5 Tips to Find Alpha in Your Infrastructure
The demands placed on financial infrastructure today are more challenging than they have ever been in the past. Firms now realize they can no longer view the world through siloed single-asset infrastructures and are looking to implement new multi-asset trading systems. Increasingly complex trading and workflow requirements make it almost impossible to scale in real-time, placing undue pressure on the underlying infrastructure to support the business. In addition, a renewed regulatory focus on operational risk controls is forcing firms to re-evaluate infrastructure-level safeguards. At the same time, business requirements are expanding the connectivity layer of the infrastructure as a whole.
In the new world of multi-asset strategies, access to the largest number of trading venues over the fastest fiber can mean the difference between success and failure for trading. This shift in trading requires a new type of infrastructure that’s not only capable of providing the level of flexibility called for, but that can also produce alpha that would be lost in the legacy infrastructures of yesterday.
Firms are now looking to gain operational advantage from their infrastructure and can follow these five tips to find it:
1. View the infrastructure as a whole
A major source of error occurs when firms fail to view the entire infrastructure as a whole. The network is connected to the servers, which are connected to the operating system. Those connections require a deep understanding of how these components have an impact on one another. If the infrastructure is built or managed by disparate teams, discovering errors will be challenging. Alpha can be found by ensuring that the infrastructure is handled by a unified team that can identify cross-system failings early, instead of discovering them too late and losing large investments in patching up a preventable problem.
2. Colocation is a must
Often colocation is viewed purely as a speed advantage, necessary only for the most latency sensitive firms. Although colocation reduces baseline latency, it also allows for an increased level of connectivity -- since customers and venues gather in a single center, and connectivity is reduced to simple "cross connected" fiber. Without colocation, firms will lose speed as well as the alpha that comes from sharing close quarters with neighbors in a datacenter.
3. Build only when necessary
A decade of innovation in electronic trading has brought a world of new choices in terms of commercial off-the-shelf offerings for financial services. FIX engines, feed handlers, ticker plants, OMS, and EMS systems are often best purchased rather than built. This rings true for infrastructure as well. New technology has allowed the industry to move past the days of spending millions of dollars, and many months building out inflexible infrastructure. The multi-asset strategies emerging will not allow for this outdated model, since to test trading in FX, for example, a firm would have to build out an entirely new infrastructure focused specifically on FX trading. This is very unlikely to be approved if it is just for a trader’s experimental purposes.
4. Instrument or die
If every component of the infrastructure cannot be measured, diagnosing the system will be a huge problem. Companies can use Dtrace, a real-time OS level instrumentation framework, to diagnose any level of the system in real time by pinpointing which programs are causing performance issues. Troubleshooting performance problems can be a burden, but firms can utilize technology such as Dtrace to determine the cause of the problems so they can be isolated and corrected quickly.
5. Have a low-level, big red button
Failures are inevitable, and it's important to ensure that in the event of an extreme failure (e.g. a runaway algorithm that has made a system unresponsive) there is a low-level stop mechanism that can be implemented to protect the system. The kill switch is a hot topic that the industry is currently debating. A kill switch that turns off the entire infrastructure would be incredibly costly to the business that could achieve better operational alpha if it could isolate the problem. A zoning approach that allows the company to pause only infrastructure that is failing will allow for incoming information to enter the system while outgoing information is halted.
As the financial industry continues to change and trading strategies evolve, infrastructure is not just about improving latency. Firms can look to their infrastructure for the flexibility and scalability to support multi-asset trading strategies, as well as for operational advantage. Speed-to-market in launching new strategies, performance of that infrastructure to support those asset classes, and reliability and failover are critical aspects that must be considered to remain competitive in this new environment. They can serve as guidelines on how to gain operational alpha from your infrastructure for those of you looking to take the first step.
Jacob Loveless is a proven technology innovator with 15 years' experience in automated trading and advanced systems design. As co-founder and CEO of Lucera, Mr. Loveless is focused on introducing radical innovation in engineering that will change the future of financial ... View Full Bio
Tweet This