Getting That Single Version of the Truth on Wall Street
Amid the credit crunch and worries about improperly priced CDOs, best-execution requirements, and real or perceived risk management issues; and facing freshly aggressive audits from regulators and exchanges, firms all across The Street are pursuing data management overhauls. "In my 25 years at Accenture, I've never seen so much focus on data," reports Beth Roberts, managing director, North American capital markets at the consulting firm.
The pressure on securities firms to get their data in order is coming from all sides, Roberts notes. It goes without saying that traders and quants need to act on timely and accurate data -- and later be able to prove they met best execution requirements. And wealth management advisers also are under pressure to provide accurate gauges of their performance. "Customers may complain about whether or not they got the best price three years ago," Roberts explains. "Firms need to go off and locate that information and try to produce it."
To meet these data demands, Wall Street firms traditionally pursued a vision known as the "golden copy," in which a single, enterprisewide repository holds an authoritative version of every customer file, trade, quote and other scraps of relevant data. But the golden copy's image has been tarnished by disappointing early projects -- it has turned out to be too difficult (particularly as firms have merged and were acquired) to keep a golden copy up to date and too easy to make mistakes. As a result, most firms have abandoned the golden copy ideal as impractical (except for certain uses, such as the securities master file).
"There's no one technology or architecture for integrating data," notes Mark Clare, VP of data integration services at TIAA-CREF. "The large data warehouse vendors may tell you to load all your data in one warehouse because that's what enterprise data warehouses are designed for. That has its use, and we have that component in our architecture. But there are other times when you can leave your data where it is and virtually integrate."