Rob Daly
Traders Magazine Online News

OPINION: CAT NMS is Out of Options

The SROs have no choice but to meet their deadlines.

Traders Poll

Are you ready to comply with the new updates required by the amended Rule 606?

Free Site Registration

January 21, 2010

High-Frequency Debate Continues

By James Ramage

Also in this article

  • High-Frequency Debate Continues
  • Page 2

Two recent studies address fears traders have about high-frequency trading, relating to pattern recognition and adverse selection.

The first, by brokerage Investment Technology Group, said high-frequency trading strategies hurt best execution in the dark pools they frequent. The second, by Quantitative Services Group, a provider of transaction-cost research, said these strategies pick off volume-weighted average price orders and lead to higher impact costs.

Both studies attempt to quantify how high-frequency trading strategies affect the equities markets. These studies also follow an earlier joint study conducted this past summer by electronic broker Pipeline Trading Systems and the money manager AllianceBernstein, which examined dark pool aggregators and adverse selection. All three studies said that interacting with high-frequency flow in certain scenarios can be costly.

Prior to the two recent studies, a number of buyside traders said they were not convinced of the hazards of interacting with high-frequency flow. That conclusion came from an October buyside survey by Greenwich Associates. The survey reported that 90 percent of the buyside traders surveyed said they need more information before forming an opinion on how the practice impacts their trading costs.

It remains to be seen whether these studies will have an impact on the buyside's views. Still, the studies addressed a question that has generated tons of debate over the last year, both inside and outside the industry.

"We see clear signs that the dark pools that have more high-frequency flow have a lot more adverse selection than the dark pools that have lower high-frequency flow," said Hitesh Mittal, who manages ITG's crossing and algo products.

As many dark pools allow unfettered access, regardless of flow type, high-frequency trading activity therein has surged, according to the ITG study. And as high-frequency traders use short-term alpha models, incidences of adverse selection have also risen, ITG reported. Adverse selection refers to trading with the wrong counterparty, such as a more informed trader. For its part, ITG runs a dark pool that does not let in orders from high-frequency traders. The same goes for Pipeline.

To reach its conclusions, ITG measured adverse selection in dark pools using three methodologies. The first measured price drift, from a profit-loss perspective. The second method examined the price movement in a stock from the order arrival to the end of the day--which is called alpha. It compares the alpha of executed orders in that stock to the alpha of unexecuted orders, and quantifies the difference as adverse selection. Finally, ITG developed a Time-Weighted Average Mid-Quote--or TWAM--benchmark. TWAM calculates the time-weighted average mid-point price both before and after the execution time and compares it with the execution price to gauge adverse selection.

But to others, the matter of adverse selection may not so easily be pinned to high-frequency traders. There are varying types of adverse selection, argues Owain Self, head of UBS's algorithmic trading for the Americas, emerging markets, Europe and Asia.

"Some people would argue that trading with somebody that's bigger than you is potentially adverse selection," he said. "At the end of the day, once you've traded, they're going to continue to trade in the market and push the stock against where you just were filled."