Third-party software that can analyze market data consumption on the trading floor is reaping huge benefits for banks and brokerages, which have begun renegotiating contracts with top data providers like Reuters and Bloomberg. Some firms have cut data costs by 30 percent.
This technology breakthrough is particularly exciting for brokerages since market data costs often represent their third-highest bill after salaries and real estate. A single firm can spend as much as $100 million per year.
Former Reuters technologists-knowing that few brokers used all the services they got billed for-were the ones that first zeroed in on this market opportunity. Stuart Harvey and Steve Cowler, who had been with Reuters since the late 1980s, co-founded London-based Harco Technologies in 1996, and spearheaded the development of Harco's Data Access and Reporting Tool (DART), which was first used in 1999 to monitor Reuters' Triarch usage. That deployment was followed by versions that audited Reuters' TIB data platform in 2000, Bloomberg in 2002, and Reuters' latest, RMDS, this year.
Harco customer Bear Stearns has been a DART user since launching a firmwide "Efficiency Management Program" in 2001. The investment bank measures data services consumption among workstations and across products arrayed on Reuters' Triarch, the broker's "backbone" data platform.
Bear adopted DART based on a proposal by Peter Esler, the broker's global head of market data services. He recommended the system soon after joining the investment bank in August of 2000 after four years using it while global head of market data services for Perot Systems.
Esler culled optimization reports from a DART mechanism called "Product Best Fit" to compare usage of Reuters' Markets 2000 with NACAP, a less costly, more limited Reuters dataset, to determine which users were only, or mostly, using data fully covered by the cheaper system.
While he would not provide specifics on cost reductions, Esler says the resulting changes based on DART analysis has paid "multiple times over," and that DART has enabled the removal of unused IDs, converted real-time exchanges to static services, eliminated third-party commentary services from Triarch users, and transitioned "many" users from Markets 2000 to NACAP.
Other solutions traditionally packaged by the data vendors, like Reuters' DACS, MoneyLine's MTRS Observer, or TIBCO's Entitlements, have functioned primarily as permissioning checks to thwart shared or free usage and ensure vendors get paid when new users are added. In many cases, their functionality failed to inspire much use among customers.
"DACS reports are unwieldy and cumbersome to review; I mean lots of paper," Esler says. "When I came to Bear Stearns the business unit controllers didn't show much interest in reviewing 'un-used permission report' data from DACS, and I wasn't surprised based on my previous experience."
Most brokers at the time had only just begun to rationalize market data platforms. It took awhile for data executives to get through to business managers about accurately assessing and controlling systems usage. But banks are now getting more proactive, and other solutions are helping bridge the gap between broad-brush service contracts and true consumption; these include Duncannon Group's GLOW product, which is strictly focused on auditing Bloomberg, and modules from International Trading Room Systems (ITRS), which monitor both Bloomberg and Reuters.
Brian Slater, a Duncannon principal, said GLOW is focused on Bloomberg because Bloomberg's pricing and distribution "justifies our number-crunching and our development costs," whereas Thomson Financial, another big data player, does not.
Esler says DART for Triarch is designed to monitor usage of any feed over Triarch, including non-Reuters services such as Telerate; however, "Thomson does not presently reside on our Triarch backbone." Harco's Harvey said that his firm is "actively looking" at usage tracking and analysis extensions to DART on "several of these environments."
Bloomberg has proved a harder nut to crack for all the usage monitoring vendors since the information required for reverse-engineering remains locked inside the terminal, which Bloomberg controls. Whereas in the case of Triarch, "I control the entitlements down to the end user," Esler says. Bear, in fact, has paused from implementing "Bloomberg takedowns" recommended by DART, pending further analysis.
The vendors, it may come as no surprise, have been less enthusiastic about the technology than the brokers. The customer service strategy seems striking to some market observers, particularly as adoption of on-demand services and support grows in financial services IT overall.
The services of DART, GLOW and ITRS, for instance, seem a lot like IBM's utility computing initiative writ small. ITRS announced last month it was joining IBM's "Autonomic Computing" alliance, an effort involving automatic management of services based on usage, suggesting a broader strategy. Harco, Harvey reports, did its first deal directly with an outsourcer, which he wouldn't name, this year.
The Perot-UBS-DART link cited by Esler however, seems to have already set the precedent.
So then, how is Harco's relationship with Reuters and Bloomberg, the vendors the firm gets paid to watchdog, and arguably, nitpick? "Reasonably good I believe," Harvey says. "Many vendors are starting to see a competitive advantage in giving transparency on how their market data is actually used. [This is] the first year in which we've received interest from vendors themselves in connecting to DART."
Harco's primary development center is based in Belfast, Ireland, but the firm opened a New York office in 2002, and about 50 percent of its revenues come from North America. Other clients include Goldman Sachs, UBS Warburg, Royal Bank of Scotland, Societe Generale, Natexis Bleichroder and Banco Santander.
As for next steps, Harvey says Harco is "working with a number of clients" to track the internal redistribution of "fee liable data." The hope is to apply precise cost analysis to licensed information, which brokers add their research to and then redistribute internally and externally.
The tracking is part of a larger move many banks and brokers are making to build long-term or multi-year historical databases to better profile how market data terminals are used. Real-time exchange data from the NYSE, for instance, may not be so pertinent for a bond desk.
