The inside story of DerivHack and how the ISDA CDM might be the beginning of something big

You may not have heard of DerivHack, an event run by Barclays in London and New York on September 21^st and 22^nd. Multiple teams entered the event in both locations with sizes from one person to four, each team including technology and business experts. Teams were from technology firms with varying backgrounds, some being users of ledger technology and some being the builders and suppliers of the ledger platforms.

The goals of the event for Barclays were:

• Viability: Demonstrate ISDA CDM™ as a workable standard for post-trade processing
• Adoptability: Implement CDM as an event model in enterprise and DLT platforms to provide an integrated view of trade economics and business processes
• Opportunity: Develop a pathway for CDM based asset and functional infrastructures to enable optimisation across market infrastructure.

For background the ISDA CDM is a new project intended to move beyond FpML into a codified model of both the data and processes which represent OTC trades. For more background on the CDM visit https://www.isda.org/2018/06/05/isda-publishes-digital-iteration-of-the-common-domain-model/

Sunil Challa Director at Barclays and instigator of the event evaluated the outcome and said “CDM is proven to be viable for these use cases, of course CDM needs to be extended in scope and depth but the teams showed the concept to be sound. We saw that a wide variety of platforms both ledger based and conventional could adopt the CDM approach. And as a result. We see this as the beginning of the CDM opportunity to transform processing of OTC products in the long term.”

The teams were set a series of use cases, scenarios to be implemented with test data for trades and counterparties. The use cases included:

• Setting up counterparties on the platform
• Creating a new rate swap and credit default swap trades
• Trade events such as a notional change, 3 and 4 way novations, partial & full terminations, rate resets, accruals and fee payments
• Settlements with payment netting
• Reporting on trades for a specific trade date, or the life history of a single trade

Each team was expected to take the test data provided, and also a connection to a live simulated settlement service, and implement each use case to produce output results. A panel of judges met each team in private to review the results of each team and talk over their activity. The event began on Thursday morning, with the teams evicted from the venue at midnight that day – some teams used every minute available to make progress and had a short night’s sleep on Thursday. On Friday they resumed work leading to the judging sessions during the afternoon and a series of three minute pitches with a single visual slid, in front of the entire audience of attendees.

Many of the London teams based their activity on the Corda platform from R3, some used Hyperledger, a team from Digital Asset participated, and one team used Ethereum giving a wide range of technology platforms. One team even developed an Alexa skill enabling voice driven trade booking “Alexa book me a new rate swap with Barclays paying LIBOR flat against 3.2213%” – not what happened in reality, but this was a proof of concept showing the possibilities.

Teams entering in London included Xceptor with JDX and R3 (JRX), Digital Asset, TradeHeader, Derivite (which was really bluebank.io and Finteum), Natixis, ‘Team 7’ (which was really Barclays), Clause.io, ModelDriver, Rajeev Mavinkurve as an individual, iPushPull, Industria, and blk.io. In New York amongst 15 teams were Synswap, OpenRisk.io, Baton Systems, JP Morgan Chase, Chatham Financial and Cyptonomics.

Comments from the teams during the London three minute pitch sessions included:

• Storing trade data on a ledger brings the possibility of banks sharing a ‘golden source’ of data on a trade including cashflows. This would enable measuring exposure and credit events much easier.
• Managing data permissions on a ledger is a key problem. How do you control access to some or all of the data on a trade to third parties like a regulator or other service provider?
• Corporate actions in the equity market need more attention from regulators. They aren’t currently modelled in any language at all.
• Enriching the option payout sections and equity basket operations in the CDM is desirable.
• When based on Corda, some of the uses cases only required 50 lines of code
• The mapping between legal documentation and the data held about a trade is a disconnect and needs to be made closer
• Ledger platforms necessitate group activity, and that the Ethereum platform has a big developer community that could take up and support the CDM

Viv Diwakar Technical Sales Engineer at Digital Asset

The judging panel in London & New York picked out some of the firms for prizes, although all the firms pointed out that this event was a learning exercise for them, and winning wasn’t a specific goal. Recognition was given for these criteria:

London

• Best overall solution: Digital Asset. Digital Asset completed all the use cases plus an extended stretch goal
• Completeness: JRX (Xceptor, JDX, R3). The JRX team built CDM primitives which other teams used to complete their use cases
• Solution architecture: blk.io
• Best pitch: Industria

New York:

• Best overall solution: Baton Systems
• Completeness: JP Morgan Chase
• Solution Architecture: Chatham Financial
• Best pitch: Cryptonomics

Honourable mentions went to:

• TradeHeader for detailed and incisive feedback
• Derivite / Finteum for looking beyond the use cases for new applications for the CDM
• iPushPull: For a novel use of a secure messaging platform
• Rajeev Mavinkurve for individual effort as a one man team
• Synswap for the being the only team in New York that solved the privacy issue

The OTC Space spoke with Scott O'Malia CEO of ISDA at the event about the idea of making the ISDA CDM a product within a licensing company, in order to fund certification testing. Other feedback suggests that making the CDM a paid-for product might reduce take-up, the debate on this will continue with ISDA. Scott did confirm he would be providing feedback to the ISDA Board on the DerivHack event, and in his short speech to the audience described the event as 'terrific' with 'fabulous work' and 'having vision'. He explained that the idea of the CDM has been developed for the past two years, the purpose of which is to fix the problems of 'inefficiency' and 'bespoke data'. Scott felt that ISDA members need a fundamental standard to underpin their processing environments.

Brian Nolan Founder and CEO of Finteum said “I foresee that having a common data representation for products and events, similar to CDM, will be a valuable building block for markets outside derivatives.”

Mohammed Cherif COO at Synswap said “CDM isn’t necessarily tied to ledger technology and keeping the two apart reduces risk for taking the CDM forward. We found that the event enabled us to explore practical uses cases with other firms.”

Stuart McClymont, Managing Director at JDX said “It was a great event, not only did the ISDA CDM language hold up well throughout the use cases but there was a genuine collaborative environment between all participants to share their experiences and refine the language. Being awarded first place is a tremendous milestone for JDX as it shows how we as a firm have worked together across our service lines to lead the industry from a strategic concept to a technical solution in under 18 months. It will fundamentally transform the OTC derivatives industry post-trade landscape for years to come and puts JDX in pole position to work with industry participants on the journey to roll out the use of this new industry standard data language across their post-trade architecture”

Scott O'Malia, CEO ISDA

Nara Kodihalli CEO of OpenRisk.io said “We took part to explore how post-trade events could be implemented on a ledger platform. We found the experience valuable to see how ledger based software deployment could lead the way to high levels of automation between market users in future. We also demonstrated the semantic approach for extending the CDM models to a wider post-trade application such as Collateral Management.”

Richard Brown CTO of R3 had these thoughts:

• CDM is one of the first standardisation initiatives we’ve seen that applies the insight that true consensus between firms – the holy grail that enterprise DLT seeks to deliver – depends not only on common data and process models but also a common approach to verification of data and updates. There is more work to do but it’s definitely heading in the right direction.
• I was delighted by both the number and variety of teams who chose to build on the open source Corda blockchain at DerivHack 2018. Corda is an open platform upon which startups, established firms and other innovators can build new applications and it was wonderful to see so many of them do just that last week.
• I have long predicted that the world will rapidly consolidate down to a very small number of enterprise blockchain platforms and last week’s event provided strong evidence that it will indeed be the open source, openly governed such platforms that prevail. I was impressed by all eight of the teams (out of eleven presenting) who chose to build on Corda.
• Three projects that stood out in particular were those that pushed the boundaries of the technology. For example, several teams including that from Natixis showed how CDM-modelled trades in Corda could generate cashflows, which could then be netted and settled atomically by Corda across the same shared ledger. At the other end of the process, the team from clause.io impressed me in how they used Corda’s long-standing support for legal/technical interlock to store trades that linked to their underlying legal documents. But many other teams also impressed with their creativity, including Derivite, who showed, using an open source extension to Corda called Cordite, one possible future, where obligations arising from derivatives positions are settled with tokens that also lived on the same shared ledger.
• Finally, it would be great to see the licensing of the CDM model liberalised: just as the market’s consolidation onto Corda as the go-to platform for professional deployments has been enabled by Corda’s liberal open source license, it is likely that adoption of and interest in CDM amongst developers will significantly increase if/when the CDM license is similarly liberalised.

Viv Diwakar Technical Sales Engineer at Digital Asset in London and leader of their team said “Digital Asset fielded the only team to complete all 6 demanding ISDM CDM challenges, including a stretch goal, within the allotted time at DerivHack in London. This demonstrates how writing smart contracts in DAML is extremely productive as developers only need to focus on the business problem at hand, not the underlying complexity of distributed ledgers. We were delighted to have been named best overall solution in London as this was a challenging hackathon with complex subject matter and many competitive teams."

Petko Karamotche Director and Solution Architect at INDUSTRIA said: “DerivHack tested the use of the CDM with a real ledger platform (Corda). Some of the use cases, necessary for the event were new for us and two days were not enough to really understand the value of the CDM. We will re-run the hackathon within our own business as a training exercise to learn more about the CDM. The event allowed teams to test using different ledger platforms. DLT is increasingly becoming a more and more competitive space, so apart from the Corda teams congratulations to Digital Asset for their level of completion. The colleagues using the Ethereum ledger were also very competitive. Therefore it can be said that the CDM is design to be platform-agnostic. The hackathon gave the organisers feedback about the level of interest in CDM. Our recommendations to them was to provide even more documentation, more interactive training and deliver marketing about the model.”

Sunil Challa at Barclays also said: “The success of DerivHack means we will consider running the event again next year. We’d like to see uses cases including end-to-end workflows for rates and credit and exploring the dependencies within the CDM. We see the CDM as part of a new foundation for the OTC market with further functional layers above and below.”

My own take away from the event was this:

• The current state of the back office for OTC products needs to move in a new direction, becoming highly automated and efficient. Costs must be reduced to match the reduction in trading revenues at many banks.
• The ISDA CDM is at an early stage and will expand in scope and complexity towards wider coverage of products and lifecycle processes
• The CDM has the potential to become an industry standard reference point from which any firm using or implementing OTC products could benefit
• The application of the CDM using conventional technology (ignoring ledgers) is entirely feasible, and may be a first step to standardising back office platforms
• The development of ledger based tools to enable OTC back office processing will take longer, as this has a disruptive effect on everyone concerned
• The event was an excellent opportunity to test the CDM with a wide variety of platforms, showing that the concept is sound
• Events like it should be repeated and build towards bigger challenges

Do you think that the OTC market is destined for a ledger based future? Would the CDM and a ledger bring the increases in efficency and reductions in cost that firms need?