How Do Agile Architectures Reshape High-Performance Trading Systems

Agile architecture in trading systems is no longer a futuristic aspiration—it’s a necessity. At the recent panel during TradingTech Summit 2025 in London, experts from Citi, HSBC, Rapid Addition and velox explored the realities of transitioning from monoliths to microservices, the myths surrounding agile, and what true agility means for high-performance financial systems. In this article, we outline the key takeaways from the discussion moderated by Irina Sonich-Bright, MD at UBS.

Is Agile Architecture More Than a Buzzword

For many technologists, “agile” immediately conjures images of post-it notes, standups, and sprints. But in the context of trading systems, the word has a deeper, more demanding meaning. Agile design is about navigating the trade-offs between cost efficiency, adaptability, and performance at scale. There’s no free lunch.

The challenges in trading systems are unique. Unlike streaming platforms or e-commerce sites, trading platforms must balance extreme performance with reliability, explainability, and regulatory compliance.

Microservices as the Engine of Agility

For Deepak Dhayatker, CTO of Rapid Addition, microservices are synonymous with agility in architecture.

“For me, microservices is a preferred term than agile architecture. Event Driven, Stateless, low-latency messaging, using off-heap memory, optmising network, cpu OS, using lock-free data structures are the fundamental pillars of High-Performance Systems.”

Dhayatker emphasised the importance of removing non-determinism from systems. By minimising garbage collection and using patterns like Disruptor and off-heap memory allocation, systems become more predictable and responsive. In trading, where microseconds matter, that predictability is gold.

Transitioning from Monoliths

Despite the promise of microservices, migrating from legacy monoliths isn’t a walk in the park.

“You have to accept a hybrid coexistence of monolith and microservices. Migration takes time. It’s costly. And it can span months or even years,” Dhayatker explained.

He advocates for the use of the Strangler Fig Pattern—incrementally extracting functionality from monoliths and routing it through modern services. Deepak also warns of hidden challenges related to maintaining consistency, monitoring, and dealing with distributed data.

Deepak concluded “The golden rule to architecting a high performant Trading systems is having the right people, tools and dedication to find bottle necks in your applications and fixing it.”

Why Big Banks Need Predictability Over Pure Agility

At Citi, agility isn’t just about speed—it’s about control.

“We want testability, explainability, and a unified, engineering-driven approach. Regulators require predictability,” said Anthony Warden, Head of High Performance Architectures at Citigroup.

Warden points out that unlike consumer tech giants, trading platforms don’t have the luxury of throwing systems away and starting over. The cost of mistakes is too high. Agile architecture in trading must combine the bleeding edge of performance with the stability of enterprise-grade systems.

The People Factor: From Projects to Products

Agility isn’t just architectural. It’s cultural. At HSBC, product-focused roles are replacing traditional IT structures.

“It’s about creating cross-functional teams that break silos and deliver end-to-end business value,” said Minhaj Ahmed, Product Owner for MSS Equities.

Ahmed highlights how banks are shifting from output-oriented projects to outcome-oriented product thinking. This means tighter collaboration between business and tech, faster feedback loops, and shared accountability.

Monoliths Aren’t Evil—Bad Implementation Is

Despite the push toward microservices, not all panellists were ready to throw monoliths under the bus.

“It’s easy to say legacy is bad because it’s a monolith, but that’s not true. Legacy is bad because of poor implementation or tech debt,” said Jon Butler.

He proposed the idea of the micro-monolith—a compact, well-designed monolith with clear separation of concerns that avoids the overhead and complexity of distributed systems when unnecessary.

Scaling Success Through Patterns and Reuse

One major takeaway is that organisations need structures to share knowledge and successful components across teams.

“Some of the best high-performance developers are terrible communicators. So we facilitate, harvest, and share best practices to scale success,” said Warden.

At Citi, once a high-performing component is built, it’s supported and shared so teams can reuse it and focus their engineering on true differentiators.

Conclusion: Context is King

Agile architecture in trading systems isn’t about chasing trends. It’s about making thoughtful, context-aware decisions. Whether you embrace microservices, evolve your monolith, or create something in between, the ultimate goal is to build systems that are fast, reliable, testable, and—most importantly—aligned with your business strategy.

“You can’t come at these things with a solution. You have to come back to the problem,” said Butler.

As firms race to modernise, the ones that succeed will be those who understand that architecture is not just about code—it’s about people, process, and purpose.