In previous blog posts, we have discussed the importance of trading infrastructure testing and the regulatory and operational impetus driving its necessity. But when push comes to shove, how easy is it to implement these complex systems? What types of testing solutions are available, and what are the biggest challenges to their adoption?
Today’s trading operations have become complex networks of applications often running on different technology platforms tied together into an overall workflow infrastructure. This diversity has led to a fragmented approach to infrastructure testing, with the result that many firms lack an overarching ‘enterprise’ perspective or coherent, holistic testing strategy. Instead, they rely on a variety different testing agents. The cost of managing all these disparate testing systems can be exorbitant. Firms can end up spending more time and money managing their complex testing environment than on the testing process itself, resulting in increased expense without a concurrent improvement in trading system reliability.
At a time when regulators are demanding better and more frequent testing of trading infrastructures, the complexity and cost of these processes continue to increase and the number of tests and their associated results is expanding rapidly. Analyzing all this data is a serious challenge – especially as the teams tasked with analyzing the data are often too busy running the systems themselves. As a result, even firms with a good testing program often struggle to reap the benefits.
Unsurprisingly, consolidating and improving this fragmented testing environment scenario present a number of challenges. From test design, test management, test configuration, and test results analysis, the goal is to add value without increasing cost. But before embarking on a wholesale rip-out-and-replace project, it’s essential to understand what’s required.
Although testing of algorithmic trading strategies tends to be the star of the show, there are several other forms of trading environment testing that are also relevant to both regulators and operators. Functional testing is a key element. For example, when a firm starts work with a new client, it must ensure its trading systems and connectivity to that client operate as intended. In another example, sell-side firms often need to support many flavors of FIX to receive orders from their buy-side clients, which requires them to create output messages in different versions of FIX to route an order to exchanges and trading venues – and this sometimes requires translations to native protocols. Most venues use proprietary messaging formats and protocols; add in the growing number of Systematic Internalisers (SIs) and the situation becomes even more complicated. All of this connectivity must be set up correctly in the trading environment and then tested before live trades can be undertaken.
Other types of testing include resiliency testing. Is the firm able to manage and inject faults into a trading system to simulate rare conditions, exception behaviours, and intended malfunctions? Performance testing (or throughput/scalability testing) is equally important. How many orders can the system process before it falls over? What is the level of jitter and inconsistency on virtual machines? Firms need to understand the impact of these issues on service levels, and this points to the need for repeatability.
Finally, audit trails are also an issue. To achieve MiFID II certification, firms must test whether their counterparties are sending what they say they are sending. Yet on many legacy testing systems, the absence of an audit function means there is no view of which tests have been performed, making it difficult to validate counterparty systems.
With so many diverse challenges and processes, the concept of the trading system as a single management layer is clearly invalid. Multiple trading systems and testing solutions operating within the same environment create high levels of complexity, and firms urgently need the ability to operate and control this highly disparate platform. Solutions such as VeriFIX offer the ability to manage and test on an enterprise scale, produce auditability and validation of test results, while at the same time embracing a wide set of testing tools and agents.