The scheduled implementation this June of the Securities and Exchange Commission's Regulation National Market System is causing brokerage firms to revisit an issue many thought was resolved years ago: the adequacy of their trading-data storage capacity.
Regulation NMS's order protection rule will require sell-side firms to identify immediately the trading venue with the best execution price and to route an order there. The rule is expected to encourage even greater use of automated trading platforms, electronic routing systems and algorithmic trading -- and to increase the already ballooning traffic of price quotes, corrections, messages and order cancellations.
Storing all this data for compliance and trading purposes has become "a very big issue" for securities firms, says New Yorkbased Philip Cushmaro, chief information officer for investment banking and the Americas at Credit Suisse. He notes that storage costs already "have increased dramatically."
The need for still more storage is a certainty. Robert Iati, a partner at Tabb Group, a Westborough, Massachusettsbased consulting firm, estimates that the volume of trade-related traffic will double every year through 2008. Between now and then, he contends, the securities industry must increase its storage capacity at a 100 percent compound annual rate.
"This is moving so fast that even the OPRA [Options Price Reporting Authority] and SIAC [Securities Industry Automation Corp.] can hardly keep up," says Iati, who recently wrote a report on data storage. "There are now 40 quotes per equity trade and 2,000 quotes per options trade." The consultant predicts that as order flow increases in electronic markets, data could multiply tenfold as a result of trading algorithms that cancel and replace orders as they look for liquidity.
Because every stock price change results in many more quotes for the underlying option, the options exchanges have experienced a data onslaught.
"Options-quote growth has almost doubled in each of the past five years," says Thomas Knorring, vice president of trade operations at the Chicago Board Options Exchange and current chairman of the Consolidated Tape Association, the exchange-created body that oversees collection, processing and distribution of market data. "The number of options messages transmitted hit 1 billion a day for the first time in October 2005," he says, "and has touched that level a couple of times since."
For securities firms, which must retain all pricing data for compliance purposes, storage capacity is not an insignificant cost or logistical concern. "But the issue isn't plain old storage anymore," says Iati. Lightning-fast accessibility to trading data is now equally important, he maintains, because the algorithms that can be updated first are likely to be the ones with the greatest chances of success. Explains Iati, "If you can move the data fast, you can make money."
Since conventional database software typically runs too slowly to update trading algorithms, specialized software vendors have emerged to access, analyze and store the enormous trading-data flow in real time. Vhayu Technologies, headquartered in Los Gatos, California, markets Vhayu Velocity, a Windows/Intel-based product used by global brokerages and hedge funds. In June, says John Coulter, vice president of marketing at Vhayu, the company plans to introduce a product capable of real-time options data analysis and storage. Simon Garland, chief technology officer at Kx Systems of Palo Alto, California, says his firm's Kdb+ high-speed database can use "commodity servers" and satisfy both trading and long-term retrieval needs. Other vendors include Progress Software Corp. of Bedford, Massachusetts, maker of the Progress Apama Algorithmic Trading Platform, and StreamBase Systems in nearby Lexington.
Customers will be watching the performance of brokerage systems closely. Michael Buek, a senior trader at the Valley Forge, Pennsylvaniabased Vanguard Group, says he will hold the sell side to a higher standard of best execution than Reg NMS.
"We're not going to examine every 100-share trade, but I'll be judging whether we get best execution that's deeper than top-of-the-book and looking to see whether algorithms are dynamically adapting to the market," Buek says. "I think we'll be getting more from brokers."