Smashing Through The Speed Of Light

The physicists among you will no doubt be shaking your heads at this headline, which, to be fair, is a bit of a push. Einstein's Theory of Relativity, although challenged recently by a few upstart neutrinos at CERN, still holds as a constant - the speed of light is the maximal velocity at which anything can physically move.

This becomes more pertinent when discussions of latency crop up ─ and let's be honest here, barely a conversation regarding the process of trading goes by where it doesn't. The fact of the matter is that we may be reaching the limits of what is physically possible when it comes to reducing latency for trades, in that we're simply pushing against the barrier of how fast things can move in a very real sense.

Earlier this year, I was at a press briefing for the launch of Citi's Velocity 2.0 product, their own high-frequency platform for foreign exchange (FX). The numbers that they quoted were, if you'll excuse the hyperbole, astronomical. Round-trip transaction times from London to New York were measured by them at around 52 milliseconds, a stunning level of alacrity when you consider that the speed of light in a fiberoptic cable, for that distance, is 18 milliseconds. Microseconds, of course, are now common parlance when talking about latency.

Plateau
Following on from that, last week I spoke at length with GreySpark Partners about their latest research, entitled Low Latency: Faster Than Light. Again, apologies to anyone with a physics degree reading this. During the discussion, there was talk about latency reduction being at a point where it might plateau, given the ever-decreasing distance between where we are now and Einstein's inviolable rule.

The term plateau is a strong one, and it's perhaps, in this context, not appropriate to the nuances of the technology. "Latency is always an issue as it's an arms race," Tim Carrington, global head of FX at RBS Global Banking and Markets told me a few days ago, when I put this to him. "The degrees of latency that we're trying to eliminate now are miniscule compared to the degrees of latency that we were trying to eliminate two years ago. You could argue it's plateaued, but I'd argue that while the degrees of latency that you're trying to eliminate are much, much smaller, the benefit of eliminating them is much, much larger. I don't think it's changed as such; it's just the numbers that have changed."

The fact of the matter is that we may be reaching the limits of what is physically possible when it comes to reducing latency for trades.

In many ways, it depends on the measurement of latency, which as we all know is an ongoing debate. But there's absolutely no doubt that it will continue to be a conversant topic, albeit one that refines itself as the conversation progresses. Co-location is a hot issue, but for many who perhaps aren't on the bleeding edge of speed but in the relative ‘safe zone' in between that and those who are too slow, the quality of cabling is perhaps a stronger draw than the actual geographic location. You might be able to put a server on the top of BT Centre, for instance, but if you're not using the appropriate materials in your connection, there's a chance that someone in Welwyn Garden City may just pip you to the post.

Diminishing Denominations
As always, technological innovation will continue to work around this and squeeze every possible nanosecond, femtosecond, zeptosecond and attosecond out of latency that it can. As Carrington said, it's a numbers game. Although, as Anil Prasad, Citi's global head of FX and local markets admitted, quite a few people in this space might be keen for Einstein's theory to hold true, whatever the neutrinos may think.

From everyone here at Waters and Sell-Side Technology, have a great weekend and a successful week.