As buy-side traders increasingly voice concerns about information leakage, many brokers have coded all of their algorithmic strategies to search for liquidity in dark pools. These private networks match trades anonymously without publishing a quote, preventing trades from moving the markets.

Up until about a year ago, brokers provided "pure-play" dark pool algorithms that were specifically designed to source and consolidate liquidity from dark pools. Credit Suisse, for example, offered Guerrilla and Sniper, while Citi offered Scouter and Instinet had Nighthawk.

Because of the success of Guerrilla, however, Credit Suisse Advanced Execution Services (AES) decided to separate out the dark pool aggregation code from Guerrilla and add it to other algorithms and electronic trading strategies. "What we saw is that dark pools became more popular," says Dmitri Galiametidinov, a director at AES.

As a result, in January 2007, Credit Suisse launched CrossFinder+, an algorithm that posts liquidity and takes liquidity from multiple dark pools. While clients can leverage the stand-alone algorithm, all other AES algorithms also use the CrossFinder+ technology. "The whole objective with CrossFinder+ is to maximize liquidity from dark pools, not move the market and get the best prices," explains Galiametdinov.

Similarly, Instinet's success with Nighthawk -- a dark pool algorithm that aggregates liquidity from multiple markets, including NYFIX Millennium, Liquidnet H20, Credit Suisse's CrossFinder, Fidelity's CrossStream, the ISE's MidPoint Match and other ECN/ATS pools -- led the agency broker to integrate dark pool aggregation into all of its algorithms in October 2007, following the completion of its merger with Nomura Securities.

"Dark pool access is clearly a very significant part of algorithmic trading, and we thought our clients should not have to access two separate algorithms to get dark pool access," explains John Comerford, EVP and global head of trading research at Instinet. After Instinet integrated its algorithms with the Nomura algorithmic trading platform last October, it began to systematically roll out this capability, he adds.

"The dark pool aggregation [now] is native to all [of our] algorithms," Comerford continues. "Every single algorithm that [clients] have can interact with dark pools, including our portfolio algorithms."

With the exception of VWAP, for which the client's goal is to obtain the volume weighted average price in a stock across the entire day or part of the day, most electronic trading strategies can be designed to interact with dark pools, according to Comerford. "With implementation shortfall, the challenge is to trade as much [volume] as possible without moving the stock price, and that is what dark pools are excellent for," he notes. "You can trade a tremendous amount of volume and no one sees [it]."

Comerford says Instinet currently executes around 20 percent of its orders purely in dark pools. But the firm also uses hidden order types on ECNs for another 30 percent to 40 percent of its orders, he adds.