Change is almost never without uncertainty. And major change—especially in the typically conservative world of investment banks—creates even greater uncertainty. So it should come as no surprise that Refinitiv’s plans for the next evolution of its TREP market data platform are being met with a mix of excitement and apprehension.

At its core, TREP is the ubiquitous platform—more than two decades old—that was created to solve the problem of getting market data from point A to point B. Think of TREP as the heart, lungs, and entire nervous system of any trading operation. When it functions properly, the body runs smoothly. But if it experiences problems, parts of the body are starved of oxygen—in this analogy, oxygen represents data—and don’t work properly. Even worse, trading on stale data could cause losses as the market moves but the data in the system remains stagnant.

There’s another reason this analogy represents TREP well: Trying to replace it would be a lengthy—not to mention risky—operation, to be performed only by teams of skilled doctors, probably involving multiple surgeries spread out over a long time to mitigate any dangers.

“You can’t just rip TREP out; it’s a much longer-term plan,” says one market data consultant who has worked directly on TREP at multiple end-user firms.

Now, to add to the difficulty, imagine that all those surgeons are constantly worrying about who owns and runs their hospital, while performing these intricate operations. Refinitiv’s two-year existence has been clouded by uncertainty around its ownership.

In 2018, Thomson Reuters sold its Financial & Risk division to a group of investors led by Blackstone Group. Under Blackstone, this division took shape as a completely separate entity, badged Refinitiv. But after around 18 months, Blackstone agreed to a deal to sell Refinitiv to the London Stock Exchange Group (LSEG). The deal has since been blessed by LSE shareholders—after a failed counter by the Hong Kong Exchanges and Clearing Ltd. (HKEx)—and is expected to close later this year. After a run of uncertainty, those data surgeons may be as worried about job security as they are about their patients.

And that’s a big part of the problem: Financial clients who rely on Refinitiv’s systems feel that the vendor has been distracted—more focused on a sale than on customer service. One data manager spoken to by WatersTechnology chronicles a litany of support incidents, while another grumbles about a lack of communication regarding the vendor’s roadmap for TREP. Both work at large financial firms with significant investments in—and exposure to—the platform.

Rival technology providers, spotting an opportunity, are fanning the flames of uncertainty—not necessarily outright scaremongering, but seizing any opportunity to promote themselves as long-term alternatives to the incumbent, whose origins date back to the 1990s-era TIB and Triarch platforms. But Refinitiv does have a roadmap, and is now working hard to communicate that TREP is not only safe, but is being developed to offer more flexibility and efficiency in the future. And to do that, the vendor is making a big bet on the cloud.

We are committed to TREP in a deployed fashion that everybody has used for more than 20 years, and we are looking at how to deploy it in our private managed cloud and in the public cloud. I think clients may hear about one of these and think we’re abandoning another—but that’s not the case.
 

Anthony Mazzocchi, Refinitiv

“Refinitiv’s focus point whenever we’re doing any forward-thinking is that we’re always looking at backwards compatibility … and ease of architecture moving forward. We have a vision of a unified platform, where—whether you’re a deployed customer, use managed services, or the public cloud—the connections will be transparent to you … with tiers of latency, content, and value-add,” says Anthony Mazzocchi, director of enterprise integration at Refinitiv, who joined the vendor last year after 30 years as a senior engineer on some of TREP’s forerunners and as a consultant who has worked on market data infrastructures at some of Wall Street’s largest financial firms.

“We are committed to TREP in a deployed fashion that everybody has used for more than 20 years, and we are looking at how to deploy it in our private managed cloud and in the public cloud,” he continues. “I think clients may hear about one of these and think we’re abandoning another—but that’s not the case.”

Next Steps

Indeed, of seven items on TREP’s roadmap for the first half of this year spread over versions 3.3.1 and 3.3.2, equal effort is spent on enhancing the core functions of deployed-TREP instances and on enabling a move to the cloud. Deployed enhancements include specific customer-driven changes, as well as support and tuning for new hardware, such as Intel’s Skylake processors and HP’s G10 servers. Meanwhile, cloud-supporting enhancements include the ability for hosted instances of ADH (Advanced Data Hub, TREP’s datafeed server) and ADS (Advanced Distribution Server) to access Refinitiv’s Elektron Real Time (ERT) Cloud, and use of VMWare virtual machines, and containerization, while threading will significantly increase throughput on hosted instances of ADS (ADS/POP) from 50,000 to 500,000 instruments.

“The public cloud has forced us to look at becoming more efficient with our deployments, to move outside our comfort zone of the deployed world, and to look at how we can be more effective in the world today. As we move into the public cloud, we’re hoping to offer something with more economic value to customers, with a lower footprint,” Mazzocchi says. “One thing we really need to nail down for customers before we can move [to the cloud] is latency. If we can offer a true end-to-end latency measurement through that public cloud, and be able to say to clients ‘You’re not just getting resiliency and redundancy; you’re also only adding five microseconds—not 10 seconds,’ then I think we’ll see some real adaptation there.”

The market data consultant also notes that Refinitiv’s plan doesn’t just make sense for clients seeking the next evolutionary leap for their data platform; it also makes a lot of economic sense for the vendor itself. “The more clients they can get onto a shared environment, the more cost-effective it is for them, and they can be much more profitable,” he says, adding that the vendor is “aggressively” pushing hosted TREP options—possibly as a way to sign up recurring revenue that will make it even more attractive to the LSE, or to build momentum behind the new services.

TREP is the beast of Wall Street. You need it. But they’ve lost a lot of good people and salespeople, and it’s hard to recover from that. There are not a lot of experts in this field.
 

Executive at a large US bank

It also makes sense from the ease of architecture philosophy that Mazzocchi describes, where, under a service-based model, clients could access real-time and reference data—and soon, delayed (i.e., fee-free) exchange data in the public cloud, as clients can already get from deployed and private cloud-hosted TREP instances—via a single API. In addition, a service-based cloud architecture means firms can do more offsite, and can reduce their hardware footprint by only taking the data they truly need into their own environments, rather than consuming all the data on the more than 84 million instruments covered by Refinitiv.

“I think our managed-services and hosted platform could be a sweet spot for us in the future because … we can very quickly deploy value-add services in that space. We can deploy even quicker in our private cloud than we can in the public cloud,” he says. “Part of the overall plan is about blending [delivery options] to make it easier to get the data, whether in a private or public cloud. I think what will drive people back to us is when it becomes too easy to refuse.”

Mazzocchi also notes that some of the vendor’s key strategic accounts may never move everything to the cloud. “They may be the ones who always stay on deployed technology, and that’s why there will always be a deployed TREP option,” Mazzocchi says. But he adds that others among Refinitiv’s lower and middle-tier clients are already set to make the move. “And as we start to add more value, they will start to see this as the way to go,” he says.

Walking a Wire

Refinitiv’s plan is ambitious, and any ambitious plan contains inherent risk. And it is end-user concerns about this risky element that other vendors are playing up, presenting themselves as safe options for the future.

For example, data vendor and low-latency data technology provider Activ Financial last year released a set of data integration tools, dubbed Enterprise Data Integration Suite (EDIS), citing end-user concerns over the long-term viability of legacy data platforms as a key driver behind the project.

“EDIS is a set of components designed to integrate with internal and third-party feeds and data platforms, and is aimed at people with exposure to incumbent platforms like [Refinitiv’s] TREP data platform, and recognizes that people are concerned about the technology roadmap and costs…and offers a path toward an alternative, based on Activ’s underpinning technology,” says Activ COO Jim Bomer.

EDIS is vendor-neutral, multi-tenant, and includes symbology cross-referencing to support other vendors’ data formats. It runs on Activ’s Activ One Platform (AOP) messaging middleware to provide a bridge between different technology platforms and data sources that can also cache, conflate, and delay data; connect to Activ’s global distribution network; and handle on-demand calculations.

Bomer says the idea for EDIS came around 2014, when Activ was figuring out how to optimize elements of its underlying infrastructure—which ultimately became AOP—and how it could best serve its clients in the future. “We asked ourselves: what is the key problem that people face when trying to de-risk their future,” he says. “We’ve often developed our systems at clients to interoperate with TREP, so we took that concept and looked ahead.”

New Entrants, Old Problems

Certainly, rivals are keen to seize on—and exploit—any perceived weakness on the part of TREP or Refinitiv. And if there’s one person who knows TREP’s weaknesses, it’s Bob Bonaguro, who is credited with being the primary architect of Refinitiv’s enterprise technology stack, having spent the past 18 years at Reuters and Thomson Reuters in a range of senior technology roles.

But he’s no longer at Refinitiv. Instead, he’s joined forces with Terry Roche, former global head of Enterprise Platform at Thomson Reuters, to set up Pegasus Enterprise Solutions. Pegasus offers a suite of APIs—available in C-sharp, C++, and Java—as well as a data viewer, a Microsoft Excel interface, and other tools that provide a neutral abstraction layer between firms’ TREP deployment, other transport layers, and client applications.

At Pegasus, Roche serves as CEO, while Bonaguro is CTO. The third co-founder and chief product officer, Brian Stephens, a former market data and technology exec at Royal Bank of Scotland and Dresdner Kleinwort, also spent almost six years at Reuters as a product support consultant for its RMDS and TIB data platforms (the forerunners of TREP).

Roche stresses that the vendor is not trying to displace Refinitiv and TREP, but rather is offering a solution to broader industry problems that leverages—and enables users to “extract maximum value” from—their existing investment in TREP. Indeed, the open “ecosystem” that Pegasus envisages relies on a multitude of components, including market data platforms—whether operated by Refinitiv or someone else.

“We’re not trying to replace TREP; we’re building software for the market to transform itself—and we think we can drastically accelerate that transformation at dramatically reduced cost,” he says, explaining that employing Pegasus’ APIs allows firms to create abstraction layers around their data platforms, opening the door to easier migration to open platforms in the future. Then, “if you want to switch to a different platform in the future, you will be able to do so with a simple configuration change,” Roche says.

One reason for this kind of time horizon is the sheer number of systems inextricably linked to TREP that must either be abstracted or rewritten. Another reason why a quick rip-and-replace isn’t practical is that other vendors simply can’t yet offer everything that TREP does.

Activ’s Bomer strikes a similar note, describing gradual migrations rather than a wholesale rip-and-replace of existing platforms.

“The big firms have very complex environments that have grown over the years. We’re not expecting that anyone will turn their back on their existing investment, but this gives them a way to support all their existing applications and migrate them gradually to a new platform,” he says. “Ultimately, we hope people will use this as … the new bedrock of their data architecture. No one is going to turn around and completely shift just like that, so this needs to integrate [with other platforms] for some time. You need to be thinking about your platform in five to 10 years’ time, and asking whether you are happy sitting on the same technology stack that you have now.”

One reason for this kind of time horizon is the sheer number of systems inextricably linked to TREP that must either be abstracted or rewritten. Another reason why a quick rip-and-replace isn’t practical is that other vendors simply can’t yet offer everything that TREP does.

“Firms all have folks looking at sources of risk, and have identified Refinitiv as a potential area of risk, so they’ve been given their marching orders to identify how risky it is, and where else they can find that data,” says Steven Roe, CEO of New York-based systems support and software provider West Highland Support Services.

Firms are enlisting alternative desktop and feed providers to mitigate their risk and dependency, but, “clients are all finding that there is no 100% replacement for Refinitiv’s content and technology. You can get 85% to 90% of the way there, but that still leaves you at least 10% to 15% dependent on Refinitiv. And the cost of doing all that is a big deal,” Roe says.

Mind the Gap

Since TREP’s origins extend back more than 20 years, while others are new to the enterprise platform space—even if they have been active in certain areas of data platform functionality—it’s not surprising that Refinitiv’s full lineup of functionality is tough to match like-for-like.

“We’re not trying to duplicate all of the esoteric features of a platform like TREP, but instead we’re focused on addressing the core functionality that everyone needs,” Bomer says, adding that firms can adopt elements of EDIS at their own pace to meet specific needs, then expand its use within their firms, integrating with their existing data sources, and transferring applications to run natively on the new platform.

The platforms offered by Activ, Pegasus, and others will become functionally richer over time—but, under Mazzocchi’s vision, so too will TREP. And while Activ’s 10-year view gives time for firms to migrate slowly, the current reality is that TREP leads the field. So the reason why abstraction layers and APIs are important is not only to potentially migrate away from TREP, but also to integrate multiple third-party systems that firms may need to combine to replicate the full capabilities of TREP.

Another such third-party system is the ONE cloud-based platform for consumption and distribution of real-time, historical, and reference data, and other content, such as news, provided by Frankfurt-based BCC Group International, which is working on two proof-of-concept projects to replace TREP.

“The financial markets are becoming more adventurous, and see the future as not one big platform, but many different platforms,” says BCC Group CEO and founder Mauricio Gonzalez Evans. BCC Group also provides an adapter, dubbed Elisa, to connect to TREP sites.

Gonzalez Evans says they were originally looking to integrate with existing TREP installations, but then they got the idea of replacing, using BCC Group’s ONE platform, while still connecting to Refinitiv’s content but via the cloud instead, or using the Elisa adapter to replace connections between applications and TREP with connections to other sources. The platform also allows users to publish and sell their own datasets, and BCC Group aims to provide a marketplace offering in partnership with Amazon Web Services (AWS), he adds.

Indeed, the big picture for data transport layers in the future, says Pegasus’ Roche, is not focusing on a single component with a firm’s data architecture, but interaction between multiple open platforms, using standards that eliminate vendor lock-in and mitigate users’ exposure to proprietary data platforms.

“With our software, you can create a consumer or publisher in hours, not months, and our dialog channels allow full peer-to-peer messaging through a firm’s entire environment,” Roche says. In addition, Pegasus is already developing gateways to brokers, as well as a broader data platform and an entitlements system to address this.

“We see this as going well beyond TREP and data, so anyone can deliver any service to anyone else in an efficient manner. Our end-goal is an open enterprise ecosystem whereby the market operates in a standardized manner for open interaction. What we’ve created so far is an important step toward that, which allows migration to open standards while getting more value from existing services,” Roche says. “The point is to remove the friction associated with providing a service and delivering data, because most of that now is under proprietary lock-in.”

And Pegasus isn’t just looking to replace lock-in to one vendor with a lock-in to its own services instead—Roche says the vendor will provide its source code to clients so they can maintain and develop the platforms internally, should they want.

“We don’t want to own or define standards. We will provide the technology foundation for the industry to do so in an open way. We don’t want clients to be locked into us any more than they are locked into anyone else, and thus we provide source code to our clients,” he says.

This isn’t the first time abstraction has been on the agenda—various groups have tried this approach in the past, with limited success—but Roche says the difference is a new sense of urgency as firms become uneasy about their current platforms and feeds. And a big part of what’s making people uneasy is, well, people.

Brain Drain?

Much of Pegasus’ value is tied up in the experience of its founders, and Roche views the united front of himself, Bonaguro, and Stephens as something of a coup. “If you’re a user of an enterprise platform, or an entitlements platform, it’s likely you’re using something we have built in the past,” he says.  “We’ve spent decades—not just years—understanding these solutions and tackling the needs of the market. We understand, because we’ve lived the lives of service providers and data and liquidity consumers.”

TREP is the beast of Wall Street. You need it. But they’ve lost a lot of good people and salespeople, and it’s hard to recover from that. There are not a lot of experts in this field.
 

Sell-Side Data Executive

This is another factor that has led to concern among consumers over Refinitiv. Because while Pegasus is touting its experienced staff, sources say Refinitiv has been losing its own expert sales and support staff, including a number of redundancies at its Oak Brook, Ill., office—widely viewed as the “brain trust” behind TREP.

“TREP is the beast of Wall Street. You need it. But they’ve lost a lot of good people and salespeople, and it’s hard to recover from that. There are not a lot of experts in this field,” says one executive at a large US bank.

Mazzocchi acknowledges some “consolidation” of roles at the location, but says this mostly accounted for instances where overlap existed between roles there and at its campus in St. Louis, MO, in areas such as quality assurance, and that the core team remains in place.

“We’re aware of our customers and their concerns. But the key decision-makers and the key people writing code have been retained. They were not let go,” Mazzocchi says. “We’ve trimmed some fat, not cut to the bone.”

And even some of Refinitiv’s rivals acknowledge that, whether or not some firms get cold feet, the future of multiple platforms and ecosystems they envisage is one that still incorporates TREP, and definitely still relies on Refinitiv’s data—that support for 84 million-plus instruments that is the “jewel in the crown” of the vendor’s assets, which would be tough for anyone else to replicate, Mazzocchi says.

“I think TREP will be around for the next 20 years, because it is doing a good job in many places,” says BCC’s Gonzalez Evans. “But once in the cloud, clients can cherry-pick. And vendors will have no choice but to compete on the quality of their data, rather than on selling technology.”

Certainly, TREP and its predecessors were never devised to lock clients in to Refinitiv’s data, but to solve the challenges of getting market data from A to B when similar data transport mechanisms simply didn’t exist. And almost certainly, if industry-standard, cloud-based technologies emerge that can perform that task more cost effectively, Refinitiv will also ultimately stand to gain from adopting those practices over time.

Clients’ concerns over “legacy” on site-deployed data platforms is really just exposing their reliance on old-school practices. Once firms are willing to commit to entertaining new approaches, all providers—no doubt including Refinitiv—will be ready to fulfill those needs.