Fragmented Liquidity Hurts Marketplace
November 3, 2008
NEW YORK - The fragmented, free-for-all liquidity in the marketplace is aggravating the extreme volatility in the stock market, and experts say a more organized, specialized system is needed.
"We are going through a revolution in how we manage data," said Larry Tabb, founder and CEO of the Tabb Group research and advisory firm, speaking at a panel discussion at Nasdaq headquarters in Times Square, hosted by Interactive Data.
"The fragmentation of the market increasingly forces trading to be done electronically," he said. "The more electronic these orders are, the more they can be interchanged, plugged and played."
Only a few years ago, most trading was sent to exchange floors over the phone, Tabb said. Now, most orders are handled electronically by algorithms, operating in a plethora of individual dark pools.
While dark pools allow traders to move large numbers of shares without revealing themselves on the open market, the rapidly increasing opacity has cast a shadow over them.
"At the end of last year, only 37% of orders were handled by the sales desk, compared with 79% in 2005," Tabb said. "Machines are constantly hunting for trades. As machines take control of execution, you're ending up with a reallocation of order flow."
The result has been daily share volumes surging 10% to 20%, he said.
"Currently about 8% of liquidity is executed in dark pools," Tabb said, and that percentage is growing. "I think that 25% is as much as you can do in dark areas, but once we get there, I don't know if we'll be able to turn back."
Tabb said there are currently more than 50 dark pools, and even within individual exchanges, there are dark orders and multiple dark pools. So far, dark trading is mostly limited to equities, but experts expect it to expand into other asset classes, as well.
"Our customers are interested in getting access to dark pools," said John Partridge, co-founder and vice president of automated trading software designer StreamBase Systems. "They have a desire to bypass intermediaries and get directly to data. There will always be an appetite for looking at all of the data in its rawest form. Quants tell us next to nothing. These are very secretive hedge funds."
Partridge said his customers tell him that it is burdensome to parse trades out to multiple Alternative Trading Systems, adding that if the 68 ATSs were to consolidate and specialize, it could help to reduce the fragmentation. "There is a shifting investment in IT towards automation," Partridge said.
He said his firm relies on Moore's Law (computers will double their processing ability every two years) and Metcalfe's Law (the more users a network has, the more valuable it is to its users), but IT providers have been blindsided by the staggering speed at which this automation is occurring.
Companies plan for expansion, but not for an 800% surge in demand, Partridge said. An eight-fold increase in volume "is terrible for IT," he said.
"We are seeing message traffic we couldn't have predicted," Tabb said. "How do you deal with this?"
Noting that firms are suddenly being asked to handle two million messages per second, Partridge said such "data figures were unthinkable two years ago. It's 10 million messages per second next year that worries me. Change is our friend, but I'm not sure I want this level of change on an ongoing basis," he added.
In the struggle to keep up, firms are boosting their IT budgets at the expense of other areas.
"Cost management is an increasingly important issue," Tabb said. "As people cut back budgets, there will be a move from manual traders to automatic trading. Firms will continue spending on electronic trading and risk management, and folks who don't invest in electronics will lose business and become technologically disadvantaged."
Another growing problem is communications latency: the time it takes for information to travel over long distances. Automated trading commonly occurs in milliseconds (one thousandth of a second) and microseconds (one millionth of a second). While there are some inherent limitations, such as the speed of light, firms are looking for innovators who can keep this latency as low as possible.
"Who can offer low latency?" asked Ken Barnes, vice president of NYSE Euronext, adding that an ideal latency rate is currently about 20 microseconds.
"Many people don't need to be in this high-speed, low-latency environment," Tabb said. "It's important to understand who your market is and what they need. You should buy the right data for the right people." "The U.S. has the tightest spreads in the world, and also the most fragmentation," Barnes said. "We will always see a demand for innovation. There is plenty of technology out there and plenty of opportunity."