Brokers’ electronic trading desks have been as busy as ever this year developing algorithms. Cases in point include Bloomberg with B-Dark, ConvergEx with Abraxas and Deutsche Bank with SuperX. SunGard even took a shortcut by purchasing Fox River to offer its algos.
But in 2010, the algos many brokers were developing were those tailored specifically to their buyside clients. For algorithms, it was the year of customization.
That’s certainly the case at Credit Suisse, said Eugene Choe, head of sales for Advanced Execution Services in the U.S. Clients often asked AES to design algos that make real-time adjustments based on the market conditions around a specific sector or an index in an environment that has been highly correlated for the last half of the year. The firm calls the algos "relative value" customizations.
"There are a lot of clients who require specific customizations," Choe said. "Over the last two years, really all we’ve been doing is customized strategies. One size does not fit all."
Algorithm use by money managers hasn’t reached a plateau, as it continued to rise in 2010. And it should only become more widespread. According to an industry study of electronic trading trends by Tabb Group, the buyside used algorithms to execute 29 percent of its flow. The number is estimated to rise to 35 percent in 2011-putting it neck and neck with high-touch volume.
As algorithms have evolved, the buyside has been getting more comfortable using them and, in turn, has been expecting more out of them, industry execs say. Accordingly, the buyside has been letting brokers access information on their inner workings to build them better and more customized algos.
Firms such as Goldman Sachs, Citi and Pipeline Trading Systems said their customers want tools that will help them trade based on data from their own past trading decisions. The brokers are working more closely with buyside traders and even portfolio managers to construct the algos that suit their trading and investment strategies. It’s the next level of customization, said Greg Tusar, head of Goldman Sachs Electronic Trading in the Americas.
"Many clients have asked for help in making the algorithm selection process more scientific, mainly by working together to profile by a variety of characteristics: urgency, size, market cap, etc.," Tusar said. "In working together to do that, we in effect customize an algo which may reflect orders from different funds, PMs, traders, etc."
Ultimately, buyside traders wanted algos that suited the different objectives of their many portfolio managers. As an example, Pipeline introduced a service it developed alongside J.P. Morgan Asset Management and AllianceBernstein that claims to do this by bringing execution tools one step closer to portfolio managers and their traders. (J.P. Morgan Asset also worked with trade-cost analysis provider Abel/Noser to help it develop algos tailored to portfolio managers’ investment and trading styles.)
The service helps traders develop and deploy custom trading strategies, said Fred Federspiel, Pipeline’s chief executive. The technology starts by analyzing up to hundreds of thousands of trades an institution has executed, to build a profile of its desk. Then, after Pipeline receives an order from a trader’s order management system, the technology takes the instructions and current trading conditions and compares them to the institution’s profiles of historical trading activity to identify analogous conditions.
It predicts the order’s intraday alpha during those conditions. Then Pipeline devises custom strategies that maximize the institution’s alpha capture, based on those analogous trading conditions.
For its part, Citi is also creating an algo that maximizes alpha to specific asset managers. The technology starts by creating profiles of an institution’s PMs, using two years of trade data. It then uses that to map out the "alpha expression" for each PM, said Dan Keegan, co-head of Citi’s electronic trading division. But then the algo will build the institution a "rules engine, which will then guide a lot of the order placement, and micro-strategy order placement to all of their internal orders," he said.
Traders still use the core algorithms, as well, Credit Suisse’s Choe added. But customized algos are popular because they’ve given the buyside more say in the creation of the very tools it will soon be using to trade.
