Measures - Atoti Hub

This page documents the Ratings cube’s measures. The Ratings cube publishes 14 visible measures of its own across five folders, not counting the engine’s contributors.COUNT and update.TIMESTAMP, and exposes them through the IRatingsCubeMeasures API. The scorecard results sit at the root of the Ratings folder, so the folder a user opens holds only the figures the template reports; the calculation intermediates and the surfaced configuration are in sub-folders beneath it.

Folder	Measures	Accessor group	What it holds
`Ratings`	5	`baseMeasures()` for `Metric Observation Value`, `scorecardMeasures()` for the rest	The figures the scorecard template reports: the raw metric observation value, the metric value each metric is graded from, the metric and category composite ratings, and the internal rating of a counterparty.
`Ratings\Components`	0	`scorecardMeasures()`, except `Period Metric Value` and `Reported Metrics Count` on `baseMeasures()`	The hidden calculation building blocks behind those figures: `Period Metric Value`, `Reported Metrics Count`, `Latest Metric Value`, `Latest Metric Comment`, and, for each rating component, `Metric Value (<method>)`, `Metric Rating (<method>)`, `Composite Rating (<method>)`, `Score (<method>)` and `Total Weight (<method>)`. A component with several aggregation steps adds one rating per step after the first.
`Ratings\Scorecard Configuration`	6	`scorecardMeasures()`, except `Metric Name` on `baseMeasures()`	The configuration in force for the selected Scorecard Date, surfaced as measures: metric and category weights, period names and weights, the metric name, and the Scorecard Date itself. The measures that an application’s lookup tables publish are returned by `scorecardMeasures().lookupMeasures()`.
`Ratings Attributes`	2	`attributeMeasures()`	The counterparty’s one-to-one display attributes: the country of risk, as a code and as a name. They are measures because, as levels, they would only mirror the counterparty axis.
`Counterparty Relationships`	1	`counterpartyRelationshipMeasures()`	The approval measure describing which relationship types a counterparty holds.

A measure’s folder and the accessor group it is declared on are independent: the folder says where a user finds it, the accessor group which interface returns it. Reported Metrics Count is the clearest case: it publishes into Ratings\Components but is declared on IRatingsBaseMeasures alongside the other base metric measures.

What does Metric Value show?

Metric Value is the value each metric is graded from. It switches on Rating Method: each rating component’s value factory reads the value, and the measure answers with the value of the component the metric declares. See Rating components for the value key that selects the factory.

With Period End Date on an axis, a metric read as a period average shows each period’s own value.
A metric read as its latest value shows that value only where Period End Date is not on an axis.
Selecting a subset of periods renormalizes the period weights over the periods selected.

This display rule applies to Metric Value alone. The ratings and scores read the latest value on every reporting period, so with Period End Date on an axis each period’s Internal Rating adds the scores of the components read as period averages, built from that period’s own values, to the scores of the components read as latest values. The building blocks are hidden. A hidden measure stays queryable by name in MDX and is left out of discovery.

Measure	What it holds
`Period Metric Value`	The average of a metric’s observations in each reporting period.
`Reported Metrics Count`	The number of observations feeding `Period Metric Value`.
`Latest Metric Value`	The latest value on or before the Scorecard Date. Observations with no value are skipped.
`Latest Metric Comment`	The comment of that latest value.
`Aggregation Metric Weight`	The weight a metric aggregates by: its configured `MetricWeight`, or 1 where the configuration leaves it blank. It sits in `Ratings\Scorecard Configuration`, and a step’s `weight` can name it.
`Metric Value (<method>)`	The output of the component’s value factory, kept only where a metric declares `<method>`. The grading rule reads it.
`Metric Rating (<method>)`	The grade of the metric under the component, where the metric declares `<method>`.
`Composite Rating (<method>)`	The first aggregation step’s output under the component, at the Category level.
`<Level> Rating (<method>)`	The extra rating that each aggregation step after the first publishes, named after the step’s last level. Its value is what the step reads: the previous step’s output for each member of the step’s last level. The standard route’s second step gives `Category Rating (<method>)`. It sits in `Ratings\Components`.
`Score (<method>)`	The score of a counterparty under the component.
`Total Weight (<method>)`	The total weight that the component’s score is normalized by, and that it carries into the `Internal Rating`.

Every measure carries its own description in the cube metadata, which Atoti UI shows on hover.

Limits measures

The Ratings cube publishes four kinds of limit measure, only when at least one limit method is configured under limit-methods. They are exposed through IRatingsCubeMeasures.limitCapacityMeasures(), which returns null when no method is configured.

Measure	Folder	Visibility	What it shows
`Limit Method`	`Limits`	Visible	The limit method that the `LimitMethod` column of `RelationshipTypes` gives the relationship type.
`Limit Capacity`	`Limits`	Visible	The capacity under the relationship type’s limit method. No value where the relationship type has no method.
`Limit Capacity (<method>)`	`Limits\Components`	Visible	The smallest of the method’s bounds, whichever method the relationship type is mapped to. No value if any bound has no value.
`Limit Bound (<method>: <bound>)`	`Limits\Components`	Hidden	The measure the bound’s factory built for that method.

Each measure shows a value only where the cell resolves to one counterparty and one relationship type the counterparty holds, however each is picked: by Counterparty ID, name or structure, and by Relationship Type or Relationship Type Name. Above that, the measures show nothing, since capacities do not add up across relationship types. For how bounds are written and the capacity is formed, see Limit capacity. A bound that returns a value that is not a number fails the query with an error naming the bound key and the method.

​What does Metric Value show?

​Limits measures

​Related pages

What does Metric Value show?

Limits measures

Related pages