A group of financial firms believes it has cracked the code, quite literally, on how to make usage rights for market data contracts machine readable.

The Rights Automation for Market Data Community Group—comprising the likes of the Chicago Mercantile Exchange (CME), Deutsche Bank, JP Morgan, Google Cloud, and Refinitiv—will at the end of this month unveil the first draft of its finance-specific language model for coding usage expressions in market data contracts.

The group has used the last weeks of June to put the finishing touches to the language ahead of its showcasing, says Jo Rabin, head of technology of the Innovation Network for the UK and Ireland at Deutsche Bank.

“We would say that most of the construction is done, but the decorators haven’t been in yet. It’s at the stage where you can definitely see that most of the bits are there, but there’s a little bit of polishing that is needed and there are definitely still some typos that needed ironing out,” says Rabin, who is also co-chair of the Rights Automation for Market Data Community Group.

A lot of the initial challenges related to fear, uncertainty, and doubt have been dispelled, and people realize that this is a possibility—this can technically be achieved. Yes, there’s a lot of nuance and details to be worked out, but it’s possible.

Ilya Slavin, JP Morgan

The project has involved more than a years’ worth of work. Since April 2020, over a dozen firms across the data supply chain—including suppliers, intermediaries, and users—have set out to develop a finance-specific language model to help simplify data licensing activities and automate the entitlement process. Today, the group includes 36 participants from 19 organizations.

In the past, these end-to-end data licensing ch­­allenges were deemed by many to be too daunting to tackle, says Ilya Slavin, who leads global market data engineering at JP Morgan. “A lot of the initial challenges related to fear, uncertainty, and doubt have been dispelled, and people realize that this is a possibility—this can technically be achieved. Yes, there’s a lot of nuance and details to be worked out, but it’s possible.”

The vehicle that underpins this initiative is the Open Digital Rights Language (ODRL), an extensible markup language (XML) and open-sourced data model used for coding policy expressions, created by the World Wide Web Consortium (W3C), an international community that develops open standards aimed at driving the growth of the web. The ODRL is already used to standardize the policy rights of published content for sectors such as telecommunications, online publishing, and television broadcasting. Now, building upon that work, the Rights Automation Group has used it to develop a finance-specific digital language.

In the early stages of the initiative, the group agreed upon a basic criterion for the first version of the language. Ben Whittam Smith, managing director of Deontic Data and co-chair of the group, says it wanted the language to be rich enough to express 20 of the most common market data licenses and answer between 60 and 70 of the most-frequently-asked questions that market data professionals grapple with on a day-to-day basis.

A coded language is nothing without technologies to support it, however. Slavin says the finance-specific ODRL will function as the glue between implementations and allow different technical solutions to interoperate—avoiding what he describes as a Tower of Babel effect. Once the first draft of the language has been released, a primary focus of the group over the next few months will be to build and fine-tune the various technologies designed to resolve key inefficiencies in the data licensing supply chain.

From theory to reality

The project is currently focusing on six proofs of concept (POCs), with two or more group members working on each one.

For example, in one of the POCs, members are zeroing in on the entitlement process for redistributing data—such as from a different part of the business or from a data aggregator to a user—looking at how to automate activities including consent, rights attribution, and disclaimer duties.

“With this case study, we wanted to see how the ODRL travels down the supply chain,” Whittam Smith says. “Three of us are working on that and [as part of that] we have had our computer systems talk to one another.”

We are hoping that over the course of the summer implementations will emerge, which will then go through proper interoperability testing. [The plan is that] by the end of the summer, going into the autumn, we have established inter-working between member group systems, whereby people are doing useful things with it.

Jo Rabin, Deutsche Bank

In another example, the group is developing exploratory tools for cross-referencing new use cases for datafeeds with existing licensed usage rights to check whether they comply with the contract or there is a need to repaper.

Additionally, in collaboration with two of the world’s biggest cloud providers—Amazon Web Services and Google Cloud—the group is also building technical solutions for automating rights management of data distributed and retrieved from exchanges via the cloud. Whittam Smith says the big tech firms have played an important role in driving these projects forward, in part by leveraging their technical brainpower.

Across some of the other POCs, the group is developing tools that can apply specific controls on how a user can access data, translate policy expressions in market data contracts into ODRL using machine learning, and, beyond that, generate machine-readable rights statements by scanning human-language rights information from a data repository.

The technical solutions are still in the early stages and the crucial next step for the project is to trial the quality of the integrations and their interoperability, Rabin says. “We are hoping that over the course of the summer implementations will emerge, which will then go through proper interoperability testing. [The plan is that] by the end of the summer, going into the autumn, we have established inter-working between member group systems, whereby people are doing useful things with it.”

Whittam Smith adds that before the calendar year ticks over, the group wants to release several ODRL-compliant tools for industry-wide consumption.

Elsewhere, there are several internal projects in development and production within the four walls of banks (see box: An inside job). But while some firms have been hammering away at their own systems, Whittam Smith says their efforts are better served by collaborating with the community group.

“I think the most important lesson I’m learning is that, because the solutions are so new and the space is so big, collaboration trumps competition at the moment. The more services that we can link together between different organizations, the better we are at supporting the entire supply chain,” he says.

Taking on time

“You may be compliant now, but were you compliant six months ago?”

This is a typical question that compliance teams and data vendors need to answer, whether on a real-time basis or in the event of an audit. Contracts are not permanent in nature; they have fixed timeframes in which they are active. So, to make the ODRL model applicable to data licensing, the Rights Automation Group and the W3C had to go back to the drawing board. First, they had to figure out a new way to embed temporality into the original ODRL, which was first created over 20 years ago.

“We went back to the main W3C community group that manages the ODRL standard and said, ‘We need to strengthen the ODRL so that it can support a temporal model.’ So, we are working with them to design that, alongside a lot of people out of universities and other industries,” Whittam Smith says.  

What we’ve done is create another profile of ODRL—a temporal profile—which allows us to embed all of that valid time metadata into the machine-readable rights statements. So, they describe not only what’s true today, but also what was true in the past, and potentially what will be true in the future.

Ben Whittam Smith, Deontic Data

To do this, the collective had to build an extension to the ODRL and code key concepts of a contract—such as permissions, duties, or an asset—as temporal objects, meaning each concept has an associated beginning and end date for when they are active. This added capability has the potential to ease the burden on compliance specialists and data vendors when trying to assess whether a datafeed was used in a compliant way at any point in time.  

“What we’ve done is create another profile of ODRL—a temporal profile—which allows us to embed all of that valid time metadata into the machine-readable rights statements. So, they describe not only what’s true today, but also what was true in the past, and potentially what will be true in the future,” Whittam Smith says.

This means the ODRL can also be used to model contractual updates from an exchange or data supplier. The temporal profile can capture the upcoming changes and provide visibility on when they will come into force. The group is currently in the process of implementing the updated ODRL version that supports temporal attributes into their POC tools.

The temporal issues have mostly been resolved, says Whittam Smith, but there is still an ongoing debate about ODRL version control. One argument is that the group could use a live document that will house up-to-date usage rights in written form, while other firms favor an elaborative version-management approach, whereby each firm manually updates the document as contractual changes come into force.

He says the group is engaged in conversations with the wider W3C members to figure out how they would create a living document “without breaking the emerging tools built on top.” This challenge also exists when it comes to updating systems to read new versions of the ODRL as the language matures and to reflect changes made by exchanges to market data contracts.

For now, says Whittam Smith, agents that consume the ODRL—meaning in-house technical solutions—need to be able to alert a human in front of the screen if they do not recognize the ODRL version or nuances in the language. “We have to be very clear over the competence of any particular agent [so that] it makes sure it only enforces or makes decisions when it completely understands what that decision is based on—and if it doesn’t, it has to have a human on the other side,” he says.

Rabin adds that members are aware that they are working on an unfinished standard.

“The people who have started to prototype this or develop systems that work with what we are doing will be fully aware that they’re working on an unfinished standard, and that anything they do is subject to correction or restatement in a different form,” he says.

Exchange appetite

As with any industry project, securing buy-in and adoption is crucial, and when it comes to market data licenses, a major piece of that puzzle is naturally the exchange community. In September, WatersTechnology spoke to JP Morgan’s Michelle Roberts about the significance of getting exchanges onboard. 

“I think everybody must participate. From the vendor and exchange side, ultimately, we want them to translate their contracts into ODRL, but until there are tools to generate ODRL, and the language is sufficiently rich to allow the nuances of market data to be captured, then it’s a bit of a chicken-and-egg situation,” she said.

As ODRL solutions continue to evolve, we’ll be collaborating with all of our data providers … to ensure that the rights standards drive opportunity for suppliers and consumers of data alike.

Steve Case, LSEG

To date, CME and the London Stock Exchange Group (LSEG) are the venues participating in the Rights Automation POCs. Steve Case, global head of data acquisition and rights management at LSEG, says both Refinitiv, a founding member of the initiative, and the exchange group are committed to the adoption of digital rights tools.

“As ODRL solutions continue to evolve, we’ll be collaborating with all of our data providers … to ensure that the rights standards drive opportunity for suppliers and consumers of data alike,” he adds.

CME did not respond to requests for comment.

Outside of CME and LSEG, it is understood that other exchanges have expressed interest in participating in the initiative but have adopted a wait-and-see approach until industry pressures force their hand.

But Whittam Smith says exchange participation isn’t the determining factor in the project’s success. Rather, he says, adoption will be driven by where the ODRL and associated solutions can present value, and that will be apparent in different parts of the data supply chain over time.

Additionally, he says, there are ways to alleviate concerns about a user inaccurately translating usage rights into ODRL that take inspiration from testing solutions. “Just because it’s gone into computer code doesn’t mean we can’t be extremely explicit about what that translation means, and we can borrow a technique from software called unit testing.”

In this context, data licenses could be treated like software, and a bank could write unit tests against the translations made by the ODRL-supported solutions. Taking that a step further, the bank could cross-check the translations not only with its compliance team, but also with the data supplier to see if it is comfortable with those interpretations of the usage rights.

However, the group has not started to ask exchanges whether they would accept unit testing as a way of approving the interpretation of licenses. This is yet another area that will need to be hashed out at a later stage of the project.

For now, Whittam Smith says, the main task is for the group to present the first iteration of the ODRL to the industry in June, with the aim of driving more adoption.

Some members of the Rights Automation Group will participate in an SIIA FISD-hosted webinar on June 24 to discuss their work so far.