-- =DJ 'Trading Algorithms' No Longer Just Jargon On Wall Street --

By Mohammed Hadi
Of DOW JONES NEWSWIRES

NEW YORK (Dow Jones)--Trading algorithms, long the domain of brokerage firms,
are coming soon to a money manager near you.
Not to be confused with trading models, which help managers identify what to
trade, trading algorithms are software programs that mimic human traders by
taking orders and executing them according to a managers' goals.
These programs aim to improve execution by taking into account the most minute
changes in the market, breaking trades into tiny pieces and seeking out
liquidity wherever it exists on the open market, all at a much faster speed than
a human trader and with relative anonymity.
Trading algorithms programs aren't new. Brokerage firms have been using them
for several years on behalf of their clients. But now, the software is being
redesigned and repackaged for fund managers or their traders. One expert
compared the new version to travel Web sites. They are still powered by the same
kind of technology, but with an interface that practically anyone can use.
Large brokerages like Bank of America Securities and Lehman Brothers have
established what are essentially in-house software development firms that
produce the repackaged algorithms. Meanwhile, others like Algorithm Trading
Solutions, a boutique firm formed about a year ago solely to develop algorithms
for fund managers, are offering up some competition.
Still, when it comes money managers directly using these programs, it's early
going and fund managers who do use the software say that it isn't foolproof. It
requires oversight and, if improperly used, can end up costing a fund more in
poor execution than it saves in low commissions.
Putnam Investments, the mutual-fund giant that manages about $200 billion in
assets, has used algorithms for about two years, said Rich Block, director of
Global Equity Trading. The software, "allows us to maximize the effectiveness of
trading a liquid order, at a lower commission rate, and at excellent execution
levels and have the trader spend a greater percentage of his time working harder
to trade names," he explained.
However, for Block, it's really all about saving money; with that in mind, the
software is most effective when used to trade liquid stocks. On such trades, "a
sales trader won't be able to add significant value," so Putnam's traders will
use an algorithm. Putnam currently uses the software for about 10% to 15% of its
trades.
As for the broader asset management community, "basically we are in an early
adoption and experimentation phase," said Rob Flatley, managing director at Banc
of America Securities' electronic trading services group. Flatley estimates that
only about 5% of trades placed by money managers are executed on the new
in-house algorithms.
But, "we expect a doubling of order flow each year into these tools over the
next three years," so that by 2007, 20% of hedge fund and mutual fund trading
will be conducted using an in-house algorithm, he added.
David Cushing, managing director of execution services analytics at Lehman
Brothers, thinks one growth driver is the rapid increase of hedge funds. He said
the software appeals to this group because it can handle the mundane trades,
allowing a funds trader to focus on more difficult stocks, and it provides a
degree of anonymity because by breaking the order up into tiny pieces, it keeps
the market from sensing that a large order is brewing.
Also, "if you are a hedge-fund manager and you are gathering assets at light
speed, you could basically grow five times and not have to hire another trader,"
Banc of America's Flatley said.
As far as costs go, Banc of America charges about one-and-a-half to two cents
a share for trades placed through its algorithms, depending on various factors,
Flatley said. That's below the three cents typically charged for a human
trader's services.
This means the big brokerages are likely eating into their own trading
commissions by offering trading algorithms directly to fund managers. But they
don't have a choice, because competition for trading algorithms is getting
intense. "I think it's pretty clear that this was going to grow with or without
us," said Cushing.
That's partly because the big brokers aren't the only ones offering the
algorithms to fund managers. George Rodriguez is a managing director at
Algorithm Trading Solutions and is a trading algorithm bull. He thinks the
technology will become the primary form of stock trading among institutions,
although he does note that traders can't be replaced completely. His firm only
offers algorithmic trading services to clients, although it can "fill in the
blanks" when an order cannot be completed by the software.
There is evidence of the increasing use of algorithms.
For example, trading algorithms increase message traffic on the exchanges as
they adjust and readjust orders. According to one measure provided by Nasdaq,
message traffic has doubled since about a year ago and is up more than three
fold since the beginning of 2004, which the exchange said is a sign of the
increased use of algorithms.
Plus, trade sizes are getting smaller and smaller, which is a mark of the
trading algorithms. For example, average execution size for algorithmically
placed trades at Banc of America is down to 180 shares, Flatley said. That's
from orders of several thousand shares.
But Lehman's Cushing noted the drop in trade sizes is both a reason for the
increased use of algorithms, as well as a result of it. "The decreasing trade
size was primarily kicked off by decimalization and algorithms evolved to fill
the need for trading in smaller sizes."
Also, the software is likely one factor contributing to the rise in program
trading. Already, "a significant proportion of program trading, definitely a
non-trivial percentage gets ultimately carried out via an algorithm," Cushing
estimates. Program trading already typically accounts for more than 50% of
trading on the New York Stock Exchange, and that figure is bound to climb as
more fund managers trade stocks in baskets, because trading algorithms allows
them to do so with greater ease. "A lot of the brokers that offer algorithms
today, are starting to offer algorithms for programs," Flatley said.
Even if the use of algorithms doesn't reach the proportions that Rodriguez
expects, increased use of the software is going to change a stock traders' role
in many transactions. Sales traders at brokerages "will become more of a
consultant or a coach," Flatley said, advising clients on how trades can be best
executed and which algorithm to use depending on market conditions.
Still managers have much to consider when deciding which algorithms to employ,
or whether to use them at all.
Analytic Investors, a manager of more than $7 billion in assets, has been
using algorithms for about three years. The firm only invests in large-cap
securities in developed markets, "which is particularly useful for these types
of algorithms because that's their strength," said portfolio manager Dennis
Bein.
Analytic Investors' use of algorithms highlights the highs and lows of the
tool. When Analytic Investors first started using algorithms, it executed about
half of its trades using this system. Today, algorithms account for less than
10% of the firm's trades. Bein said the sharply pared-down usage reflects the
firm's experience that, unsupervised, algorithms can increase execution costs
that outweigh saved commissions.
"We find them to be valuable tools that, when used by a trader, provide
optimum levels of execution. But when used blindly - as they were designed to be
able to function - they frequently make foolish decisions," he said.
"There is some common sense that needs to be used," of which the algorithms
are incapable Bein added.
-By Mohammed Hadi, Dow Jones Newswires; 201-938-4049

(END) Dow Jones Newswires
08-02-05 1029ET
Copyright (c) 2005 Dow Jones & Company, Inc.(AP-DJ)--08-02-05 1029EDT