Are traders getting hosed by algorithms?
A new study contends that the indiscriminate selection of algorithms by the buyside drives up trading costs. Pros who opt for certain "nave" algorithms risk tipping off predatory traders. Algorithms that trade blocks in a nave, or non-random, fashion can be "reverse engineered" by these other traders.
"There are a lot of people out there sniffing the tape and looking for these patterns," says John Wightkin, a principal at the Quantitative Services Group (QSG), a transaction cost consulting firm and the publisher of the report. Wightkin says traders at hedge funds and broker dealers monitor the market in hopes of detecting large trades being parceled into the market. They can program computers to front run the trades.
QSG examined more than 120,000 trades totaling over $1 billion conducted by a large money manager attempting to match VWAP. It found those trades executed in a non-random fashion cost 26 basis points. Those fed into the market in various sizes and at random time intervals cost only two basis points. There are no estimates of how much volume is pushed through nave algorithms. But, Wightkin notes, some vendors sell trading platforms that include basic algorithmic strategies. These can be detected.
Wightkin also notes that the TWAP, or time-weighted average price, algorithm offered by many large brokers is susceptible. "People can easily identify a strategy which offers up stock every two minutes or so," Wightkin says. One brokerage exec agrees. "It's no surprise that very simple or non-randomized algorithms get picked," says Dan Mathisson, who heads up Credit Suisse First Boston's Advanced Execution Services group. "The game is about developing algorithms that are unpredictable."
