Shaving response times by nanoseconds can produce value in high-frequency trading, but the cost of achieving that size of an improvement in latency, in resources and time, can be too high for trading of more complicated types of securities, according to low-latency services and market access platform providers.
“High frequency traders are responding at a level of 200 nanoseconds,” says David Snowdon, chief technology officer and co-founder of Metamako, a Sydney-based low latency technology company. “If you want to get it down to 190, 193 or 195 nanoseconds — get those last few nanoseconds out of the system, you have to measure very accurately what time events happen on your network, so you can then understand what your response time was.”
Firms also should look at variance in their response times around the 200 nanosecond level, according to Snowdon. “Being able to measure that variance is extremely important to exchanges, to guarantee that they’re providing fair access to the market,” he says.
While frontiers of speed can still be trimmed, as Snowdon states, having a certain level of speed and a certain low level of latency has become a given in the industry — and one that need not be improved upon, as Dan Hubscher, director of strategy at Object Trading, a direct market access platform, describes.
“Speed is still important in that for anyone who has a strategy that depended on speed, they can’t get slower and still be profitable,” he says. “They still have to maintain that minimum level. The problem for most traders is that they’ve reached a commercial limit, where it doesn’t pay. It doesn’t return dividends to get it any faster.”
Furthermore, trying to lower latency when dealing with asset classes other than equities requires clearing additional hurdles, according to Hubscher. “Latency arbitrage on multiple exchanges doesn’t really exist in futures,” he says. “Trading a wider array of products across many more geographies — different types of derivatives and asset classes — pushed the game into one of scale, bringing in cost control.
“When you’re scaling up to different destinations, especially if you still need some degree of low latency, managing pre-trade risk, positions and exposures … is harder if you’re constantly adding new things that aren’t familiar.”