What does a baseball all-star like Derek Jeter, or a tennis great like Roger Federer have in common with the best high-frequency traders?
In each instance - from Jeter swinging at a 90 mile-an-hour fastball, to Federer returning a 100 mile-an-hour serve; or even a trading firm firing off a transaction at warp speed – they're actually putting off their execution to the last possible millisecond.
That ability to "optimize delay," is what separates the pros from the amateurs in sports, and is what fattens the profits of the best high-frequency trading operations, according to University of San Diego law and finance professor Frank Partnoy's new book, "Wait: The Art and Science of Delay."
"The idea of optimizing delay is just to figure out how much delay will generate the highest profit," Partnoy explains. "A lot of high-frequency trading firms, if they get fast enough with their particular strategy, it not might be worth it to pay huge amounts of money to get faster and faster."
Partnoy used the story of trading tool provider UNX to illustrate this point. In 2006, he says the firm was teetering as a result of a dated technology offering that drove an alarming number of its clients into the welcoming arms of competitors. Its chairman Andre Perold knew that in order to survive, the firm would need to quickly raise money to build a new platform. And they'd have to do it with the best technology possible, Partnoy writes.
Soon after naming Scott Harrison Chief Executive in July of that year, the firm accomplished its goal, having armed itself with faster, more efficient software to sell to clients who were looking to execute trades as fast and cheap as they could. Roughly a year later it appeared that UNX had won the high-frequency trading arms race, Partnoy points out.
But the firm wanted to get even faster, and Harrison decided to move the firm's trading operation to New York from Los Angeles in order to trim latency. Yet even after shaving 35 milliseconds following the move, Partnoy says the firm saw trading costs actually go up. Its trades got faster, but executions weren't as good.
"Finally, we gave up and decided to slow down our computers a little bit, just to see what would happen," Harrison told Partnoy. "We delayed our operation. And when we went back up to 65 milliseconds of trade time, we went back to the top of the charts. It was really bizarre. I mean, there we were in the most efficient market in the world, with trillions of dollars changing hands every second, and we'd clearly gotten faster moving to New York. And yet we'd also gotten worse. And then we improved slowing down."
There's been plenty of innovation in high-frequency trading in the years since, but Partnoy says that the smartest firms have reached the same conclusion: that although being fast is good, optimizing delay can be better. And he argues that a growing number of firms are designing their algorithms to approach the markets under this train of thought.
"Several leading high-frequency traders told me they encourage employees to take more time to reach decisions and they program their computer algorithms to anticipate how other traders will react, and then wait for those reactions," Partnoy writes.
"Sometimes these computer algorithms are set to minimize delay. But more often they are optimizing delay – buying and selling only a few shares during the first milliseconds, like the initial feint of a fencing duel, to test how other traders respond."