Understand Metrics and Statistics

Understand analysis result metrics and statistics

Overview

The Risk Data API exposes operations for retrieving metrics, statistics, and metadata from analysis results. Distinct operations may be used depending on the model (ALM, DLM, or HD) used to model the result data.

Platform API supports operations for retrieving analysis result data:

  • A loss table returns output data (metrics, statistics, and metadata) for an EP or non-EP analysis. Loss tables are generated from analysis using both probabilistic and deterministic models. DLM analysis data is stored in an event loss table (ELT). HD analysis results are stored in a period loss table (PLT). EP metrics and statistics are based loss table data.
  • EP metrics define quantitative measurements that represent probability that losses will exceed a certain amount. EP metrics are computed for DLM or HD analysis using probabilistic models.
  • Statistics is a calculation based on loss table data. Platform returns statistics for both probabalistic and deterministic models.
  • Result metadata is data about analysis including the financial perspectives, regions, cedants, and treaties that factored into the analysis result.

Distinct data is available in the result depending on the model (ALM, DLM, HD) and type (EP, non-EP) of the analysis.

Analysis Types

The Intelligent Risk Platform supports a variety of deterministic and probabalistic analysis types for each model type (ALM, DLM, HD).

Every analysis of an exposure is defined by a model profile and output profile. The output profile specifies analysis settings including the analysis type, engine, and the peril.

Analysis TypeClassEngineEntitlementPeril*
AccumulationEX, RM, UW
CA DOIDLMRM, RL, UWEQ
Exceedance ProbabilityprobablisticDLM, HDRM, RL, UWCS, EQ, FL, WC, WS, WT,
Footprint FiledeterministicDLM, HDRM, RL, UWEQ, FL, WC, WS
Hazard CurveDLMRM, RL, UW
HistoricaldeterministicDLM, HDRM, RL, UWEQ, WS, WT,
Maximum CredibleDLMRM, RL, UWEQ, WC
Maximum HistoricalDLMRM, RL, UWWS
Non-RunnableRM, RL, UW
ScenariodeterministicDLM, HDRM, RL, UWEQ, WC
SimulatedDLMRM, RL, UWCS, FL, WS, WT,
RollupTQ
User DefinedDLMRM, RL, UWEQ, WC

Different analysis types are available for different perils. The analysis type may be available for some regions and unavailable in others.

TermDescription
Accumulation
CADOIAnalysis type that applies to California (CA) earthquake analyses. It analyzes exposure to generate the CA Department of Insurance (DOI) Form A report, used for reporting primary insurance to the DOI. This analysis applies damage ratios, which are selected based on a number of factors as determined in the CA DOI PML table as provided by the DOI for the purpose of Form A submissions.
Exceeance ProbabilityCumulative distributions showing the probability that losses will exceed a certain amount, from either single or multiple occurrences. Losses are expressed in the occurrence exceedance probability (OEP) and the aggregate exceedance probability (AEP) curves.
Hazard Curve
HistoricalAnalysis that calculates loss based on parameters defined for one or more actual historical events from the event catalog.
Maximum CredibleAnalysis type that searches through all possible events to determine how each may affect exposed locations and selects the event that would cause the worst damage for the selected financial perspective.
Maximum Historical AnalysisMaximum Historical Analysis type that searches through all historical events that may affect exposed locations and selects the event that would cause the worst damage for the selected financial perspective.
ScenarioAnalysis that calculates loss based on parameters defined for one or more individual events from the stochastic event catalog.
SimulatedSamples the ground up damage ratio for each location-coverage from the modeled severity distribution for each simulated event in order to represent secondary uncertainty.
RollupAn analysis that calculates summary and segmented metrics and statistics. Intelligent Risk Platform supports rollup analysis for programs, program variations, and business hierarchies.
User DefinedAnalysis type that creates a custom earthquake event with basic parameters such as the magnitude, region, latitude, longitude, depth, rupture length, orientation, and attenuation to analyze the impact of an event that does not exist in the current stochastic or historical event set and does not have a footprint file.

Loss Tables

A loss table is a table of output data (metrics and statistics) computed for loss-causing events.

Platform analysis generates loss tables for both deterministic and probable analysis models. Analysis data based on ALM and DLM models are stored in event loss tables (ELTs). Analysis data based on ALM and DLM models are stored in event loss tables (ELTs).

Metrics and statistics are based on output data.

OperationModelAnalysis TypeDescription
Get ELTALM, DLMEP, Footprint, Historic, ScenarioReturns an ELT (event loss table). For each event, returns metrics and statistics that define computed losses for that event.
Get PLTHDEP, Footprint, Historic, ScenarioReturns PLT (period loss table). For each event returns metrics and statistics that define computed losses for that event.
Get Non-EP Sampled LossesHDFootprint, Historic, ScenarioReturns losses by sample ID.

Event loss tables

The event loss table (ELT) is an output table that loss-causing events in a DLM analysis, including the mean loss standard deviation (split into an independent and a correlated piece), exposure value, and event rate.

For each event, the table lists metrics, statistics, and metadata that characterize that event. This data is used to compute EP curves and statistics based on this output data. The ELT is the basis of losses for all financial perspectives at all exposure levels and is used in computing output statistics.

The Get ELT operation returns an array of objects that returns metrics, statistics, and metadata for each event.

[
  {
    "analysisId": 83343,
    "sourceId": 22257,
    "eventId": 276,
    "positionValue": 3473657.880534969,
    "perspectiveCode": "GU",
    "stdDevI": 1066771.2741421806,
    "stdDevC": 400576.7724183246,
    "expValue": 4343900.0,
    "rate": 1.2056769946866552e-6,
    "peril": "Windstorm",
    "region": "North America",
    "oepWUC": 1.205676267823641e-6,
    "exposureResourceId": 0,
    "exposureResourceType": "UNRECOGNIZED",
    "exposureResourceNumber": "string"
  }
]
MetricDescription
analysisIdID of analysis result.
sourceId
eventIdID of event, a representation of a peril that may cause catastrophe losses.
positionValue
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
stdDevIStandard Deviation Independent. The independent standard deviation for a portfolio assumes that all locations are completely independent, which means that knowing the size of the loss at one location does not provide any information about the size of the loss at the other location.
stdDevCStandard Deviation Correlated. The correlated standard deviation for a portfolio assumes that all locations are completely correlated, which implies that if the losses are large for one location, they are likely to be large for the other location.
expValueValue of exposure.
rate
perilOne of Accumulation, Earthquake, Fire, Flood, Hurricane, Infectious Disease, Multi-Peril, Severe Convective Storm, Terrorism, Tornado, Unknown, Unrecognized, Wildfire, Windstorm, Winterstorm
oepWUC
exposureResourceIdID of exposure analyzed.
exposureResourceTypeOne of ACCOUNT, AGGPORTFOLIO, LOCATION, POLICY, PORTFOLIO, STEP_POLICY, TREATY, UNRECOGNIZED.
exposureResourceNumber

Period loss tables

The period loss table (PLT) is an output table that simulates event losses over the course of a time period for a HD analysis. The PLT provides greater flexibility to evaluate loss metrics than the analytical calculations based on event loss tables (ELTs).

By simulating events through time, an HD model computes total loss as well as maximum event occurrence loss for each simulation period in the table, and generates loss statistics based on the distribution of losses across the large number of simulated periods. This methodology can calculate the impact of all contract terms, including terms with time-based features, such as contracts that are shorter (or longer) than a single year.

The Get PLT operation returns period loss table calculated using an HD model for the specified HD analysis result.

PLT data may be filtered by exposureResourceType. One of ACCOUNT, PORTFOLIO, LOCATION, POLICY, STEP_POLICY, TREATY, AGGPORTFOLIO

[
  {
    "periodId": 503,
    "weight": 2.0e-5,
    "eventId": 3508644,
    "eventDate": "2020-08-07T00:00:00.000Z",
    "lossDate": "2020-08-13T00:00:00.000Z",
    "positionValue": 111642.35349968076,
    "peril": "Earthquake",
    "region": "string",
    "exposureResourceNumber": "string"
  }
]
MetricDescription
periodIdID of the period in EP analysis.
weightFor the full weighted period event table, the weights sum to one. However, if the weighted period loss table does not contain all the events in the model, due to those events not taking loss based on the portfolio’s exposure or if there are no events in that year, then the sum of the weights is less than one.
weightPeriod loss tables (PLTs) represents primary uncertainty using the weight field. The weight represents the likelihood that one period occurs when compared to the other periods. For climate models, all periods are equally weighted and are calculated as one (1) divided by the number of simulation periods used for the analysis. For earthquake models, the weight can vary by period and is derived during the development of the period event table (PET)
eventIdID of event within stochastic event set.
eventDateDate that event occurred. When event dates are used in the context of a simulation period for HD analyses, they operate in connection with loss dates. For example, in simulation period 100, event 200 starts on April 1 (the event date). The event affects segments of Portfolio 1 on April 2 and April 4 (the loss dates shown in the PLT for Portfolio 1). The same event affects segments of Portfolio 2 on April 3 and April 4 (the loss dates shown in the PLT for Portfolio 2). Event 200 may recur in simulation period 400, on a different date.
lossDateDate that policy payout occurred. In the HD financial model, a policy payout is assigned only a single date. This date is recorded as the Loss Date in the Period Loss Table (PLT) and reflects the earliest date that a location covered by that policy experienced damage by that specified event occurrence.
positionValue:
perilName of peril. One of Accumulation, Earthquake, Fire, Flood, Hurricane, Infectious Disease, Multi-Peril, Severe Convective Storm, Terrorism, Tornado, Unknown, Unrecognized, Wildfire, Windstorm, Winterstorm
regionID of region .
exposureResourceNumber

Non-EP Sampled Losses

The Get Non-EP Sampled Losses operation returns sampled non-EP losses from the PLT (period loss table) of the specified analysis result.

This operation takes four required parameters: analysisId, perspectiveCode, exposureResourceType, and events, a comma-separated list of events identified by event ID. The optional lossSampleLimit parameter...

An event is a representation of a peril with the potential to generate catastrophe losses. For a given peril, all potential events are synthesized into a stochastic event set, which defines the occurrence and magnitude of each event.

For each event, the response returns the eventId, sampleId, lossType, and loss, as well as the names of the cedant (cedantName) and line of business (lobName) associated with that loss.

[
  {
    "cedantName": "string",
    "lobName": "string",
    "lossType": "string",
    "sampleId": 0,
    "eventId": 0,
    "loss": 0
  }
]
MetricDescription
cedantNameName of risk-holding party (insurer/reinsurer) that transfers a portion of its risk to another risk-holding party (reinsurer/retrocessionaire).
lobNameName of line of business classification, e.g. auto, commercial, fire, personal, residence. Defining LOB is required for some types of reinsurance to apply."
lossTypeCoverage type of loss. One of 1 (Building), 2 (Contents), 3 (BI), 4 (Combined Coverage).
sampleIdID of sample in non-EP analysis.
eventIdID of event.
lossValue of loss associated with the financial perspective.

EP Metrics

An EP analysis computes EP curves based on loss table data. An EP curve is a cumulative distributions showing the probability that losses will exceed a certain amount, from either single or multiple occurrences. Losses are expressed in the occurrence exceedance probability (OEP) and the aggregate exceedance probability (AEP) curves.

AEP and OEP curves are two different curves that have two distinct uses and offer different information. Both curves show the probability that losses will exceed a given threshold.

TypeModelDescription
AEPALM, DLM, HDProbability that the total losses across all events in a year will meet or exceed a loss threshold.
OEPALM, DLM, HDProbability that at least one event will occur in a year that causes losses greater than or equal to a certain amount.
TCE-AEPALM, DLMExpected annual or maximum loss given that this loss is greater than or equal to the return period loss corresponding to RP. Also known as Tail Value at Risk (TVaR).
TCE-OEPALM, DLMExpected annual or maximum loss given that this loss is greater than or equal to the return period loss corresponding to RP. Also known as Tail Value at Risk (TVaR).

Metrics are quantitative measurements." "Metrics are the raw data you collect to measure aspects of the product, and analytics is the process of interpreting the data to gain insights and make data-driven decisions." An EP analysis returns an EP curve which consists of computed metrics.

OperationModelCurvesDescription
Get EP MetricsALM, DLM, HDEPReturns EP curves calculated using ALM, DLM, or HD models.
Get Interpolated EP MetricsHDAEP, OEPReturns EP curves based on return period or critical probability values interpolated into an HD EP analysis.
Get Marginal EP MetricsDLMAEP, OEP, TCE-AEP, TCE-OEPReturns EP metrics for the specified DLM analysis.

EP Interpolation Metrics

The Get EP interpolation metrics operation returns EP metrics based on interpolated return period or critical probability values.

This operation accepts a new returnPeriod or criticalProbability value, which it inserts ("interpolates") into the PLT of the specified HD EP analysis. The Intelligent Risk Platform then computes new EP curves based on the interpolated value and returns updated EP metrics for the new return period. The request can also specify the perpectiveCode that defines the financial structure used to calculate modeled losses.

For example, the following request interpolates a new return period and specifies that metrics are calculated based on the GR (gross loss) financial perspective.

{
  "returnPeriod": 13,
  "criticalProbability": 0.075,
  "result": [
    {
      "perspectiveCode": "GR",
      "values": [
        {
          "metricName": "TCE-OEP",
          "metricValue": "68418.2476820577"
        }
      ]
    }
  ]
}

An exceedance probability (EP) curve indicates the probability that losses will exceed a given threshold within a year. This operation returns AEP curves and OEP curves.

This operation returns the following result properties:

MetricDescription
returnPeriodPoint on an EP curve that describes the likelihood of exceeding a loss threshold from the single largest event (OEP) or the aggregation of one or more events (AEP). Return period is defined as the inverse of the annual exceedance probability.
criticalProbabilityLikelihood of exceeding a loss threshold at a specific return period in an EP curve. Defined as the inverse of the annual exceedance probability.
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
valuesArray of result objects that consist of a metricName and metricValue.
metricNameEP curve for modeled losses, e.g. AEP, OEP, TCE-AEP, TCE-OEP.
metricValueModeled losses at interpolated return period.

EP metrics [EP Metrics (DLM Models)]

Returns exceedence probability (EP) metrics for the specified <glossary:ALM> or <glossary:DLM>> analysis result.

DLM and ALM EP analyses write modeled losses to an event loss table (ELT), compute EP metrics, and calculate summary statistics based on ELT data.

EP metrics define EP curves that represent the probability that losses will exceed a certain threshold. ALM and DLM analysis supports five types of EP curves: OEP (occurence EP), AEP (aggregate EP), CEP (stochastic conditional EP), TCE-AEP (tail conditional expectation AEP), and TCE-OEP (tail conditional expectation OEP) curves.

The Get EP Metrics request returns an array EP curves:

[
  {
    "jobId": 8,
    "epType": "AEP",
    "perspectiveCode": "GU",
    "exposureResourceId": 0,
    "exposureResourceType": "PORTFOLIO",
    "exposureResourceNumber": "8",
    "value": {
      "returnPeriods": [
                        1.0,
                        2.0,
                        3.0,
                        4.0,
                        5.0,
                        ...,
                        5000.0,
                       ],
      "positionValues": [
                         0,
                         0,
                         0,
                         1.4336826286215238,
                         2898128.185059362,
                         ...,
                         2923391.5767636155
                        ]
    }
  }
]

For each EP curve, the response shows the following:

MetricDescription
jobIdID of model job that generated the analysis result.
epTypeType of exceedence probability curve. One of AEP, CEP, OEP, TCE_AEP, TCE_OEP
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
exposureResourceIdID of data resource analyzed.
exposureResourceTypeType of data resource analyzed. One of ACCOUNT, AGGPORTFOLIO,LOCATION, POLICY, PORTFOLIO, STEP_POLICY, TREATY, UNRECOGNIZED
exposureResourceNumber
returnPeriodsArray of return periods. A point on an EP curve that describes the likelihood of exceeding a loss threshold from the single largest event (OEP) or the aggregation of one or more events (AEP). Return period is defined as the inverse of the annual exceedance probability. For example, a return period of 100 years corresponds to an annual exceedance probability of 1%. In the context of peril events, return period refers to the number of years between occurrences of an event of a given size in the region. Short and long return periods enable modelers to estimate risk at both short and long-range exceedance probabilities.
positionValues

Marginal EP Metrics

The Get Marginal EP Metrics operation returns projected differential losses expressed as marginal EP curves for the specified analysis result. This operation may be used to retrieve marginal EP metrics modeled using ALM, DLM, or HD models.

"Marginal impact analysis enables insurers to identify higher-risk policyholders who may require revised premiums or additional risk management measures."

An exceedance probability (EP) curve indicates the probability that losses will exceed a given threshold within a year. This operation returns four different types of EP curves: AEP, OEP, TCE-AEP, and TCE-OEP curves.

A marginal EP curve shows projected losses and differential losses (as raw numbers and percentage) between X and Y at each position in the curve. These numbers enable you to view the projected loss for X and Y at any given position, the difference between the value of X and Y at that position, and the percent difference between X and Y.

The request accepts three required parameters an analysisId, perspectiveCode exposureResourceType that specify the scope of the query.

[
  {
    "exposureResourceId": 0,
    "exposureResourceType": "UNRECOGNIZED",
    "exposureResourceNumber": "string",
    "perspectiveCode": "Empty",
    "epType": "UNRECOGNIZED",
    "value": {
      "returnPeriods": [0],
      "eps": [0],
      "positionValues": [0],
      "positionValueDiffs": [0],
      "positionValueDiffPercents": [0]
    }
  }
]
PropertyDefinition
exposureResourceIdID of exposure resource analyzed.
exposureResourceTypeType of exposure modeled. One of ACCOUNT, AGGPORTFOLIO, LOCATION, POLICY, PORTFOLIO, STEP_POLICY, TREATY, UNRECOGNIZED.
exposureResourceNumber
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
epTypeType of exceedence probability metrics. One of AEP (aggregate EP), CEP (stochastic conditional EP), OEP (occurence EP), TCE_AEP (tail conditional expectation AEP), TCE_OEP (tail conditional expectation OEP)
returnPeriodsList of return periods. Each return period is a point on EP curve that describes the likelihood of exceeding a loss threshold from the single largest event (OEP) or the aggregation of one or more events (AEP). Return period is defined as the inverse of the annual exceedance probability. For example, a return period of 100 years corresponds to an annual exceedance probability of 1%. In the context of peril events, return period refers to the number of years between occurrences of an event of a given size in the region. Short and long return periods enable modelers to estimate risk at both short and long-range exceedance probabilities.
epsList of EP curves?
positionValuesList of position values.
positionValueDiffsList of diffs between position values.
positionValueDiffPercentsList of percentage diffs between position values.

The number of marginal EP curves returned depends on engine to model the results.

Different EP curves may be returned depending on the model used in the analysis:

ModelCurves
ALMAEP, OEP, TCE-AEP, TCE-OEP
DLMAEP, OEP, TCE-AEP, TCE-OEP
HDAEP, OEP

Statistics

A statistic (AKA loss statistic) is a calculation performed on output data, e.g. in an ELT or PLT. A standard deviation, mean loss, coefficient of variation are all statistics. Compare metrics.

OperationAnalysis TypeModelDescription
Get EP StatisticsEPDLMReturns statistics calculated base on an EP analysis, including cv, netStdDev, premium, purePremium.
Get Non-EP StatisticsFootprint, Historic, ScenarioALM, DLM, HD?Returns statistics calculated based on a non-EP analysis, including stdDev, meanLoss, cv
Get Marginal StatisticsEPALM, DLM, HDReturns treaty-level statistics including cv, totalStdDev, and purePremium.
Get Location-Level StatisticsDLM, HDReturns stdDev, aal, cv

Get Statistics

The Get Statistics operation returns a list of DLM-based metrics by analysis result.

Unlike other operations that retrieve metrics and statistics, this operation returns data from multiple analysis results.

Statistics include the average annual loss, standard deviation, and coefficient of variation.

[
  {
    "analysisId": 54,
    "exposureResourceId": 765,
    "exposureResourceType": "ACCOUNT",
    "perspectiveCode": "TY",
    "epType": "AEP",
    "purePremium": 0,
    "totalStdDev": 0,
    "cv": 2.808502733065068,
    "netPurePremium": 85213.63686810755,
    "activation": 0.19734192801321804,
    "exhaustion": 0.06465721482368603,
    "totalLossRatio": 0.17429420184027278,
    "limit": 900000,
    "premium": 25000,
    "netStdDev": 231466.464883724,
    "exhaustAllReinstatements": 0.0004301273657816296,
    "exposureResourceNumber": "Cat_EntirePort"
  }
]

For each analysis result, the response returns the following properties:

MetricDescription
analysisIdID of the analysis result.
exposureResourceNumberID of analyzed exposure resource.
exposureResourceTypeType of exposure modeled. One of ACCOUNT, AGGPORTFOLIO, LOCATION, POLICY, PORTFOLIO, STEP_POLICY, TREATY, UNRECOGNIZED.
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
epTypeType of EP analysis. One of AEP, CEP, OEP, TCE_AEP, TCE_OEP
purePremiumExpected value of the aggregate loss distribution. Pure premium can be computed in two equivalent ways: first, as the area under the aggregate exceedance probability (AEP) curve, or second, as the sum-product of the individual event losses and rates." "Total pure premium. Equivalent to expected loss on an annual basis.
totalStdDev
cvVolatility in the annual losses. CV (coefficient of variation) provides a comparative basis for assessing diversification of risk and compares volatility across analyses, such as between portfolios.
netPurePremiumEstimate of premium required to balance risk, considering the reinstatement premiums to be paid. Base figure for catastrophe treaties with reinstatement provisions.
activationProbability that limit value is activated based on the OEP curve.
exhaustionProbability that limit value is exhausted based on the OEP curve.
totalLossRatioProbability aggregate losses in a year are greater than or equal to the premium. Based on the AEP curve.
limitELT expValue treaty layer amount, i.e. the maximum loss that can be incurred for each event at the perspective and aggregation level with which the exposure value is associated.
premiumPremium specified in the treaty.
netStdDevStandard deviation associated with the portion of catastrophe treaty loss before reinstatements.
exhaustAllReinstatementsProbability that aggregate treaty payout exceeds Reinstatements (%) (k+1) \* treaty occurrence limit, where k is the number of reinstatements. Applicable to catastrophe treaties only.
exposureResourceNumberUser-defined ID of modeled resource.

Marginal Statistics

The Get Marginal Statistics operation returns marginal statistics for the specified analysis result.

Treaty Losses: Risk Modeler makes losses viewable at the individual treaty level so that you can better assess capacity and coverage and make corresponding pricing decisions. You can view losses at the treaty level for the treaty types listed in Table 67. Add treaties pre- and post-analysis for DLM, ALM, and HD analyses. Risk Modeler does not include the ability to view individual facultative losses.

The loss dashboard displays the following: Return period losses, along with Aggregate Exceedance Probability curves (AEP) and Occurrence Exceedance Probability curves (OEP), Pure Premium (AAL), Standard Deviation, Coefficient of Variation (CV).

The request takes three required parameters: analysisId, perspectiveCode, and granularity.

ParameterTypeDescription
analysisIdstringID of analysis result.
perspectiveCodearrayFinancial perspective used to model losses at each granularity.
granularityarrayLevel of detail of marginal statistics. One of ACCOUNT or POLICY

Responses may be optionally filtered and sorted using query string parameters.

The response returns marginal statistics.

[
  {
    "exposureResourceId": 26113,
    "exposureResourceType": "TREATY",
    "exposureResourceNumber": "Cat_EntirePort_04",
    "perspectiveCode": "TY",
    "epType": "AEP",
    "limit": 900000,
    "exhaustion": 0.06465721482368603,
    "activation": 0.19734192801321804,
    "cv": 2.808502733065068,
    "cvDiff": 2.7163077811358347,
    "cvDiffPercent": 0,
    "totalStdDev": 3676637.4187586145,
    "totalStdDevDiff": 0,
    "totalStdDevDiffPercent": 0,
    "totalLossRatio": 0.17429420184027278,
    "netStdDev": 231466.464883724,
    "premium": 0,
    "purePremium": 493148.5493424345,
    "purePremiumDiff": 0,
    "purePremiumDiffPercent": 2.7777777777777777,
    "netPurePremium": 85213.63686810755,
    "exhaustAllReinstatements": 0.0004301273657816296
  }
]
PropertyDefinition
exposureResourceIdID of exposure.
exposureResourceTypeType of exposure modeled. One of ACCOUNT, AGGPORTFOLIO, LOCATION, POLICY, PORTFOLIO, STEP_POLICY, TREATY, UNRECOGNIZED.
exposureResourceNumber
perspectiveCodeFinancial structures considered in calculation of the loss statistics. See Financial Perspectives.
epTypeType of exceedence probability metrics. One of AEP (aggregate EP), CEP (stochastic conditional EP), OEP (occurence EP), TCE_AEP (tail conditional expectation AEP), TCE_OEP (tail conditional expectation OEP)
limit"Estimated catastrophe risk expressed as a rate against capital committed. A “Risk on Line” statistic. Calculation—Pure premium as a percentage of the layer amount. Application—Shows the expected annual burn on the total limits exposed. Compare to the Rate on Line n Use in comparative analysis"
exhaustionProbability that limit value is exhausted based on the OEP curve.
activationProbability that limit value is activated based on the OEP curve.
cvFor non-EP analysis, the coefficient of variation is calculated by dividing stdDev of the event losses by the `meanLoss. Event CV provides a basis for assessing the relative uncertainty in the mean estimate of a loss, such as between perspectives or portfolios.
cvDiffCoefficient of variation associated with the portion of catastrophe treaty loss before reinstatements.
cvDiffPercent
totalStdDevStandard deviation associated with the portion of catastrophe treaty loss before reinstatements.
totalStdDevDiff
totalStdDevDiffPercent
totalLossRatioProbability aggregate losses in a year will be greater than or (%) equal to the premium. Based on the AEP curve. Note: This requires that you enter layer premiums.
netStdDevStandard deviation associated with the portion of catastrophe treaty loss before reinstatements.
premiumPremium coded for the treaty
purePremiumExpected value of the aggregate loss distribution. Also known as average annual loss (AAL). Pure premium can be computed in two equivalent ways: first, as the area under the aggregate exceedance probability (AEP) curve, or second, as the sum-product of the individual event losses and rates.
purePremiumDiff
purePremiumDiffPercentEstimated catastrophe risk compared against premium collected. Estimated pure premium divided by layer premium. The pure premium used in this calculation does not consider any expected reinstatment premiums. Measurement of premium adequacy.
netPurePremiumEstimate of the up-front premium required to balance catastrophe risk over time, after considering the expected amount of reinstatement premiums to be paid. Provides a base, unloaded premium figure for catastrophe treaties with reinstatement provisions. Provides a base, unloaded premium figure for catatastrophe treaties with reinstatement provisions.
exhaustAllReinstatementsThe probability that the aggregate treaty payout exceeds Reinstatements (%) (k+1) * treaty occurrence limit, where k is the number of reinstatements. Applicable to catastrophe treaties only.

Depending on the model, returns treaty-level losses for the following treaty types:

Treaty TypeModels
Surplus StoreDLM, HD
Quota ShareDLM, HD
Working ExcessDLM, HD
Non-CatastropheALM
CatastropheALM, DLM, HD
Corporate CatastropheALM, DLM, HD
Stop LossALM, DLM, HD

This operation does not return individual facultative losses.

Get Non-EP Statistics

The Get Non-EP Statistics operation returns key losses by event for the specified analysis result.

A non-EP analysis is a deterministic analysis based on a user-specified set of events. The Platform API supports several types of deterministic analysis, including footprint analysis, historical analysis, and scenario analysis.

The response returns an array of key loss metrics for each event modeled:

[
  {
    "eventId": 2864907,
    "stdDev": 1663.1549792686967,
    "meanLoss": 0,
    "cv": 6.62779067641994
  }
]

For each event, the response returns cv, meanLoss, and stdDev metrics:

PropertyDefinition
cvFor non-EP analysis, the coefficient of variation is calculated by dividing stdDev of the event losses by the `meanLoss. Event CV provides a basis for assessing the relative uncertainty in the mean estimate of a loss, such as between perspectives or portfolios.
meanLossThe mean loss represents the expected loss for an event for the corresponding position or financial perspective.
stdDevThe standard deviation associated with a mean loss value characterizes the secondary uncertainty associated with that meanLoss value, that is, the uncertainty in an event loss, given that a certain event has occurred.

Location-Level Statistics

The Get Location-Level Statistics operation returns average annual loss (AAL), coefficient of variation (CV), and standard deviation statistics for modeled location exposures. Using these statistics, underwriters can identify the location exposures in their portfolios that are at the greatest risk.

To help inform pricing and underwriting decisions, generate the following location-level results for detailed loss model (DLM) and high-definition (HD) results. Location-level losses are not available for ALM:

Location-level statistics are available for analysis results based on both DLM and HD models. These statistics are not available for analysis using ALM.

Location-level ELT and EP curves can be generated using DLM models. , you can also generate EP and Loss Table output at the location level.

The request must specify an analysis ID that was computed using an output profile.

To generate these location-level losses, create an output profile that uses Granularity=Risk for EP and Loss Table output (DLM only). For Statistics output, risk-level granularity is already selected by default (available for both HD and DLM).

For each location, the response returns the locationId, locationName, locationNumber, and AAL statistics.

[
  {
    "locationId": 58,
    "locationName": "Location_Name",
    "locationNumber": "Location_Num",
    "aal": 294.25022798552993,
    "cv": 5.652179067641994,
    "stdDev": 1663.1549792686967
  },
  {
    "locationId": 60,
    "locationName": "Location_Name",
    "locationNumber": "Location_Num",
    "aal": 274.25022798662993,
    "cv": 6.62779067641994,
    "stdDev": 1234.1549792686967
  }
]
MetricDescription
locationIdID of location.
locationNameSystem-defined number of location.
locationNumberUser-defined number of location.
aalThe average annual loss (AAL), sometimes called pure premium or burn cost, is the expected value of the modeled loss distribution. It is the loss one would expect to see in a year on average. Since the AAL represents only an average, the actual annual losses will fluctuate around the AAL in any given year. AAL does not include expenses, non-modeled loss, profit, or risk load. One may be interested in AAL for ground-up, gross, net of reinsurance, or other views of risk.
cvFor exceedance probability (EP) analyses, the coefficient of variation (CV) statistic is a measure of the relationship between the pure premium (AAL) and the standard deviation of the annual losses and reflects the volatility in the annual losses. It is calculated by dividing standard deviation of the annual losses by pure premium. CV provides a comparative basis for assessing diversification of risk and compares volatility across analyses, such as between portfolios.
stdDevThe standard deviation associated with a mean loss value characterizes the secondary uncertainty associated with that mean loss value, that is, the uncertainty in an event loss, given that a certain event has occurred. The event loss table includes correlated and independent standard deviations:The correlated standard deviation for a portfolio assumes that all locations are completely correlated, which implies that if the losses are large for one correlation, they are likely to be large for the other location.The independent standard deviation for a portfolio assumes that all locations are completely independent, which means that knowing the size of the loss at one location does not provide any information about the size of the loss at the other location.Portfolio standard deviation is calculated as a weighted average of these two extreme cases.

Analysis Metadata

Analysis metadata is data about any analysis that provides information about the exposure modeled or analysis configurations defined in the model profile e.g. the region, cedant, or treaty applied to the analyzed exposure.

This data is helpful for understanding the significance of the computed losses, metrics, and statistics in an analysis result.

The Risk Data API supports the operations that enable you to retrieve following analysis metadata.

OperationModelDescription
List Financial Perspectives by AnalysisALM, HD, DLMReturns financial perspectives by result.
List Regions by AnalysisALM, HD, DLMReturns regions by result.
List Cedants by AnalysisALM, HD, DLMReturns cedants by result.
List Treaties by AnalysisALM, HD, DLMReturns treaties by result.
Get Treaty Details by AnalysisALM, HD, DLMReturns treaty details by result.

List Financial Perspectives by Analysis

The List Financial Perspectives by Analysis operation returns financial perspective data about the specified analysis.

A financial perspective (also called a position) is a financial structure that determines how loss statistics are calculated.

For example, the ground up loss (GU) financial perspective represents total loss to the exposure and excludes any insurance or reinsurance terms in the loss calculations. The gross loss (GR) financial perspective represents the loss to the insurer, accounting for the application of all insurance terms but without consideration for any reinsurance recoveries.

This operation returns an array of the financial perspectives used to calculate losses in an analysis. For each financial analysis, returns the code and name:

[
  {
    "perspectiveCode": "GR",
    "perspectiveName": "Gross Loss"
  },
  {
    "perspectiveCode": "GU",
    "perspectiveName": "Gross Up Loss"
  }
]

Multiple positions can be output for the same analysis, and the relevance of each position varies between users and use cases. Some financial perspectives affect estimates of property (P) loss or worker's compensation (WC) loss. To learn more, see Financial Perspectives

List Region Details by Analysis

The List Region Details by Analysis operation returns information about an analysis including details about exposures modelled and the model used in the analysis.

This operation returns an array of the analysis details by region. For each region, this operation returns information about the exposure modeled and model profile details.

[
  {
    "region": "string",
    "subRegion": "string",
    "peril": "Unrecognized",
    "eventRateSchemeId": 0,
    "framework": "ELT",
    "analysisId": 0,
    "modelProfileId": 0,
    "petId": 0,
    "numSamples": 0,
    "periods": 0,
    "applyContractFlag": true,
    "engineVersion": "HDv1.0"
  }
]
PropertyTypeDefinition
regionStringRegion of exposure analyzed.
subRegionStringSubregiion exposure analyzed.
perilStringNatural or man-made phenomenon that generates insurance loss.
eventRateSchemeIdNumberEvent rate scheme ID.
frameworkStringFor example, ELT
analysisIdNumberID of analysis.
modelProfileIdNumberID of model profile.
petIdNumberID of period loss table. A collection of simulated periods and the events simulated to occur within each period. Each period contains the same fixed number of years.
numSamplesNumberNumber of samples in analysis.
periodsNumberNumber of periods in analysis.
applyContractFlagBoolean
engineVersionStringVersion of modeling engine used to model exposure. For example, HDv1.0

List Treaties by Analysis

The List Treaties by Analysis operation returns treaty data about the specified analysis.

A treaty is an agreement between a primary insurer and a reinsurer in which the primary insurer cedes a portion of risk to the reinsurer.
The terms of the treaty

List Treaty Details by Analysis

The Get Treaty Details by Analysis operation returns detailed information about a treaty applied to a specific analysis.

A treaty is an agreement between a primary insurer and a reinsurer in which the primary insurer cedes a portion of risk to the reinsurer.
The terms of the treaty

This operation returns the following detailed infomation about the specified treaty:

{
  "treatyId": 0,
  "treaty": "string",
  "treatyName": "string",
  "cedant": {
    "cedantId": 22,
    "cedantName": "CSW"
  },
  "producer": {
    "producerId": 3,
    "producerName": "rena"
  },
  "treatyType": "CATA",
  "currency": {
    "id": 0,
    "code": "string",
    "name": "string"
  },
  "attachmentBasis": "L",
  "attachmentLevel": "PORT",
  "premium": 0,
  "occurrenceLimit": 0,
  "attachmentPoint": 0,
  "riskLimit": 0,
  "retentionAmount": 0,
  "percentagePlaced": 0,
  "effectiveDate": "2020-01-01T00:00:00.000Z",
  "expirationDate": "2020-01-01T00:00:00.000Z",
  "percentageRetention": 0,
  "percentageRiShare": 0,
  "percentageCovered": 0,
  "priority": 0,
  "numberOfReinstatements": 0,
  "reinstatementCharge": 0,
  "maolAmount": 0,
  "isValid": true,
  "userId1": "string",
  "userId2": "string",
  "aggregateDeductible": 0,
  "aggregateLimit": 0,
  "uri": "string",
  "lobs": [
    {
      "lobId": 9,
      "lobName": "FACTORY",
      "uri": "/riskdata/v1/exposures/5/lobs/9"
    }
  ],
  "lossOccurrences": [
    {
      "id": 0,
      "treatyId": 0,
      "uri": "string",
      "regionPeril": {
        "id": 0,
        "code": "string",
        "name": "string"
      },
      "lossOccurrenceTime": 0,
      "lossOccurrenceRadius": 0,
      "radiusUnit": {
        "id": 0,
        "code": "string",
        "name": "string"
      },
      "multiLossOccurrence": {
        "id": 0,
        "code": "string",
        "name": "string"
      }
    }
  ],
  "analysisId": 0,
  "tagIds": [0]
}

This operation returns the following detailed infomation about the specified treaty:

PropertyTypeDescription
treatyIdStringID of treaty.
treatyNumberStringNumber of treaty. Moody's RMS recommends coding the treaty number as a combination of cedant name, treaty type, and treaty terms.
treatyNameStringName of treaty.
cedantObjectRisk-holding party (insurer/reinsurer) that is transferring a portion of risk to another risk-holding party (reinsurer/retrocessionaire). Includes cedantId and cedantName. See Create Cedant.
producerObjectAgent or brokerage that produced a policy. Includes producerId and producerName. See Create Producer.
treatyTypeStringType of treaty. One of CATA, QUOT, SURP, WORK, CORP, STOP, NCAT. See Create Treaty.
currencyObjectCurrency used by treaty.
attachmentBasisStringDefines how insurance policies have their losses covered. If L (losses occurring), claims are covered that occur during the period of this treaty. If R (risks attaching), claims are covered if policy inception occurs during period of this treaty.
attachmentLevelStringLevel at which treaty is attached. One of ACCT (account), LOC (location), POL (policy), PORT (portfolio).
premiumStringAmount an insurer or reinsurer charges to provide the coverage described in the policy or treaty.
occurrenceLimitStringMaximum amount that paid out for an event.
attachmentPointStringDescription
riskLimitStringMaximum amount paid out for risk in an event. Applies to QUOT and SURP treaties.
retentionAmountStringAmount of risk or loss kept by the ceding company for its own account or for others.
percentagePlacedStringPercentage of a treaty’s coverage that has been accepted by participating reinsurers.
effectiveDateStringEffective date of this treaty in format, e.g. 2020-01-01T00:00:00.000Z. From this date losses are eligible for recovery.
expirationDateStringExpiration date of this treaty in format, e.g. 2020-01-01T00:00:00.000Z. From this date a loss-occurring policy/treaty/program no longer takes loss and a risks-attaching policy/treaty/program no longer allows risks to attach.
percentageRetentionStringPercentage of the reinsurer’s share of treaty losses not covered by retrocessions. Determines the losses for the RN financial perspective and the net premium a reinsurer receives for a treaty.
percentageRiShareStringPercentage of treaty coverage assigned to reinsurer. Used in calculation of RG perspective losses.
percentageCoveredStringPercentage of loss greater than the attachment point covered by a treaty. Not applicable to SURP treaties.
priorityStringInurning priority of this treaty. Lower numbers take losses prior to higher numbers.
numberOfReinstatementsStringNumber of reinstatements. Restoration of the treaty occurrence limit to its full amount after an event that led to a non-zero reinsurance payout so that the reinsured has the full limit coverage for the next event occurrence. Used in calculating catastrophe and corporate catastrophe treaty statistics, including average annual loss. The event loss table and the EP curves assume unlimited reinstatements
reinstatementChargeStringAmount of premium the insured must pay the insurer to reinstate the coverage of the treaty after it is exhausted. Applies to CATA treaties only.
maolAmountStringMaximum Any Oe Life Amount
isValidString
userId1String
userId2String
aggregateDeductibleStringFor aggregate deductibles, losses from multiple events over a treaty’s in-force period can contribute to eroding the deductible. Applies to HD models only.
aggregateLimitStringLimit that applies across all loss events during the treaty’s in-force period. Applies to HD models only.
uriString
lobsArrayList of lines of business.
lossOccurrencesArrayList of loss occurences. A treaty loss occurrence is an an event, a portion of an event, or a collection of events to which reinsurance terms are applied to determine the split of liability. A loss occurrence is represented as a one or more rows in a period loss table (PLT) where each unique combination of the properties periodId, eventId, and eventDate defines a loss occurrence. For each loss occurence, includes id, treatyId, uri, regionPeril, lossOccurrenceTime, lossOccurrenceRadius, radiusUnit, and multiLossOccurrence
analysisIdStringID of analysis.
tagIdsArrayList of tags to apply to this analysis treaty.