Brokerage houses are getting creative with new algorithms

NEW YORK —The next wave of trading algorithms is hitting Wall Street as both smaller, specialty brokerage houses and their larger counterparts seek to gain an edge in that area in catering to their money manager clients.

Firms as diverse as Instinet LLC, Rosenblatt Securities Inc., ITG Inc. and Banc of America Securities LLC are among firms that have unveiled or are working on new algorithms to address client needs to trade more efficiently.

"A lot of these algorithms have been driven by hedge funds, but as more traditional asset managers enter this marketplace we're going to see more customization," said Sang Lee, managing partner at Aite Group, a Boston-based financial industry consulting firm.

Customizing algorithms and creating algorithms that analyze trading from a portfolio level rather than a stock-by-stock level are two main areas brokerage firms are focusing on.

"If you had to trade 20 to 25 stocks, algorithms will give you an analysis of each stock — the market impact, etc. … these new algorithms will give you the market impact for the entire portfolio," said Harrell Smith, manager of the securities and investments group at financial services consulting firm Celent LLC, New York.

Brokerage firms swamped the market with algorithms in 2005. But traders at money management firms quickly realized most of them were basic tools that, while useful in certain circumstances, did not offer much in helping handle more difficult trades.

"VWAP, TWAP, everyone has one and probably to some degree they're commoditized, but there's probably a time and place for them to be used," said Michael Plunkett, president, North America, at New York-based Instinet, referring to volume-weighted average price and time-weighted average price algorithms.

He said such rules-based algorithms have a predictable pattern and can result in poor performance if other market participants, often using more sophisticated algorithms, pick up on the pattern and trade against it.

Mr. Plunkett added that while all algorithm providers need to offer basic tools like VWAP and TWAP, traders are looking for more advanced tools that mimic their own actions.

Ted Oberhaus, head of trading at Lord, Abbett & Co. LLC, Jersey City, N.J., said that given the explosion of algorithms in 2005 and the inevitable similarities among them, brokerage firms are now attempting to create unique algorithms. "A lot of these people are responding to their customers, which is really good to see."

He said he has asked "a lot" of his providers to customize algorithms because "my midcap value trader wants something different than my large-cap value trader."

Richard Rosenblatt, president and chief executive officer of the brokerage firm that bears his name, agreed that clients are looking for customized algorithms.

"What our clients really seem to be asking for are algorithms customized to their specific needs, and they simply don't exist," Mr. Rosenblatt said in an e-mail. "We are designing our interactive algorithms to not only provide custom execution solutions but also to adapt to a particular trader's style."

Mr. Plunkett said: "We're starting to try and think more like a trader, but it's difficult to put all that into a model or into a rule. If a trader gets an order in the middle of the day, instinctively he'll look at the last sale, opening price, high of the day, low of the day and who's been in the stock. In a matter of seconds, his brain will digest all that information.

"We've been trying to focus more on emulating what a trader does when he gets a stock," he said, adding that Instinet's Cobra algorithm, launched Jan. 23, aims to do just that.

Cobra is an order management algorithm that attempts to capture liquidity in a stock without being detected. It does so by waiting until liquidity builds up, then executing a trade without eliminating the entire pool of liquidity.

At Credit Suisse, New York, Dan Mathisson, managing director and global head of the firm's advanced execution services unit, and his team have been focusing mostly on expanding the use and functionality of the firm's algorithms across the globe.

Currently, clients can trade in 23 different markets and settle trades in any currency. Mr. Mathisson said the firm hopes to expand to four more countries this year.

"More and more, we're seeing clients setting up trading operations in a handful of cities and wanting to be able to trade around the globe from one place," he said.

For Rosenblatt and ITG, flexibility is a key feature that traders want in this next generation of algorithms.

Joseph Gawronski, chief operating officer at Rosenblatt, said one problem with current algorithms is that the trader has little opportunity to alter how the order is handled if new information comes in during the trade.

"If you decide to change your mind because of new information and you take (the trade) out of the algorithm and then put it back in, everything gets muddled up," he said. "They're not designed to work that way. We're incorporating certain color you might have from other sources, whether it's the (trading) floor or wherever," into new algorithms.

Tony Huck, a managing director and co-head of sales and trading at ITG, said the firm is working to leverage its experience in pre-trade and post-trade transaction cost analysis to provide more robust algorithms.

ITG is working to give "traders tools that help them decide in pre-trade which algorithm or algorithms a trade should go to and, during execution, being able to move between them," Mr. Huck said. "Then, after the trade is done, (we are) looking at the post-trade (analysis) and using that feedback to loop back to the beginning when it comes to the next trade."

Lord, Abbett's Mr. Oberhaus said the post-trade analysis piece is key to making algorithms work better.

"You have to keep in mind that most orders on the institutional side have to get done," he said. "Pre-trade is less important to us than post-trade analytics because post-trade teaches the buy side how to better use these store-bought tools."

Those "store-bought tools" are developing rapidly, with some firms creating algorithms to handle futures and options as well as incorporating algorithms into other asset classes such as fixed income and foreign exchange.

On Jan. 24, Banc of America Securities, a unit of Bank of America Corp., Charlotte, N.C., launched a smart order routing algorithm for equity options that allows users of its InstaQuote direct market access platform to trade equity options across liquidity pools.

Less than a week later, the company announced it was acquiring Financial Labs LLC, a Cambridge, Mass., firm that specializes in creating trading algorithms in the foreign exchange market.

Gavin Little-Gill, research director at TowerGroup Inc., Needham, Mass., called cross-asset class algorithms the third generation.

"That's the next generation — applying the liquidity that's being created across not just the equity market but increasingly across the fixed income, derivatives and foreign exchange markets," he said. "That's where it gets fun."