Speed Is Not Enough, Even in HFT

It is more important to make the most of the speed once technology has enabled it, say the panelists at the event, hosted by sibling publications Inside Market Data and Inside Reference Data.

"It's like the evolution theory," says Peter van Kleef, managing director at Lakeview Capital Market Services, based in Starnberg, Germany. "It's not the fastest or the biggest ones that win, but the ones that adapt best to change."

And adapting to change, in the context of today's volatile marketplace, means that cost and risk management must be weighed in discussions about low latency and high-frequency trading (HFT) strategies.

"Speed is important, but speed is not the only thing needed to win," says Donal Byrne, CEO of latency-monitoring software vendor Corvil. "For people doing electronic trading in today's environment, the elements of cost management and risk management are equally important. This is not to say that the time for speed has passed—quite the opposite. There's no real limit, barring the speed of light, on how fast these things can get. It's really just how fast everyone else decides to move, and that's ultimately limited by cost and risk, so there's a balancing action that happens here—we're right in the midst of that."

Lakeview's van Kleef says firms must either find different data from their competitors or come up with ways of looking at the same data differently—the latter being more of a challenge.

To come up with something that nobody else has thought of is always difficult. If you go to a new continent and no-one else has landed there, then exploring that continent is easy, but to find a new continent these days is pretty tough.

"To come up with something that nobody else has thought of is always more difficult," he says. "If you go to a new continent and no one else has landed there, then exploring that continent is easy, but to find a new continent these days is pretty tough."

It’s All Semantics—Or Is It?
One differentiator is semantic analysis, which could give firms a global edge, according to van Kleef, but he says he doubts its ability to impact decisions on a microsecond basis, which would be necessary for it to be of use in speed-reliant strategies.

"Semantic analysis, in my book, really doesn't work and will never work in terms of finding out what something really means for the market," he says. "I don't think it will be possible from a machine, at least in my lifetime. That will not happen. You can make an informed guess, but to make an informed guess you need a sample, averages, distribution, and you need to see how it changes, and that is nothing that will ever happen in microseconds."

Corvil's Byrne, meanwhile, says semantic analysis can still be of use in the trading domain, but he says the idea of using algorithms based upon sentiment changes in social media to effect trading decisions would end in disaster.

"I think sentiment analysis has a role to play in trading because sentiment fundamentally changes volume, and volume fundamentally changes the characteristics of the systems that we trade on, and when those systems are under that level of pressure, different latency arbitrage opportunities arise," he says. "But I think, traditionally, artificial intelligence, inference engines and all those things have proven a disaster because, in the world of trading, people tend to take small time increments and they want predictable outcomes, and when you get to social media and sentiment inferencing, those two factors don't apply."

The Bottom Line

• Speed is not the be-all and end-all when it comes to getting ahead of the game.
• Cost management and risk management are equally important considerations.
• The jury is still out on the potential benefit of sentiment analysis and skepticism abounds regarding using it as part of a trading algorithm.

Steve Dew-Jones caught up with Peter van Kleef after the panel discussion for a more in-depth discussion. Click here to see what he had to say.