If you thought the revelations— perhaps claims is a more accurate term— that came out in the Michael Lewis book “Flash Boys: A Wall Street Revolt,” would dampen the enthusiasm for building faster technology to get trades to market and quotes from markets you would be mistaken.
The story of modern financial trading architecture has been an arms race to the fastest possible execution and recently Australian company Metamako has released a new configurable layer 1 Switch with a latency of 4 nanoseconds. It replaces the standard layer 2 Switch, which typically has a latency of 200-300 nanoseconds. What is a nanosecond? It is 1/billionth of a second. What is a layer 1 Switch? A switch is a computer networking device that is used to connect devices together on a computer network and layer 1 refers to a hub. The layer 1 Switch is placed in a co-location facility and sends out market data to multiple locations. The firm is not unique in offering the superfast layer 1 Switch.
It is aimed at proprietary traders that are either high-frequency traders or latency sensitive traders, which are not the same according to David Snowdon, founder of Matamako. High-frequency traders are not necessarily latency sensitive and latency sensitive traders are not necessarily high-frequency traders,” Snowdon says.
Some traders like to trade off of economic reports and it is important for them to see the initial quotes first. And some HFTs are not looking to cut every nanosecond.
Snowdon did express concern that the controversy over HFT could lead to regulations that could affect his business but says people will always want speed. In terms of fairness, he says the exchanges can help by creating more deterministic switches of their own to prevent any latency advantage.
He is a little bewildered by the controversy over HFT and the claims in Lewis’ book by Brad Katsuyama, founder of IEX Group. The story in Lewis’ book is how Katsuyama grew frustrated by seeing bids and offers disappear before he was able to execute large orders. “They weren’t doing their job right,” Snowdon says, who blames many of the problems attributed to HFTs to the market structure in equities.
Snowdon’s implication is that there were ways Katsuyama could have placed trades to not alert other market participants once they were sent from one exchange to another as part of the Reg NMS structure.
He pointed out that the old floor structure had latency advantages and there were folks who tried to get ahead of other traders. Electronic markets are just allowing this same activity.
Perhaps the larger question is, how low can you go? Latency wise that is. Four nanoseconds is pretty low.