This article was originally published on Intelligent Trading Technology on June 14, 2012.
The number of alternatives for low latency infrastructure to support global securities trading across all asset classes has exploded over the past several years. From improvements in servers and switching, to the fiber and now wireless networks that connect them, the arms race continues to escalate. A growing base of carriers has entered the low latency transport market to provide wide area networks connecting key liquidity centers worldwide. In many cases carriers are deploying brand new networks shaving handfuls of microseconds to multiple milliseconds at costs from several million to several hundreds of millions.
The Race to Zero: At What Cost?
Among the most well known of these network deployments are Spread Networks New York to Chicago fiber route and the pending Hibernia Express submarine cable route connecting NY and London. The latest developments include the use of low latency microwave to reduce latencies even further on key trading routes. The race continues unabated both within and between major financial markets, as carriers and the trading firms they serve work to one up each other with the next best network, seeking to capture the coveted low latency premium.
With these major investments come large increases in operating costs for those trading firms leveraging the coveted routes. Some routes have seen greater than a ten-fold increase in costs over the past 18 months alone. No sooner does a firm complete migration to a new route when the next better route is already being marketed, leading to a cycle of stranded investments both for the trading firms and often for the underlying carriers as well. With the increasingly larger investments required to participate in the most lucrative areas of high frequency trading, firms need to decide where their best overall returns exist and how to achieve those returns at the lowest cost possible.
Optimizing for Best Latency
Many firms are turning to more optimized means to achieve the best latencies at the best cost. Recognizing that most firms are moving the same market data between markets to make trading decisions, trading firms are increasingly migrating to low-latency market data delivery platformsto deliver raw or normalized market data feeds, still at the best latencies, but at considerably lower cost than building out and managing constantly moving latencies and trading venues on dedicated platforms themselves.
Another growing trend is the use of ‘relative latency’ to improve trading profits for lower frequency trading firms. The growth in high frequency trading has raised the overall awareness of the impact of latency across all electronic trading markets. The arms race supporting the largest HFTs operating on the basis of ‘absolute latency’ seeking coveted zero-risk returns, has created flow down opportunities for firms whose risk models don’t require the absolute best. These firms can improve trading profits through better fill rates vs. competing firms trading similar securities and risk models by improving their ‘relative latency’ to those firms.
One of the traditional ways for firms to improve this relative latency is to leverage proximity hosting, essentially decentralizing trading decisions. One of the major barriers to this strategy, however, is the cost of acquiring proximity hosting space, then building out and managing a dedicated infrastructure including the low latency WAN connectivity between markets. This has resulted in a growing niche of proximity hosting providers who help to manage much of the complexity and some of the costs associated with this approach.
The combination of low latency routes that are no longer the best, combined with the emerging generation of high performance cloud platforms promises to further reduce both the complexity and costs associated with proximity hosting and gaining a relative latency advantage for those firms that can benefit from it. The latest generation of cloud and virtualization technologies includes the built in security necessary to support complex regulatory environments, as well as improvements in performance that begin to mirror the capabilities of dedicated servers.
Low Latency Cloud Platforms
When these technologies are deployed on a platform and within an ecosystem that includes direct market access, raw and normalized market data, and access to best of breed front, middle, and back office trading applications, performance can exceed even some of the best dedicated infrastructures at a tiny fraction of the cost. The ability for a European trading firm to spool up a dedicated or virtual server and enable their trading applications within the US equities, options, futures, or FX markets has never been better.
So as the arms race continues at the high end of the HFT market, it is ultimately enabling trading platform providers to deliver some of the dividends to downstream market participants. Those firms trading their own risk models with counterparties around the world can now leverage a growing base of high performance trading platforms designed to deliver the full range of latency options, from absolute latency to relative latency, at a cost and level of complexity designed to optimize their trading strategies and overall profits.