TRADING - Daggers, Dark Pools and Disintermediation

Massive, once-in-a-lifetime changes are sending firms, investors and regulators scrambling to survive.

In late november, Tony Huck, CO-head of sales and trading at brokerage house Investment Technology Group, got word that a competitor, Credit Suisse, was allowing its clients to access his firm’s popular Posit block-trading network at cut-rate commissions. Credit Suisse, using a computerized algorithm called Guerrilla, was scanning Posit for available shares and charging clients half its rival’s rate of 2 cents per share for executed trades. The move was costly for Credit Suisse but gave its customers less incentive to go direct to Posit.

Huck called Credit Suisse electronic trading chief Dan Mathisson and asked him to stop. Mathisson declined, and Huck promptly barred Guerrilla orders from the Posit system. Neither man will comment on the incident, though the two firms have since begun talks aimed at allowing Credit Suisse customers to once again trade with Posit orders.

Squabbles like these are playing out all over the suddenly topsy-turvy world of stock trading as brokerage firms, exchanges and investors all scramble to survive the most sweeping transformation of trading since the deregulation of commission rates in 1975. The changes pit upstarts against established players, and former allies against one another in a no-holds-barred fight for market share.

Over the past two years, the New York Stock Exchange and the Nasdaq Stock Market went public and began buying or bidding for rival exchanges; the big Wall Street firms that were once loyal NYSE and Nasdaq members responded by investing in regional and upstart markets that are automating and coming after the big two. In January, Kansas City, Missouribased BATS Trading, a new, superfast electronic trading platform, began rebating customers for trades instead of charging them commissions. Within days it had doubled its share of Nasdaq volume, to nearly 11 percent. Last month Pipeline Trading Systems, an electronic block-trading network, closed itself off to certain orders generated by the brokerage firms that are its biggest customers.

“These changes are massive,” says Gregory Tusar, head of electronic trading at Goldman, Sachs & Co. “There is a tremendous amount of very large structural transformation going on in the marketplace.”

Much of the change is being driven by institutional investors, who need to trade big blocks but have a hard time doing so because of new regulations and technological advancements that have decentralized the market and slashed the number of shares available at quoted prices, even as volumes have surged. As a result, they are employing alternative methods. One variation is using computerized algorithms to chop up big orders and send the pieces to be executed wherever the best prices lie. Big brokerage houses like Citigroup and Merrill Lynch & Co. -- some of which were openly hostile to electronic trading just a few years ago -- are now fighting an all-out arms race to offer the most sophisticated algorithms, often giving them militaristic names like Credit Suisse’s Sniper and Citi’s Dagger. Consulting firm Aite Group estimates that algorithmic trading now accounts for one third of equity trading volume, up from less than 1 percent five years ago. Another alternative tactic involves looking for block-trading partners in private trading networks, such as Pipeline and Posit. They are called dark liquidity pools because they don’t post price quotes. One year ago there were six dark pools. Today there are 40 and counting, according to consulting firm TowerGroup.

In the old regime, institutions turned to Wall Street brokerages that sent their orders to the NYSE floor or worked the phones in search of a match. That cigar-jabbing, alpha-male-braying world is all but extinct. First the cigars went; then the men. Taking their place is a virtual marketplace in which superfast computers run by rival firms and programmed by mathematics Ph.D.s compete fiercely to execute almost all trades automatically. It’s an exhilarating, bewildering, frequently messy place where elements of a nascent market structure often work at cross-purposes; algorithms fragment trading interest into tiny, dispersed pieces even as dark pools facilitate private negotiation of giant transactions in one place. As a result, the long-established roles of brokerage firms, exchanges and customers are being redefined in extraordinary ways.

“The world is not easy to understand anymore,” says Timothy Mahoney, a 20-year veteran of Merrill’s investment management unit who in October became CEO of Block Interest Discovery Service, a new dark pool formed by a consortium of big brokerage firms. “It seems you can be a competitor, a client and a collaborator to everyone that you do business with.”

The new technology is slashing transaction costs for investors and squeezing the profits of intermediaries. Commissions for algorithmic platforms and dark pools are 2 cents per share or less, compared with about 4 cents for traditional trades. Total annual U.S. equity commissions shrank from $13.4 billion in 2002 to $10.8 billion in 2006, according to consulting firm Greenwich Associates, and are projected by consulting firm Tabb Group to fall a further $1.7 billion by the end of 2008.

Investors’ demands for efficiency have forced old-line intermediaries to adapt or lose business to upstarts like Pipeline and BATS. Retaining institutional brokerage business is particularly important for big investment banks because it supports far more lucrative activities like proprietary trading and equity underwriting. Customer volume helps firms divine the market’s overall direction, a boost to their own trading efforts. Corporate clients like to see that their underwriters have relationships with big shareholders. But as commission rates fall, brokerages earn less on each trade. That makes it critical for sell-side firms to invest in technology that allows them to process maximum order flow with minimal fixed costs. Trading commissions fund not only their sales and trading infrastructures but also their equity research departments.

“Every dollar paid into a dark pool that is not run by a sell-side entity is a dollar that is not being reinvested back into research and other products they can take advantage of,” explains Peter Forlenza, head of equities at Banc of America Securities.

A further incentive for brokerages to offer the most advanced electronic execution platforms is coming from a handful of investors who are pushing to unbundle trading from other services, such as research, that are paid for with commissions. A few institutions, including mutual fund titan Fidelity Investments, have cut deals with Wall Street firms to pay cut-rate commissions that only cover execution. Fidelity writes separate checks for any research it uses. In such an environment brokers will win business solely on the basis of their trading efficiency -- not, as is still frequently the case, on the quality of their research or primary offering calendar.

Regulators are adding yet another wild card to this fast-paced, high-stakes game. Regulation National Market System, passed by the Securities and Exchange Commission in 2005 and first implemented last month, forces brokerages to execute trades at the best quoted prices across all market centers that offer automated execution. (Before, trades could be executed on floor-based markets at prices inferior to those available on electronic competitors.) The regulation gives rivals and upstarts a huge opportunity to take volume from the Big Board and Nasdaq. With that volume comes fees for listings, market data and executing trades.

“The new rules open the door for a lot of competition,” says Adam Sussman, an analyst at research firm Tabb Group. “2007 will be a battle for liquidity and market share.”

AS INTERMEDIARIES SCRAMBLE TO REMAIN RELEVANT, more money managers are waking up to the impact that transaction costs can have on their returns and are demanding the most efficient execution services. Institutions that have been experimenting with alternative trading tactics for years continue to push the envelope. Kansas City, Missouribased American Century Investments, long a proponent of electronic trading, now executes 70 percent of its orders on automated markets, double the level of five years ago. “We’ve really tried to squeeze out the unfair inefficiencies that intermediaries have tried to layer onto the process,” explains John Wheeler, head of trading at the $106 billion-in-assets mutual fund house.

Another computerized trading pioneer, $370 billion pension plan and fund family TIAA-CREF, has begun developing its own algorithms instead of relying on brokerage-developed ones. State Street Global Advisors, the $1.5 trillion investment giant, operates an internal think tank called the Advanced Research Center that employs mathematicians, engineers and economists who, among other duties, develop algorithms for the firm’s 30 traders around the world.

“When I started in the business, the technology was a yellow pad,” says Roger Petrin, State Street’s global head of trading. “We’ve evolved. The technology gets better and better, and the type of tools the traders have to differentiate how they execute is night and day from year to year.”

Beyond early adopters like these, a wide array of institutions are beginning to embrace electronic trading. After starting to experiment with algorithmic trading two years ago, Los Angelesbased value investing shop NWQ Investment Management, a $14 billion-in-assets subsidiary of Nuveen Investments, now executes at least half of its orders using these systems, reports Kirk Allen, the senior vice president who oversees trading. The Teacher Retirement System of Texas expects to expand its use of dark pools, which it previously took advantage of only sparingly. “We expect that trading volumes in dark pools will increase because they provide limited opportunity for information leakage,” says Claudia Williams, head of equity trading for the $103 billion pension scheme. “Many of our full-service brokerage firms are trying to attract order flow, so they’ve been opening up their internal pools to customers.”

The institutions are all trying to solve one big problem: the incredible shrinking price quote. The average size of an NYSE trade is down from about 1,600 shares in the late 1990s to about 300 today. The evaporation of block liquidity is a consequence of several regulatory events and the market’s response to them. In 1997 the SEC began requiring brokerages to display customer orders they previously could keep hidden; this spawned an array of electronic quoting platforms that started to siphon volume from Nasdaq and, to a lesser extent, the NYSE. In 2001 stock markets began quoting prices in decimals rather than fractions. Bid-offer spreads shrank, as did the willingness of traders to quote prices for large quantities of stock.

Trading through algorithmic systems and dark pools can yield big savings for money managers. Commissions on algorithmic trades, already lower than those for traditional transactions, fell by 38 percent in 2006, reports Tabb Group. More important, the advanced systems make it less likely that a stock’s price will move away from an investor in the time it takes to fill a big order. That can yield even bigger savings. The average cost of trading NYSE and Nasdaq stocks hit record lows last year, according to Elkins McSherry, a New York transaction-cost analysis firm (Institutional Investor, December 2006).

Hedge funds are also driving the growth of algorithmic trading. Employing rapid-fire quantitative strategies designed to exploit minute price differences, they see the systems as a way to generate returns, not just cut transaction costs. Indeed, the earliest algorithmic platforms were developed by obscure trading firms for this purpose; many grew from technology day traders used during the 1990s to profit by quickly buying and selling the same securities.

Since institutional investors began to embrace algorithms about five years ago, Wall Street has spent hundreds of millions of dollars on making cutting-edge trading technology available to clients. Five years ago just a handful of firms, led by Goldman, Sachs & Co. and Morgan Stanley, had algorithmic trading systems. Today about 30 brokerages each offer up to a dozen of these tools.

“So many institutions now have the car that enables them to get on the highway of algorithmic trading,” explains Alfred Eskandar, head of strategy at electronic block-trading network Liquidnet.

Programmers originally designed the algorithms for the most liquid large-cap shares, but now they are creating increasingly sophisticated programs that target even the most thinly traded stocks and respond to unexpected market shifts. JPMorgan in September launched Arid, an algorithm for illiquid stocks. Banc of America Securities last month introduced Instinct, an algorithm targeted at small caps. Late last year agency brokerage Instinet rolled out Wizard, which attempts to gauge how events like mergers, central bank decisions and world news will affect stock prices and then adjust execution strategies accordingly. Sophisticated clients do their own research into the most efficient ways to execute various types of orders and have brokerages design custom-made algorithms to suit each one.

“This year we have developed more customized algorithms than we have in the past five years,” notes Manny Santayana, head of sales and marketing in the Americas for Credit Suisse’s algorithmic unit, Advanced Execution Services.

Some big firms have acquired upstart algorithmic shops; Merrill, for instance, last year bought Wave Securities from Archipelago Holdings, now a part of NYSE Group. Others have hired teams of mathematicians and technologists to build their own systems. On busy trading days Credit Suisse, widely recognized as the leader in the category, executes 70 percent of client orders through AES.

But algorithms are only part of the solution. Indeed, because they chop big orders into tiny pieces, algorithmic systems exacerbate the problem of shrinking transaction sizes. That’s where the related explosion of dark pools comes into play. These private networks come in many varieties but share some traits and goals. They are all virtual meeting places where buyers and sellers of big blocks can find one another without revealing their identities or intentions. Where exchanges require users to show everyone else in the market how many shares they want to buy or sell and at what price, dark pools keep that information secret and use technology to pair off potential trading partners. They also all allow traders to move big blocks in a single piece instead of slicing and dicing them. But they achieve that end using various means. Liquidnet, a six-year-old firm that handles about 60 million shares a day, connects directly to the electronic books money managers use to manage their pending orders and automatically alerts them to potential matches; counterparties then negotiate the particulars in an online chat room. ITG’s Posit runs a series of crossing sessions -- essentially blind auctions -- throughout the day. Liquidnet limits its membership to institutional investors. Posit, Pipeline and others welcome brokers as well as institutions. Some have minimum order sizes, like Pipeline’s 25,000 shares. The majority do not. Most of the trades done in dark pools must be executed within the spread between the best bid and offer available on public exchanges for the stock in question. They also must be reported to the market, through a registered securities exchange.

“The world has embraced algorithmic trading, but it also has created a need for clients to find liquidity,” says BofA’s Forlenza. “Rather than asking brokers for capital, they prefer to find natural liquidity on the other side in anonymous fashion.”

Some brokerage houses, such as Credit Suisse and Goldman, have been operating internal crossing networks for quite a while. These systems match customer orders in-house, saving firms the fees and commissions associated with filling orders on exchanges and other venues. Now several firms are opening these internal dark pools for clients to access directly, transforming them into competitors to the likes of Liquidnet and Pipeline. Credit Suisse last year began offering its CrossFinder system to institutions; it now handles 60 million shares a day. Similarly, Goldman recently began giving clients access to its Sigma X crossing network. UBS last year rolled out a dark pool called Price Improvement Network and opened it to clients.

“People have been experimenting with new ways to trade more and disclose less,” says Goldman electronic trading chief Tusar, explaining the profusion of private networks. Adds Chris Heckman, co-head of ITG’s sales and trading group, “There is no question that the success of the existing block-trading platforms has driven the brokers to create new ones.”

Exchanges, too, are seeking to blunt competition from independent dark pools by launching their own. The ISE Stock Exchange plans to combine a dark pool with a displayed market, as does the Boston Equities Exchange, an electronic market launched last year by the floor-based Boston Stock Exchange and several brokerage houses. Traders can pass their orders through the private market en route to the exchange quote platform. To the extent that matches are available in the dark market, all or part of these orders will be executed, with the remainder going into the displayed book.

The NYSE is planning to introduce a dark pool that crosses orders at various points during the trading day, says Colin Clark, vice president for strategic market analysis. The exchange already offers an after-hours crossing session. Nasdaq, too, is building a crossing network that will offer three intraday sessions to complement its existing crosses at the open and close of trading.

IN THIS UPSIDE-DOWN TRADING WORLD THE BALANCE of power is shifting dramatically toward institutional investors, who stand to benefit the most from technological advances and the intense competition among brokerages and exchanges.

Nowhere are the stakes more evident than in the suddenly wide-open playing field for exchanges and alternative markets, chiefly as a result of Reg NMS. The rule’s dictum that traders must execute orders at the best available prices regardless of where they are quoted gives small markets a huge opportunity to get bigger -- provided they have superior technology. Markets that can display quotes and execute trades the quickest will threaten those with slower systems.

When the rules were first proposed in 2005, the NYSE announced plans for an automatic execution system that would take priority over its trading floor. Today, with that so-called hybrid system fully rolled out, some 95 percent of Big Board trades are done electronically. The rules also influenced the NYSE’s March 2006 acquisition of all-electronic market Archipelago Holdings, which in turn prompted Nasdaq to buy Instinet Group’s Inet ECN, then widely regarded as the industry’s fastest trading system. The NYSE was already losing volume to electronic markets. In February it executed 66 percent of the trades in its listed stocks, down from 76 percent one year earlier and nearly 85 percent in 2000. Much of the decline has been at the hands of Nasdaq, which now handles nearly 15 percent of NYSE volume following its acquisition of lightning-fast Inet. (It’s too soon to tell what effect Reg NMS has had on market share.)

The same firms that have been frantically launching algorithms and dark pools are hoping to capitalize on Reg NMS by investing in a host of competitors to the NYSE and Nasdaq. These rivals, including long moribund regional exchanges in Boston, Chicago and Philadelphia as well as new entrants into stock trading like BATS, the ISE and the Chicago Board Options Exchange, are using the brokerage industry’s millions to build shiny new electronic trading engines in the hopes of luring volume from the big two (Institutional Investor, July 2006).

Credit Suisse, Lehman Brothers, Merrill and Morgan Stanley are among the investors in BATS, a start-up that after barely more than one year in operation has grabbed about 10 percent of Nasdaq’s volume. Following its January rebating experiment, which cost it $5 million, BATS has returned to charging customers for trades but is still betting on aggressive pricing -- its net commission rate is an industry-low two hundredths of a penny per share. To CEO David Cummings the rebates constitute money well spent; the firm’s Nasdaq market share almost doubled within a few days of the start of the rebate program.

“We think it’s important to get critical mass,” says Cummings, adding that $5 million is a drop in the bucket compared with the billions that Nasdaq and the NYSE have spent on acquisitions to boost their market shares. “The pricing scheme was designed to get us to volume levels that are sustainable.”

Says Michael Rosen, a product manager with electronic agency brokerage UNX, “If I were running a market today, my big fear would be that overnight after the introduction of Reg NMS, the years and years of work I’ve done to get people to come to me will be totally irrelevant.”

Some experts expect that one or two of the challengers to the NYSE and Nasdaq will swipe enough market share to emerge as serious rivals, triggering another round of consolidation.

“The winners of all this will be those that develop the biggest pools of accessible liquidity,” says TowerGroup analyst Robert Hegarty. “For the ones that can, there’s a whole lot of money to be made. The ones that can’t face obsolescence.”

Certainly, the exchanges are feeling the pinch. Late last year NYSE CEO John Thain announced the exchange was cutting 520 jobs and closing a portion of its trading floor. Big specialist firm LaBranche & Co. said in January that it had cut its 190-strong floor staff in half in the previous few months. It and other floor-focused firms will have to shift into upstairs trading like other brokerage firms to thrive over the long haul. BofA, Goldman and Lehman all laid off NYSE floor staff earlier this year.

Making matters even more complicated: Exchanges battling one another are now frequently competing against their former members. Even as some Wall Street firms create dark pools to stem the tide of customer orders being lost to the likes of Liquidnet, ITG and Pipeline, they are investing in regional markets that aspire to drain volume from the NYSE and Nasdaq. Together these steps allow brokerage houses to create a potentially powerful new economic model for equity trading. Dark pools let them internalize customer orders, and ownership of regional markets allows them to cheaply report, or “print,” these executed trades to the rest of the market. By law such reporting must be through a registered securities exchange, which can then turn around and sell the data to news services and other customers. Exchanges also frequently rebate a portion of this market data revenue to customers, another incentive for brokerage firms to print internalized trades on regional markets. At the end of last year, brokerage firms were internalizing about 16 percent of their daily order flow, up from 12 percent one year earlier, according to Goldman data.

All this upheaval could also hurt investors. The proliferation of dark pools and the rise in internalized trades could create a bifurcated market serving separate classes of investors. One would consist of undisplayed, anonymous markets where big institutions negotiate block trades. The other -- exchanges and other markets that display quotes -- would become a dumping ground for retail flow. Under such a scenario orders in the displayed markets, though they would be price-protected by Reg NMS, would no longer receive the benefit of interacting with the institutional flow being diverted to dark pools. Such intermingling, by definition, can lead to better prices for all participants. Another wrinkle involves broker-sponsored dark pools that internally cross orders from a firm’s retail brokerage clients with its institutional flow. Technically, these retail orders might be price-improved by mixing with internal institutional orders, but they’re also not being published to the wider market to see if a better deal is available. Brokers must match the national best bid or offer on internalized trades, but orders that are exposed to the market often prompt other participants to adjust their quotes, creating the possibility for an execution at better prices.

“We appreciate the need for brokers to try to save costs and avoid routing orders to exchanges, but we believe the best venue in terms of price discovery is the public market,” says NYSE strategy executive Clark. “When a broker internalizes in its own dark pool, it is limited to its own customer base. The NYSE is a neutral market open to all market participants.”

The potential for a bifurcated market is something the SEC may have to consider in the near future: Should retail and institutional orders be forced to mix, or can stock trading function as most other industries do, with separate wholesale and retail markets?

In the meantime, the private sector may be coming up with ways around the bifurcation issue. The latest innovations in electronic trading, in fact, center on attempts to bring together orders in dark pools and public markets. Liquidnet’s H2O system, launched last year, allows flow from several broker-sponsored algorithmic trading systems to pass through its private block-trading network before being routed to exchanges and ECNs, in the hope of finding potential matches. Exchanges like the ISE and the NYSE are increasingly giving customers the opportunity to cross blocks in their own dark pools before sending unexecuted portions of big orders to their displayed liquidity pools. Even buy-side firms are experimenting with such tactics. American Century is building a system that will pass orders bound for exchanges through various dark pools in search of better deals. If the system finds a superior price on one of the crossing networks, says trading chief Wheeler, it stops the order from proceeding further and executes it in the dark pool.

Taking orders bound for public markets and first exposing them to private ones is, of course, exactly what Credit Suisse was trying to do by including Posit as a destination for orders generated through its Guerrilla algorithm, which probes multiple dark pools along with public sources of liquidity. In the process, however, Credit Suisse undercut ITG’s pricing model, essentially taking a 1 cent-per-share hit to attract customers. In the tumultuous world of institutional trading, every solution creates a problem for someone.