As a trend, algorithmic trading is practically passé: That's how pervasively this once-fringe activity has swept Wall Street. Almost every major brokerage house has adopted the algorithmic formula by building or buying computer systems that split up big orders into manageable pieces and execute them over time. About 11 percent of trades are processed via algorithms, according to Tabb Group; the Westborough, Massachusetts, research firm predicts that the proportion will rise to 17 percent by 2007.
These newfangled platforms can dramatically cut transaction costs -- commissions alone are about one fifth those charged for conventional trades -- and veil traders' intentions. But they're also adding a layer of complexity to trading that can confound institutional investors. Indeed, algorithms have become so adept at turning outsize orders into hundreds -- or even thousands -- of diminutive ones that the system may be perpetuating one of the very problems it was supposed to solve: a lack of sufficient big orders to facilitate block trades. Paradoxically, the mathematicians and computer scientists hired by Wall Street firms to develop these advanced systems now find themselves having to constantly improve existing algorithms and develop new ones to keep up with the continued slicing and dicing of orders.
"It's like The Terminator," explains Adam Sussman, a senior consultant with Tabb. "You have algorithms fighting against each other, figuring out what each is doing. Because [investors] are the ones using these tools that break large orders up into smaller and smaller pieces, they are in part responsible for the fragmentation of liquidity."
For buy-side traders, knowing which algorithms to employ under what circumstances -- or whether to use algorithms at all -- is becoming ever more challenging. Consequently, brokerage firms not only must offer an array of options for executing trades, but they also must be able to make sense of the choices for clients.
Many firms provide tools that let customers estimate how sundry alternatives would affect their transaction costs. Sales-traders who mostly offer market color are shrinking in number or reinventing themselves as consultants who guide clients through the maze of algorithmic options.
Increasingly, institutional clients uncomfortable with the new arrangements are asking brokerage firms to put up capital to fill block orders in single transactions, thus relieving customers of the burden of parceling orders out into the market.
"It's important for the brokers to offer a whole suite of services, everything from the algorithmic model to the full-service block-trading desk," notes Mary McDermott-Holland, a senior vice president of trading at Boston-based Franklin Portfolio Associates, a quantitative manager with $34 billion in U.S. equity assets.
Which Wall Street firms are doing the best job of steering clients through this daunting environment?
To find out, Institutional Investor surveyed hundreds of head traders at money management firms, asking which brokerages deliver the best execution for New York Stock Exchange and Nasdaq Stock Market stocks. We also had these top traders rank the best exchanges and alternative-trading systems (see table, page 50). In addition, our survey sought their opinions of various equity-trading attributes, such as the quality of algorithmic platforms and sales-trading. In all, more than 300 traders responded to our second annual survey, conducted during the summer and early fall. (For more complete results please visit our Web site, www.institutionalinvestor.com).
Lehman Brothers dominates the rankings, repeating as tops in NYSE trading and leaping from fourth to first on the Nasdaq list. Buy siders say Lehman does the best job of combining traditional "high touch" services with technologically advanced "low touch" execution tools. Specifically, the firm's clients laud its transaction-cost-analysis system, delivered through its LehmanLive online research portal, which provides pretrade estimates and posttrade reports.
"It's an example of a product that the buy side would normally have to go out and purchase," says Lawrence Peruzzi, a senior equity trader at Boston Co. Asset Management. "They've done a great job."
Lehman complements its technological prowess with attentive service. For instance, the firm has added coverage of 500 companies to its Nasdaq roster in the past year, bringing the total to 1,750 stocks.
The firm stands out, too, say buy siders, for being willing to commit capital to concluding clients' block trades in one fell swoop. Most brokerages have shrunk from using their capital in this way as equity-trading commissions have declined; they reason that they can earn higher returns on other activities, such as proprietary trading or lending to hedge funds. Lehman, however, believes that as technology makes trading intimidatingly complicated, clients will increasingly value the ease of a single-counterparty trade.
"Our goal is to provide liquidity to our clients, through capital or through access to liquidity across multiple destinations," says Gerald Donini, head of U.S. equity trading at Lehman.
Merrill Lynch & Co. holds on to its No. 2 ranking in Nasdaq trading and jumps one spot to second in Big Board stocks. Viewed in recent years as a laggard in electronic trading, Merrill last year hired Rohit D'Souza, who played a critical role in developing Morgan Stanley's acclaimed Passport algorithmic platform while trading chief there, as head of global trading. At Merrill, D'Souza, who worked for cutting-edge brokerage Investment Technology Group before joining Morgan Stanley, has been bulking up technology, chiefly by enhancing the ML X-Act electronic platform.
"We've had an intense focus on direct-access and program trading capabilities and built out significant product over the first half of the year," says D'Souza. "These efforts are coming to fruition."
Third behind Lehman and Merrill in NYSE trading is Citigroup, followed by Bear, Stearns & Co. -- which leaps four places from last year's survey -- and Morgan Stanley. The same trio rounds out the top five Nasdaq traders, but in a slightly different order: Morgan Stanley takes third, followed by Citi and Bear Stearns.
Our head traders' choice for top execution venue, including conventional exchanges and alternative-trading systems, is no surprise: Liquidnet. The New Yorkbased electronic network boasts an average execution size of 60,438 shares for large-cap stocks -- compared with fewer than 400 shares for a typical NYSE trade.
"As the institutions are finding it increasingly difficult to execute the size they require, and the Wall Street desks are becoming less effective at executing blocks, we are one of the only venues out there where they can go and find size," says Seth Merrin, Liquidnet's chief executive officer.
Clearly, block liquidity -- whose virtual disappearance from the conventional market concerns brokerages and clients alike -- is the single most important factor affecting the ratings of stock exchanges and electronic networks in our survey.
Here Liquidnet shines. The firm operates a private crossing network, launched in April 2001, that is for institutional traders only. Members let Liquidnet scan their order books, enabling it to identify matching block orders. Prospective counterparties pair off in a chat room to hammer out a price.
Liquidnet, a broker-dealer, also cracks the survey's top ten NYSE and Nasdaq brokers, finishing ninth and tenth, respectively. And it is burgeoning. The firm's average daily trading volume has doubled in the past year, to 40 million shares. Every day, Liquidnet's network has more than 1.5 billion shares eligible for execution, according to Merrin. That makes it the third-biggest liquidity pool in the U.S., behind only the NYSE and Nasdaq. Another metric of success: Members take advantage of potential matches 50 percent of the time, up from 17 percent three years ago, according to Merrin.
Other alternative trading systems also fare well in our rankings. Bloomberg Tradebook, an electronic communications network launched by the financial news and data giant in 1996, finishes second behind Liquidnet among execution venues, falling from first place last year. The Inet ATS, an ECN currently being sold by Instinet Group to the Nasdaq Stock Market, is third. Nasdaq itself ranks seventh and the NYSE tenth, behind Investment Technology Group (which operates the Posit crossing network and finishes fourth) and fifth-place Rediplus (the algorithmic system operated by Goldman Sachs Group's Spear, Leeds & Kellogg unit). The Archipelago Exchange, now being acquired by the NYSE, finishes higher than its parent-to-be, in sixth place.
Algorithms may be making matters more complicated, but they didn't create the problem of fragmented stock market liquidity. In 1997 the Securities and Exchange Commission responded to a price-fixing scandal involving Nasdaq market makers by enacting new order-handling rules that inadvertently gave a huge boost to ECNs. They mushroomed: Trading interest migrated from a few centralized exchanges to dozens of upstart venues, like Instinet.
Then, in 2000, Congress mandated that in 2001 prices be quoted in decimals rather than fractions, reducing the minimum price variation from 6.25 cents (one sixteenth of a dollar) to just one penny. This seemingly technical change in fact made it far easier for floor traders and others to jump ahead of block orders by bidding or offering just a penny better. Wary institutions began trimming their order sizes, fearful of revealing their intentions in the market.
Today's electronic-execution options are all geared specifically to this new regime. Mutual and hedge funds have flocked to systems allowing them to access electronic exchanges without a broker's having to intervene. To direct-access platforms, the more advanced systems add proprietary algorithms. One of the most popular of these attempts to beat the volume-weighted average price, or VWAP, of a stock. Others help traders behave more or less aggressively, depending on the nature of their orders, in a bid to minimize trading costs.
Algorithms are hardly the last innovation, either. Starting next year, controversial new SEC rules, known as Regulation NMS, for national market system, are supposed to more efficiently link the country's various execution venues. The theory is that traders will always be able to access and execute against the best available prices.
There is one exception: Fully automated markets such as Nasdaq get to ignore superior prices from manual markets, such as the floor of the NYSE, which can take longer to update quotes. In response, the Big Board and the American Stock Exchange are rolling out their own hybrid markets that blend floor auctions with electronic matching.
One possible outcome of all this ferment could be a bigger role for now-moribund regional exchanges: They can pick up volume across the nation if they have automated systems that display the best prices. Tellingly, a group of Wall Street firms and giant hedge fund Citadel Investment Group have bought stakes in the Philadelphia Stock Exchange.
"We don't know what the fallout will be from Reg NMS, but the way the markets work could look very different a year from now," suggests Franklin's McDermott-Holland.
Human traders may be the biggest loser, as institutions' demand for lower and lower execution costs pushes Wall Street toward greater automation.
Still, the march toward more-efficient trading doesn't necessarily trouble Wall Street trading officials. Many argue that a blend of human and automated execution services will hold sway for quite some time -- especially as technology adds still more complexity to the buy-side trader's job, even as it purports to solve his or her problems.
"Everyone talks about commissions and margin pressure," says Lehman's Donini. "But when you cut through all the noise, there is the opportunity to continue to grow your market share, provide superior value to your clients and continue to get rewarded for that."