I shouldn’t be surprised that low latency—which at the end of last year looked like it was ready for a Lance Armstrong-like fall from grace, after trading firms began publicly questioning the rising cost for ever-decreasing benefits—has so quickly returned to the agenda in 2013, judging by the content of this week’s issue.
Illustrating that firms still place great value on being able to quantify latency and how it relates to execution performance, Corvil will this week announce another deployment of its latency monitoring technology, while San Francisco-based developer 4th Story has built a tool that analyzes algorithmic execution performance against trade data benchmarks, and Wilmington, NC-based appliance vendor Cape City Command has released a free Latency Evaluator tool to identify routing improvements that clients could achieve by using its full-function product.
Meanwhile, data and trading infrastructure supplier Fixnetix has partnered with Netherlands-based network specialist Custom Connect to offer connectivity between NYSE Euronext and Eurex, using a microwave network that slashes latency by up to 40 percent over fiber. This kind of latency reduction is still so cutting-edge that Olav van Doorn, co-founder of Custom Connect, says he only knows of two other microwave networks serving this route—both built by latency-sensitive trading firms exclusively for their own use.
And it’s this exclusivity to the wealthiest players that has led regulators to scrutinize latency and its enabling technologies following high-profile events such as the Flash Crash. A colleague asked me last week whether any moves to introduce a “minimum latency” level—i.e. holding up fast orders, so as to not discriminate against slower traders—would be a feasible move to level the playing field. The concept of holding orders and executing everybody’s trades at set time intervals isn’t new (see the Open Platform from the May 11, 2009 issue of IMD, where Michael Mainelli argues that periodic auctions would reduce latency issues—albeit in the interests of lower power consumption and environmental awareness, since datacenter power consumption now accounts for two percent of global carbon emissions, and many waste 90 percent of the power they take from the grid).
What piqued my interest, though, was how this would affect technology innovation, much of which has been driven by the need for low latency in recent years. Because one might argue that if you set an arbitrary latency “standard,” what incentive is there for firms to beat that time? And from vendors’ point of view, what incentive is there for them to invest large amounts of capital on research and development to produce faster systems?
Actually, there should still be plenty of incentive, though the focus will shift from trying to be fastest because speed alone will win the race, to being faster so that more tasks can be performed within any hypothetical “minimum latency” time—because the more analysis and risk checks you can perform within that prescribed time, the better-placed you will be to not only execute a trade, but execute the right trade.
In short, the competitive aspect will shift from just achieving the best price and hoping to make a quick buck from short-term price movements, to incorporating more fundamental analysis into trading decisions to capitalize on what promise to be the most profitable and least risky short-term movements. So potentially, your competitiveness will no longer hinge only on latency, but crucially, on what you do during that latency—regardless of whether it’s imposed by regulators, marketplaces or by components of the data and trading infrastructure that hit the physical limits of what they can achieve, giving you the opportunity to perform other functions faster and in greater numbers—and how you combine latency and analytics.
See, Lance, it’s not just about being fast—it’s what you do while being fast.