Leveraging Latency
Million-dollar milliseconds
Hello everyone and welcome to the latest edition of GreySpark Insights.
Please do not hesitate to contact us with any questions or comments you may have. We are always happy to elaborate on the wider implications of these headlines from our unique capital markets consultative perspective. Happy reading!
With most financial products trading electronically, latency is now a top priority for proprietary traders and market-makers – regardless of asset class – when engaging with digitalised infrastructure for the first time. Latency is the total elapsed time between the event, or signal, that triggers a trading decision and the actual completion of that trading action. The origins of low-latency trading are traceable to the arbitrage desks of the 1990s, where speed of markets connectivity was directly used to ‘pick off’ mispriced products.
Latency is not consigned to only the trade execution itself. In fact, on exchange platforms, there are three key ways in which latency is generated:
The applications that interpret financial information and make trading decisions, equating to approximately 65% of total latency;
The middleware that distributes the various messages and signals, equating to approximately 25% of total latency; and
The network fabric that transports data from a market to the trading firm and then back to the exchange, equating to approximately 10% of total latency.
Of course, the faster an order reaches its destination, the more likely the firm will achieve best execution. Minimising latency across different jurisdictions is crucial to financial firms operating in global markets. There can be a latency of between 150-23o milliseconds between Tokyo and London depending on the chosen supplier and route, so firms can easily end up trading after other market participants have seen the real-time data. A millisecond difference in data access and trade processing times can equate to a significant monetary gain or loss. The estimated value of a one millisecond latency advantage for a brokerage firm amounts to more than USD 100 million annually.
In 2024, equities markets, globally, require low-latency trading capabilities because of the widespread use of algorithms and automated trading tools by different types of market participants, while fixed income and FX markets are best described as ‘latency sensitive’ as opposed to ‘latency dependent’ due to the extreme liquidity of some instruments or products and the tightness of bid-offer spreads in quote-driven brokerage environments. Outside of those markets, ultra-low latency is typically considered nice-to-have by most markets participants as opposed to being an imperative for a competitive trading franchise.
Capital markets trading is now considered by all industry participants as a global, 24/7 business model. Many organisations have built complex webs of network infrastructure and large data centres to support them, and – as a result – the ability of a firm/institution to leverage dedicated data centres, co-located physically together with various execution venues, using low-latency networks is commonplace. This allows trading firms to access financial data through data feeds that stream data right from the source as soon as it appears in the exchange server. Since the advent of electronic exchange platforms in the mid-1990s, the latency thresholds under which trading technology must perform to stay competitive are halving roughly every three years, and those thresholds are now at levels expressed in tens of milliseconds or less.
Ultimately, the drive for latency has become a competitive differentiator for various market participants, leading to the enhancement of colocation capabilities and services in capital markets. Early-stage colocation capabilities and services were rapidly commoditised by different types of digital markets infrastructure technology vendors – for example, equities markets exchange platform operators or cash FX brokerage venue providers – into a managed service that all markets participants could access.
GreySpark believes that the core elements of a managed colocation service are:
A Compute Platform – Specialised hardware and software components that are optimised for high-frequency trading;
Network Security – A set of technologies that protects the integrity and usability of infrastructure by preventing entry into the network of a wide variety of potential threats and their potential proliferation;
DMA & Raw Market Data – With DMA enabling the execution of trades directly within specific brokerage venues or exchange platforms, and raw market data used to enrich existing data related to historical pricing, for example, as well as the ability to derive from the raw data insights into all current, relevant information related to general market activity;
Co-location – The practice of renting space for servers and other computing hardware at a third-party’s data centre facility or server rack; and
Network Connectivity – Equipment used to combine, split, switch or direct packets of information along a telecommunication network.
For more information, please check out our latest research report here.