Cloud Prisma Margin Estimator (CPME)

Cloud Prisma Margin Estimator (CPME) calculates portfolio margin according to the Eurex Prisma methodology. The application is available to both members and non-members of Eurex Clearing, a part of the Deutsche Boerse Group. It can be used via a browser based GUI or an API.

This GitHub repository contains documentation, examples and frequently asked questions.

Key links:

Known Limitations

Frequently Asked Questions

Comparing Eurex Clearing Margin Simulation Tools

Summary Table

feature / tool CPME API CPME GUI Online Margin Calculator
Availability Everyone Everyone Disclosed clients
Asset classes covered:      
- ETD yes yes yes
- OTC IRS yes yes yes
- OTC inflation yes yes yes
- Cross-margining (ETD/OTC IRS) yes yes yes
- Single ISIN Repo yes yes no
- Equities, bonds no no no
Input formats covered:      
- Member reports CB202, CP005 CB202, CP005 CB202
- CSV files yes yes yes
- OTC sensitivities (CSV) yes yes yes
- OTC sensitivities (CC233) yes yes no
- OTC FpML yes yes no
- OTC shorthand yes yes no
Historical calculation yes yes yes
Intraday calculation yes yes yes
Local installation possible no no no
Margin accuracy - ETD production-like production-like production-like
Margin accuracy - OTC IRS approximate approximate production-like
Margin accuracy - Repo approximate approximate n/a

What is the difference between the CPME GUI and the API?

The GUI is a browser front-end to the API, but not every API feature is provided by the GUI.

When is CPME available?

There is no time restriction set, in general CPME (both GUI and API) is available 24 hours a day, however there can be short (several minutes) intervals when instance for new snapshot is starting and requests return error, in that case users have to repeat the request. Unfortunately sometimes we can have also unexpected downtimes – CPME does not have the same level of monitoring and support as our main Risk systems.

When is end-of-day or start-of-day snapshot available?

End-of-day snapshot becomes available usually before 1am, start-of-day for OTC is received in the morning, around 6am. These are usual times, actual time varies each day, CPME is dependent on production Risk systems (PRISMA, Calypso) and any delay in their snapshot processing means delay also in CPME.

What is the performance of CPME?

CPME API respondes to 95% of requests in less than one second.

Performance issues can arise:

Submitting a calculation request to API

Why am I getting 401 error?

Most likely you are not subscribed to the correct API in DBG Digital Business Platform:

or there is a typo in your API key.

How can I perform a what-if analysis?

The user may load portfolios in the API in several formats concurrently. The user can for example

Alternatively the user can prepare baseline and what-if portfolios in a single format and submit these individually.

What to do when a request fails for a big portfolio?

Especially portfolios in XML formats can be quite large which may cause connection problems. Try compressing the portfolio before submitting, see “Compressing request and response” in the API definition.

Can I margin flexible instruments?

Eurex Clearing clears so-called flexible futures and options contracts with non-standard expiry, strike or exercise style.

CPME can calculate margins on flexible contracts provided they already exist on the Eurex market. Hypothetical flexible contracts (eg: of client’s own making) cannot be processed by CPME.

Is it possible to calculate margin as-of a historical date?

Yes:

Historical requests can take several minutes to perform. A request might be refused if the CPME calculation cluster is busy as there is an upper limit on the number of concurrent global historical calculations. In case your request historical request is rejected try again in 15 minutes.

An upper limit has been imposed on the number of concurrent historical calculations across all users, so a request might be refused if the CPME calculation cluster is busy. In this case you are advised to try again in 15 minutes.

What is the difference between API versions 1.0 and 2.0?

Version 2.0 is the recommended version, it is superset of version 1.0. It has slightly more complex estimator request allowing the user to submit multiple portfolio components of the same type, e.g. two parts of OTC portfolio both in CSV format.

New features are implemented only in version 2.0, such as Single-ISIN Repo. You do not have to commit to one version - an application may use resources from both 1.0 and 2.0.

How to submit CSV or XML files?

The portfolio in CSV and XML formats is submitted as one string attribute containing all the trades or positions. Please be careful about special characters:

This is an example of body for minimal estimator request with ETD CSV, API version 2.0:

{
  "portfolio_components": [
    {
      "type": "etd_csv",
      "etd_csv": {
        "csv":"Product ID,Contract Date,Call Put Flag,Exercise Price,Version Number,Net LS Balance\nFEXD,20321217,,,,1"
      }
    }
  ]
}

Another example shows how XML can be submitted using the curl command (replace “$KEY” by your API key). Note that the single quotes inside XML are escaped because we need to enclose the JSON in single quotes on the command line. This is not required inside application or script.

curl https://risk.developer.deutsche-boerse.com/prisma-margin-estimator-2-0-0/estimator \
 -H X-DBP-APIKEY:$KEY -H "Content-Type: application/json" \
 -d '{"portfolio_components":[{ "type":"otc_cb202", "otc_cb202": {"account": "P", "xml": "<?xml version='"'"'1.0'"'"' encoding='"'"'UTF-8'"'"'?><Report xmlns:fpml='"'"'http://www.fpml.org/FpML-5/confirmation'"'"' xmlns='"'"'http://www.eurexchange.com/EurexIRSFullInventoryReport'"'"' name='"'"'CB202 Full Inventory Report'"'"'><rptHdr><exchNam>ECAG</exchNam><rptCod>RPTCB202</rptCod><rptNam>Full Inventory Report</rptNam><membLglNam>Some Member AG</membLglNam><membId>XYZFR</membId><rptPrntEffDat>2018-03-28T23:59:00</rptPrntEffDat><rptPrntRunDat>2018-03-27</rptPrntRunDat></rptHdr> <reportNameGrp> <CM> <rptSubHdr> <membLglNam>XYZFR &amp; Bank &quot; Inter &lt; n &gt; Germany &apos;</membLglNam> <membId>XYZFR</membId> <globalLEId>XYZFRXYZFRXYZFRXYZ85</globalLEId> </rptSubHdr> <acctTypGrp name='"'"'P'"'"'> <ProductType name='"'"'Swap'"'"'> <currTypCod value='"'"'EUR'"'"'> <rateIndex name='"'"'EURIBOR'"'"'> <rateIndexTenor name='"'"'6M'"'"'> <idxSource>EURIBOR01</idxSource> <CCPTradeId id='"'"'1234567'"'"'> <fpml:dataDocument fpmlVersion='"'"'5-0'"'"'> <fpml:trade> <fpml:tradeHeader> <fpml:partyTradeIdentifier> <fpml:partyReference href='"'"'CPTY_1234567'"'"'/> <fpml:tradeId tradeIdScheme='"'"''"'"'/> </fpml:partyTradeIdentifier> <fpml:partyTradeIdentifier> <fpml:partyReference href='"'"'PO_1234567'"'"'/> <fpml:accountReference href='"'"'ACC_1234567'"'"'/> <fpml:tradeId tradeIdScheme='"'"''"'"'>12345678-2</fpml:tradeId> <fpml:versionedTradeId> <fpml:tradeId tradeIdScheme='"'"''"'"'>1234567</fpml:tradeId> <fpml:version>2</fpml:version> </fpml:versionedTradeId> </fpml:partyTradeIdentifier> <fpml:tradeDate>2017-10-17</fpml:tradeDate> <fpml:clearedDate>2017-10-17</fpml:clearedDate> </fpml:tradeHeader> <fpml:fra> <fpml:buyerPartyReference href='"'"'PO_1234567'"'"'/> <fpml:sellerPartyReference href='"'"'CPTY_1234567'"'"'/> <fpml:adjustedEffectiveDate id='"'"'resetDate_1234567'"'"'>2018-10-19</fpml:adjustedEffectiveDate> <fpml:adjustedTerminationDate>2047-04-20</fpml:adjustedTerminationDate> <fpml:paymentDate> <fpml:unadjustedDate>2018-10-19</fpml:unadjustedDate> <fpml:dateAdjustments> <fpml:businessDayConvention>MODFOLLOWING</fpml:businessDayConvention> <fpml:businessCenters> <fpml:businessCenter>CHZU</fpml:businessCenter> </fpml:businessCenters> </fpml:dateAdjustments> </fpml:paymentDate> <fpml:fixingDateOffset> <fpml:periodMultiplier>-2</fpml:periodMultiplier> <fpml:period>D</fpml:period> <fpml:dayType>Business</fpml:dayType> <fpml:businessDayConvention>NONE</fpml:businessDayConvention> <fpml:businessCenters> <fpml:businessCenter>GBLO</fpml:businessCenter> </fpml:businessCenters> <fpml:dateRelativeTo href='"'"'resetDate_1234567'"'"'/> </fpml:fixingDateOffset> <fpml:dayCountFraction>ACT/360</fpml:dayCountFraction> <fpml:calculationPeriodNumberOfDays>32</fpml:calculationPeriodNumberOfDays> <fpml:notional> <fpml:currency>EUR</fpml:currency> <fpml:amount>12000000.00</fpml:amount> </fpml:notional> <fpml:fixedRate>0.03000000</fpml:fixedRate> <fpml:floatingRateIndex>EUR-EURIBOR-Reuters</fpml:floatingRateIndex> <fpml:indexTenor> <fpml:periodMultiplier>6</fpml:periodMultiplier> <fpml:period>M</fpml:period> </fpml:indexTenor> <fpml:fraDiscounting>ISDA</fpml:fraDiscounting> </fpml:fra> <fpml:documentation> <fpml:masterAgreement> <fpml:masterAgreementType masterAgreementTypeScheme='"'"''"'"'>ISDA</fpml:masterAgreementType> </fpml:masterAgreement> <fpml:contractualDefinitions>ISDA2006</fpml:contractualDefinitions> </fpml:documentation> </fpml:trade> <fpml:party id='"'"'CPTY_1234567'"'"'> <fpml:partyId>EUREX</fpml:partyId> <fpml:partyName>EUREX</fpml:partyName> </fpml:party> <fpml:party id='"'"'PO_1234567'"'"'> <fpml:partyId>XYZFR</fpml:partyId> <fpml:partyName>XYZFR &amp; Bank &quot; Inter &lt; n &gt; Germany &apos;</fpml:partyName> </fpml:party> <fpml:account id='"'"'ACC_1234567'"'"'> <fpml:accountId>XYZFR_P</fpml:accountId> <fpml:accountBeneficiary href='"'"'PO_1234567'"'"'/> </fpml:account> </fpml:dataDocument> <novDateTime>2017-10-17 16:44:55.646</novDateTime> <CCPTrdVersion>2</CCPTrdVersion> <CCPStatus>VERIFIED</CCPStatus> <preSrcSysTradeId>12345678</preSrcSysTradeId> <srcSysTradeId>12345678</srcSysTradeId> <srcSysGroupId>12345678</srcSysGroupId> <srcSysLEId>EEU_FUND1</srcSysLEId> <tradeRefID>12345678-2</tradeRefID> <oldtradeRefID>12345678-2</oldtradeRefID> <srcSysId>MarkitWire</srcSysId> <UTI>O00000000000001234567C</UTI> <UTIIssuer>E01XXXXECAG</UTIIssuer> <priorUTI>MARKITWIRE0000000000000012345678</priorUTI> <priorUTIIssuer>1234567890</priorUTIIssuer> <USI>O00000000000001234567C</USI> <USIIssuer>1050000007</USIIssuer> <priorUSI>MARKITWIRE0000000000000012345678</priorUSI> <priorUSIIssuer>1012345689</priorUSIIssuer> </CCPTradeId> </rateIndexTenor> </rateIndex> </currTypCod> </ProductType> </acctTypGrp> </CM> </reportNameGrp></Report>"}}]}'

You can also submit queries to the API with a programming language, e.g. Python. The API documentation contains an example in Python to get you started. Python method json.dumps() takes care of writing valid JSON even when the attributes contain double quotes or line breaks.

Errors when specifying ETD positions by Maturity

The “maturity” attribute is deprecated, please specify the contract by “contract_date”.

One of the symptoms that request failed because “maturity” was used is the following error message:

Line 1 contract business key or contract id isn't related to a contract; contract not found within flex options

Interpreting a margin result

What is the precision of the margin calculation?

For exchange traded derivatives, the margin is calculated by code identical to the one used in the production environment at Eurex Clearing. The result on liquidation group level should therefore be exactly the same as in the CC050 report, if calculated for the same date and time.

For OTC trades, the margin is calculated in an independent implementation of the margin methodology and may differ a few percent from official EurexOTC margin requirements.

What are liquidation groups and liquidation group splits?

Products are placed in liquidation groups depending on their characteristics, for instance in the listed equity liquidation group or the fixed income liquidation group. A list of liquidation group and the products that are assigned to them can be found at Eurex Clearing’s Risk Parameters page.

A liquidation group can be further divided into liquidation group splits. This is currently only the case for the fixed income liquidation group PFI01, which is divided into separate liquidation group splits for OTC derivatives and listed fixed income/money market products, respectively.

For a given portfolio, initial margin is calculated independently at liquidation group level and is then added to give total initial margin. Liquidation group splits are also treated independently.

How does cross-margining work?

Listed fixed income/money market positions and OTC IRS derivatives belong to the PFI01 liquidation group.

Listed FI/MM products are placed in the liquidation group split PFI01_HP2, meaning they are margined under the assumption of a two-day holding period. Similarly, OTC derivatives are assumed a five-day holding period and are assigned to the liquidation group split PFI01_HP5.

Initial margin for PFI01 portfolios is calculated independently for the liquidation group splits and are added to get total initial margin. When cross-margining is turned on, an algorithm checks will check if any of the listed FI/MM positions are economically sensible hedges to the OTC portfolio and will move these to the five-day split if doing so lowers the total margin.

How to validate CPME result against Eurex reports?

Compare the margin on liquidation group level against the CC050 report. The margin on account level (CC060) cannot be obtained from CPME as the account may contain cash market instruments that are currently not supported by CPME.

Is the margin from the simulation environment similar to production?

No.

The simulation environment has different scenarios and prices compared to production, and therefore produces a different margin for the same portfolio.

Occasionally, we set up a what-if sandbox with production market data but with different risk model parametrization, as a testing ground before the new setup becomes productive. This will be announced by the usual channels.

Security

Where is CPME running?

CPME is deployed in Deutsche Boerse VPC (Virtual Private Cloud) on datacentres of Google Cloud Platform within EU.

How is the communication designed, is it encrypted?

In essence CPME has a client-server architecture.

Client is either your browser or whatever you are making direct API calls from. Server is the backend accepting API calls. All communication between the backend and clients are running over HTTPS.

Are the requests or results archived?

On the backend side all calculations are done in memory only. No records are kept of the information processed.

How do you ensure availability and disaster recovery?

CPME is stateless, has a highly elastic architecture and fault tolerant setup.

It runs in a highly-available Kubernetes cluster that is spread across the cloud provider availability zones in EU region.

Which countries does client data pass?

CPME runs on datacentres of GCP within EU, so all processing takes part within the European Union.

Repo

Calculation of margin for single-ISIN (special) Repo was introduced October 2021, both for GUI and API.

How to submit Repo portfolio in GUI?

Please see GUI User Guide, sections Upload Repo portfolio and Enter Repo portfolio.

How to submit Repo portfolio to API?

The Repo portfolio is part of estimator request as repo_json structure. The attributes correspond to columns in F7 portfolio export. Open-ended repo trades and floating rate repo trades are excluded from margin calculation.

How to read Repo margin from API?

The response to estimator request contains rbm_margin structure. That contains: