SPAN 2 Margin Approximation Service

Starting Monday, September 11, 2023, the Account Management Service for risk limit setting for CME ClearPort and CME Globex will be updated to align with the roll out of the SPAN 2 Margin framework. The risk limit methodologies will remain the same and will be enhanced to incorporate SPAN 2 risk data to compute portfolio requirements.  This topic details the SPAN 2 Approximation File and offering.

As part of the SPAN 2 Framework launch, CME Group is creating a new SPAN 2 approximation file that can be loaded into existing SPAN replication processes. The purpose of the SPAN 2 Approximation Service is to enable market participants that perform pre-trade margin checks with the ability to run their current margin programs while computing SPAN 2 approximated margin requirements, simply by initializing this new SPAN 2 approximation file.

This service will enable all low latency margin processes that run in under 50ms today the ability to compute SPAN 2 approximated margin requirements by initializing this new SPAN 2 approximation file. This time threshold is approximate and reflects the fastest time that precise margin calculation can be achieved using CME Group Margin services, thus this approximation service is designed for use cases.

The initial product launch will focus on a subset of NYMEX Energy products and will be followed by additional asset classes such as equities, interest rates and commodities in the years to come.

See more information regarding the SPAN 2 Framework and Functionality.  

Contents

Revision History

Date

Description

Date

Description

July 5, 2023

Initial publication of this topic.

Key Events and Dates

Below is the multi-year rollout schedule for completing the CME Group migration across all asset classes.

Date

Milestone

Date

Milestone

Q3 2023

Energy Products

H1 2024

Equity Products

H2 2024

Interest Rate and FX Products

H2 2025

Agriculture and Remaining Commodity Products

Testing and Certification

Firms will be able to use their existing margin processes today, while computing SPAN 2 approximated margin requirements as this new file will be based on the existing SPAN file format. The only work required for firms would be to pull down this new SPAN 2 approximation file off the SFTP location and load into their existing processes.

SPAN 2 Approximation File Publication Details 

The SPAN 2 approximation file will be located via SFTP. 

  • SFTP Location: cme/ftp/FIRMID/pub/SPAN/data/cme/span2approx

  • File Names: 

  •  

    • CME.approx.YYYYMMDD.c.pa2.zip

    • CME.approx.YYYYMMDD.c.cust.spn.zip

    • CME.approx.YYYYMMDD.c.cust.c21.spn.zip

    • [c] indicates the file is the ‘complete’ file with all contracts represented including those without open interest.

    • Files will be available for all risk observation cycles currently published by CME for SPAN and SPAN 2 methodologies. 

First time users requiring access to SFTP or users requiring entitlements to an existing SFTP username should contact posttradeservice@cmegroup.com to request access. Please provide your existing SFTP username, if necessary.

Impacted Products

See the complete list of initial products in scope.

Category

Example Products

Category

Example Products

Regular Futures - Seasonal & Non-Seasonal 

CL, NG, NN, RB, HO

Vanilla Options on Regular Futures – Seasonal & Non-Seasonal 

LO, LN, BZO, BE

Swaps and Spreads on Regular Futures 

CS, BK, GZ, RBB

Non-Standard Options 

AO, WA, BV, 9C

CME SPAN 2 Margin Tools

As part of the SPAN 2 project, CME Group has created a new set of hosted and deployable services for market participants to compute CME SPAN 2 margin requirements. These margin services are based on offering firms a scalable way for computing post trade margin calculation for intraday and end of day risk management and bookkeeping processes. After extensive testing and external validation, CME Group has determined these margin calculation services are best suited for post-trade execution risk management as overall performance for a single calculation may range from 50ms up to several seconds. While these solutions are horizontally scalable for users to achieve performant outcomes when running large numbers of portfolios and could be considered in the pre-trade execution space to accompany the CME SPAN 2 Approximation Service, they may not be the most performant.

Services for CME Markets



Market Participants





Market Participants



Deployable Services for CME Markets

Product

End Users

Clearing Members

Service Providers

Methodology

Testing Roadmap

Production Roadmap

Deployable Margin Software

All

X

X

X

SPAN and SPAN 2

Now Live

Now Live



Hosted Services for CME Markets















CME CORE API

All

X

X

X

SPAN and SPAN 2

Now Live

Now Live

CME CORE UI

All

X

X



SPAN and SPAN 2

Now Live

Now Live

The SPAN 2 Approximation Service offers firms in the pre-trade space the option to continue using their existing implementations of the SPAN methodology with the added benefit of the SPAN 2 risk arrays for SPAN 2 impacted products. This new service is based on the premise that an existing SPAN risk parameter file can be produced by CME which has been calibrated to account for the SPAN 2 margin model levels and transpose them into the SPAN methodology.

SPAN 2 Approximation Service Impacts

To align with the current risk management practices at CME Group, CME Group plans to generate both a PA2 and XML version of the SPAN 2 approximation file 7 times a day consistent with the existing process of generating SPAN files.

The new SPAN 2 approximation files will have a different filename and firms will be required to access the modified file through a new SFTP file path.  The scope of technical changes for firms to implement the SPAN 2 approximation service should be limited to sourcing the file.  

It is recommended that market participants evaluate the usage of this file in combination with both existing SPAN approximation services and the new set of SPAN 2 margin services for exact margin replication. For example, some firms may find that all pre-trade margin processes should convert to using the new SPAN 2 approximation file to achieve a reasonable level of predictability in margin requirements.  There may be other use cases where a firm wants to leverage more of a hybrid approach where they use the approximation file for their pre-trade checks but also want to update the portfolio level risk utilization using an exact replication of SPAN 2 immediately following the approximation check. This type of use case is an example whereby the user of the file will need to also consider additional workflows when adopting new SPAN 2 margin services. 

The SPAN 2 Approximation strategy uses a SPAN file for a given SPAN cycle as input to the calibration process.  The SPAN file has defined spreads and outright margin parameters which are inputs to the approximation calibration process where CME Group creates portfolios representative of each of these defined products or strategies and proceeds to generate SPAN 2 risk requirements for each one.   

With SPAN 2 risk requirements now available for a given set of representative SPAN portfolios, risk arrays are regenerated which are calibrated to be inclusive of the SPAN 2 risk requirements.  From here, the SPAN file parameters for products outside the approximation model are copied over to the SPAN approximation file enabling a full data set of risk arrays to generate a SPAN 2 approximation file that is suitable for computing SPAN 2 risk requirements (inclusive of SPAN margin requirements).  

This process is dependent on several factors that can influence the accuracy of the margin requirements:

  • All SPAN spread strategies that are defined in the file for applicable SPAN 2 products (Energy for example) are inputs to the calibration process. Since the SPAN strategies are predefined, there are often less strategies for the SPAN 2 approximation process to use as calibration inputs relative to the number of theoretical offsets that may be formed within a portfolio of similar products under the SPAN 2 methodology. Conversely, there are 3rd and 4th order non-trivial SPAN strategies that will not be part of the SPAN 2 offset structure. These are SPAN custom spreads that perhaps do not represent a correlation offset, but are useful to market makers to describe common index baskets such as BCOM and GCSI, idiosyncratic baskets with a business purpose or specific motivation/BTCI (Basis Trade at Index Close) and other custom baskets.

  • Some of the products that are subject to the modified split allocation process within SPAN now split between both SPAN and SPAN 2 methodologies. The SPAN 2 model treats the split products differently than the SPAN 2 approximation process.

  • In the case of Futures/Options on spreads/‘combo’ products, where at least one leg resides within the SPAN 2 scope whilst at least one other leg is within the SPAN universe, the span approximation service may not reproduce SPAN 2 accurately. This is due to the SPAN model margining the spread/’combo’ product by ‘splitting’ the spread/’combo’ product into its underlying before applying intra/inter commodity spreads. In this instance, one leg will benefit from the approximation service whilst the other will go through SPAN normally and will not be approximated.

  • SPAN 2 margins are asymmetric for long or short positions and SPAN 2 approximation is symmetric.

  • Short option minimum charge is taken from SPAN methodology to inform the SPAN 2 approximation. 

  • SPAN scenarios 15 and 16 overshooting the SPAN 2 requirements.

  • SPAN super scanning spreads are not an input to the calibration process. The motivation behind a SPAN super scanning spread is to collapse exposures to tightly related products into one master product within a sector. As a result, this reduction in outright delta remains in place in the newly generated SPAN 2 approximation file (SPAN 2 will not explicitly reduce aggregate exposure in the same manner).

  • SPAN 2 Liquidity and concentration components cannot be replicated by the SPAN 2 approximation service; however the liquidity component is incorporated into the calibration process. The concentration component is non-linear and is not replicable by approximation.

Analysis has shown that the contributions to the variance between SPAN 2 approximation and SPAN 2 margin is predominantly driven by the long-short asymmetry in SPAN 2. The table below shows long and short SPAN 2 margins vs SPAN 2 approximated Price Scans. This illustrates the systematic first order variance that is created when the average approach used in the SPAN 2 approximation is contrasted to SPAN 2 differences in long and short outright margins:

NG and CL as of March 25, 2022:

The error described above is propagated further downstream to the approximation of computed strategies (intra and inter spreads), since Price Scans are used to define SPAN2 approximated intra and inter spreads. We find that across CGM accounts, error from symmetric Price Scan on calendar and offset risk has a 2nd order impact.

CME Group may continue to refine the SPAN 2 approximation framework and will make market participants aware of any changes that solve for some of the challenges outlined above that impact portfolio accuracy.

Most portfolios we have analyzed show low levels of variance but the overall range of variance can exceed 100%. 

The accuracy of SPAN 2 approximation service is dependent on the portfolio and firms who are interested need to test the service using a wide range of portfolios to ensure computing margins using the SPAN 2 approximation risk parameter files through the SPAN methodology meets their business needs. 

Contact Information

For questions or support, contact PostTradeServices@cmegroup.com.




How was your Client Systems Wiki Experience? Submit Feedback

Copyright © 2024 CME Group Inc. All rights reserved.