High-Speed Trading: My Laser Is Faster Than Your Laserby
A few weeks ago I wrote about Project Express, a new fiber-optic cable being built across the Atlantic that will give a select number of high-frequency traders a tiny speed advantage in trading times between New York and London. Currently, data take 64 milliseconds (give or take a few fractions of an eye blink) to travel round-trip between New York and London along a cable built in 1998 called the AC-1.
According to its New Jersey-based operator, Hibernia Atlantic, the $300 million Project Express will be 5.2 milliseconds faster than the AC-1, with an execution time of 59.6 milliseconds. That will make Project Express the world’s fastest transatlantic cable when it opens in 2013 and the first to achieve round-trip trading speeds of less than 60 milliseconds. Unless someone beats them to it.
As of this morning, it appears someone will. A small company called Perseus Telecom, in partnership with a subsidiary of India’s big telecom company, Reliance Communications, has announced the launch of QuanTA, a fiber-optic cable stretching from Long Island to the U.K. with an expected round-trip execution time of less than 60 milliseconds by the end of 2012. Rather than build a brand-new cable like Hibernia-Atlantic did, Perseus made improvements to an existing cable called the FLAG Atlantic-1 North, or FA-1 North, a small portion of a 17,000-mile underwater fiber-optic cable stretching from the east coast of North America to Japan. Until now, the FA-1 North was the second-fastest transatlantic cable after the AC-1.
To make the FA-1 North faster, Perseus first upgraded the cable’s “submarine optical systems,” which essentially means equipping it with faster lasers. The company also improved the backhaul systems connecting the core cable to various land-based subnetworks that spread to trading exchanges and data centers. It will next insert a giant router, or branching unit, a few hundred miles off the coast of Nova Scotia to build a shorter route to New York. With the help of submersible vehicles, a grappling hook hauled the cable off the bottom of the North Atlantic about 10,000 feet below the surface and inserted the branching unit, described as a Y-shaped device roughly the size of a conference room table.
The result will be a shorter cable powered by faster lasers. It cost $10 million to upgrade the lasers and the back-end connections. Asked how much it will cost to insert the branching unit and shorten the cable, Perseus Chief Executive Officer Jock Percy offers the following opaque calculation: It is 125th the cost of the new length of cable. Percy will say neither how much that new length costs nor how long it is. Just like the high-frequency trading industry it serves, the business of building submarine fiber-optic cables can be secretive—and highly competitive.
Although Perseus announced in January it was expanding its footprint in the data center hub at 60 Hudson Street in lower Manhattan, the company built QuanTA on the sly, announcing the project after it was done. That’s what Spread Networks did with the Chicago-to-New York underground trading cable it completed in 2010 after three years of boring through 825 miles of mostly rural, mountainous terrain in secret. Like Spread Networks and Hibernia-Atlantic, Perseus won’t disclose the identity of the trading firms it charges to use its cable, nor will it say how many there are or reveal its fees. CEO Percy will say his new project is highly cost-effective and he’s able to pass savings on to speed-trading clients.
“Market participants are always looking for advantages,” says Percy, meaning that speed traders continually look for ways to trade faster. “The cost of that advantage, though, is significant. The success of a trading strategy relies on how effectively people are able to achieve those last few milliseconds, and so a cost-effective way of delivering [faster speed] is really valuable, because just throwing money at the problem doesn’t solve it.”
The problem these new cables are solving is one of speed, not capacity. There’s plenty of fiber-optic capacity connecting most of the world’s big trading hubs, thanks to the fiber boom of the 1990s and early 2000s. Those cables were built before the era of high-frequency traders, so they rarely adhere to the shortest distance between two points: a straight line. Other than improving the caliber of the lasers shooting beams of light through these miles of cables, the only way to make them faster is to make them shorter.
Percy reckons that through a combination of improved lasers and shortened cables, the day may soon come where traders can execute a trade between New York and London at close to 40 milliseconds. Anything faster is physically impossible, save for drilling through the planet.
Assuming Einstein’s theory of relativity is correct, which the Large Hadron Collider in Europe recently reaffirmed, there’s no going faster than the speed of light, about 300 million meters per second. Since the surface of the earth is curved, a cable running along the bottom of the ocean isn’t flat. To straighten it, and thus shorten it, you’d have to drill through the earth’s crust. “If you did that you could get below 40 milliseconds,” says Percy.
So when will he finish that project? “No comment.”