I recently spoke on a panel discussion with the Securities Technology Analysis Center (STAC) on the use of infrastructure as a competitive advantage. The event offered fresh thinking on what it takes to manage high-frequency, low-latency trading environments - so I wanted to share some best practices for organization, monitoring, and how to make insights operational.
Tip 1: Organize
It is critical for firms to be able to monitor a transaction for the entire lifecycle of the trade - from the trader’s keyboard (or algorithm), to all of the different systems through which it passes on the way to the exchange, and through execution and settlement. If something doesn’t go to plan then it is essential to have a single point of reconciliation, where all of the facts leading up to the trade can be assembled, and the root cause of the issue can be resolved.
The mission is not just to avoid downtime, rather to minimise latency. We can use some of the same systems, but need to develop new methods of monitoring and managing systems, and identifying issues that can cause latency. These can be as simple as version control over software patches, malware or viruses, or as obscure as mains electricity interference or radio-frequency interference, or the routing of cables. It all makes a small difference. Marginal gains matter.
Tip 2: Monitor
Firms looking to squeeze the last drops of latency out of a system will need to consider how they approach full-stack monitoring, where every possible variable is traced, measured, and logged, so that problems can be identified, and optimization can take place across all the layers of software applications, algorithms, networks, appliances, and venues. Ideally, having a consolidated application where a firm can reconcile traces, metrics, and logs would be ideal. Firms should consider how they measure performance across different organisations and have techniques in place to measure things like the roundtrip time between an application and an exchange. Flexibility is critical, as things change so often. Firms can use this trading infrastructure to measure critical trading metrics like ‘hit and fill’ ratios, and these metrics can feed into smart order routing algorithms.
Tip 3: Make insights operational
Business analytics is commonplace, but not all firms have managed to make the insights gained through analysis available to everyone who needs them, and at the right time. The trick is to bake the analysis into the business process, to make it fully operational. That sounds easy, but requires a real-time approach to analysis, and a platform that supports real-time analysis. Not everything will be fast enough to keep up with the current pace of trading, but having analysis in place where a human is involved can pay dividends - for example where decisions are required regarding algorithm selection or limits and exposures.
Check out the STAC event on-demand to listen to the panel discussion and to explore solutions to other technical challenges in trading.
At Splunk we have been helping firms for years in their quest to reduce down-time and unplanned outages, even to the point where we can forecast an outage and proactively deal with a problem before it happens. We recently published a new eBook 15 Ways to Use Splunk for Trading Operations that addresses the key IT, security and business operations use cases you need to know download your complimentary copy today to learn how Splunk can support your firm’s efforts to leverage insights from the trading infrastructure.
We’d love to talk to you about your trading operations, low-latency or otherwise - so please reach out!