Best Execution & Transaction Cost Analysis
Best-Execution Mandates Pushing TCA Adoption | Asia-Pacific: Best-Execution Regulations on the Way? | Race Is On for Alpha-Based Algos
Race Is On for Alpha-Based Algos
For Wall Street, incorporating alpha into strategies is the next big thing
September 22, 2008
U.S. Regulation National Market System mandates that broker-dealers and exchanges ensure best execution, but, when more than one displayed venue has the same price, many institutions rely on algorithms to find an edge, or search the dark books for a better fill. As such approaches have become increasingly common, some market participants are pointing to alpha as the next algorithmic development--for firms that can afford it.
Today, electronic routers and their algorithmic cousins cast bids and offers across the market so rapidly that best price is often available on multiple platforms. Consequently, trading tools consider a destination’s speed, stability, fees and other technical factors, as well as plying dark pools for midpoint price improvements. But even that has become run of the mill.
“There hasn’t been much that’s new or novel in this space in a while,” noted Neil Fitzpatrick, COO of Citadel Execution Services. The biggest recent changes have been non-technical, he said, such as the proliferation of dark liquidity pools and broker-dealers’ strategic investments in venues. “What would be extraordinarily valuable and a material advantage,” Fitzpatrick observed, “would be the introduction of true, proprietary alpha.”
Alpha is an investment strategy’s return in excess of a market benchmark. Major trading desks, such as Chicago-based Citadel’s, have long incorporated alpha into their proprietary trading models, typically driven by automated short-term price forecasts.
Alpha-based algorithmic strategies are less widely available to buy-side firms, Fitzpatrick said, due partly to cost, but also because of the challenge of convincing users they’re getting optimal alpha, especially if the provider is a broker-dealer with its own proprietary desk.
Gary Ardell, head of financial engineering and advanced trading solutions at New York-based BNY ConvergEx Group, said the cost to develop such algorithms--mainly the technology to support the data-intensive models--has made it prohibitive for the majority of firms. “There are maybe ten of us with sufficient infrastructures to build models seeking alpha,” he said, including the bulge-bracket firms and the few agency brokerages that are as large as his firm.
Limited to Large Firms
Echoing Fitzpatrick, Ardell said that only the largest buy-side firms can accurately measure their trading performance and carry sufficient clout to demand their algorithm providers deliver full alpha, rather than see some slip away to proprietary desks. “They might be able to demand all of the trading performance; most clients aren’t that demanding,” said Ardell.
For brokers with proprietary desks, the narrow dissemination of successful alpha-driven algorithms may have practical reasons as well. “What would be the incentive for proprietary trading brokers to give their clients access to their successful alpha models?” said Tony Huck, managing director of algorithmic trading at Investment Technology Group (ITG) in New York.
Agency broker ITG is one of the grandfathers of algorithmic strategies and trade performance measurement and has had alpha embedded in its models for some time. Huck described the alpha component as “very short term, almost stat arb-like.” Statistical arbitrage typically looks for pricing discrepancies between securities and seeks to forecast their price movements--creating alpha-- a few seconds or minutes into the future, based on historical data. A few years ago, ITG launched its dynamic implementation shortfall strategy, which incorporates such an approach. Customers use it for rebalancing or other portfolio-wide trades, and it takes into account the effect of trading an individual stock on others in the portfolio, the portfolio as a whole, or a subset.
