no customisations needed. However, you can of course edit and configure the Accelerator to work with any format or file, database source, etc. that ActivePivot is compatible with.

Sample CSV input files are included in the source distribution. These files are loaded during testing of the reference implementation and provide examples of each of the file types.

Overview of Accelerator input files

The input files for the Market Risk Accelerator comprise the following set:

The MRA Input Data Diagram presents an extract of the data model, with the main input tables and how they connect.

File name patterns

The Market Risk Accelerator uses glob patterns with the (*) asterisk wildcard character to identify the relevant file names for each category of input file. So you can add characters before and after the listed names, such as timestamps or ID numbers.

For example, the pattern **/TradePnLs*.csv matches all CSV files with names beginning with the string “TradePnL” in any subdirectory.

In this guide, the File Pattern Match section for each of the input files specifies the glob pattern used. However, the glob prefix is omitted as it is now injected automatically.

You can customize the glob patterns in risk.properties.

Note on AsOfDate

The files in this document that contain an AsOfDate column will rely on that AsOfDate when loaded into the ActivePivot datastores. For the files that do not specify this column (whether described in this document or not), the AsOfDate is taken from the directory structure – these files should reside in the appropriate folder (usually ./data/20xx-yy-zz/ … /*.csv).

Key and optional fields

For each input file, the key and optional fields are indicated in the Key and Null columns respectively.

Labels and dates for pillars

For the pillar information (Tenors and Maturities), both a label and and date field are available, to allow any combination of source data:

• Pillars as labels, with no equivalent dates available in the source system

• Pillars as dates, with no equivalent labels available in the source system

• Pillars as both labels and dates

The dynamic re-bucketing converts the pillar information into a number of days, this is the basis on which the destination bucket is calculated. The primary cube level upon which this conversion is performed can be configured. The accelerator will attempt to use the configured converter to transform the value of the primary level into a number of days, only using the secondary level if the primary has no value.

To allow the slicing and dicing of input data on source pillars, both fields are selected as levels in the cube.

Examples

Note: For the scalar profile, all pillar fields are keys. Therefore 2M,N/A will be a separate fact from 2M,2019-11-05.