The role of the execution management system (EMS) in sell-side trading is expanding as banks seek to modernise their technology infrastructure while retaining control over proprietary pricing models, algorithms and analytics. Increasingly, the EMS is becoming a central control plane through which these capabilities can be integrated into electronic trading workflows, allowing firms to introduce new functionality without replacing their existing systems.
According to Dan Tsou, Global Head of Sell-Side EMS at Bloomberg, the shift reflects several converging trends, including the growth of systematic trading, increasingly demanding latency requirements and the integration of AI into trading processes. These developments are changing how banks evaluate execution technology, with architectural flexibility becoming an important consideration alongside traditional functionality.“Increasingly, our clients are asking first how they can integrate in addition to asking about our features,” he tells TradingTech Insight. “If you look back 10 or 15 years, the questions were: What can your EMS do? Which markets are you connected to? What algorithms do you have? Today, those things are increasingly table stakes. People expect an EMS to have the connectivity and trading capabilities. Now they’re increasingly looking for differentiation in terms of how easily they can integrate their own IP, such as their proprietary pricing models, algos, and analytics into the EMS over time, across different desks.”
From Execution Platform to Control Plane
Traditionally, execution platforms have operated alongside separate systems handling pricing, risk management, position keeping and other trading functions. As automated trading becomes more prevalent, these components need to interact more closely, often within increasingly demanding performance constraints.
The control-plane model coordinates these capabilities within the trading workflow. Core services such as limit checking and position management can be integrated alongside latency-sensitive pricing engines and execution algorithms, with individual components operating according to their own performance requirements.
The importance of low latency is also extending beyond asset classes historically associated with high-frequency trading. Tsou notes that colocation and performance optimisation are becoming considerations in rates trading, reflecting the growing role of systematic execution in fixed-income markets.
This creates an architectural challenge for banks seeking to introduce more sophisticated trading strategies while maintaining connectivity to multiple markets, internal systems and risk controls.
A Changing Buy-and-Build Calculus
Banks are increasingly combining vendor-provided infrastructure with their own proprietary trading technology, rather than building everything themselves or relying entirely on external systems. This allows them to spend less time and money maintaining core execution infrastructure and focus their development efforts on pricing models, algorithms and analytics that give them a competitive edge.
“Increasingly, clients are looking to differentiate through their own proprietary IP,” observes Tsou. “If you’re going to have a specialist pricing model, that’s going to be better than the general methodologies people use. So we do see our role as providing the underlying infrastructure and connectivity that enables that differentiation.”
Bloomberg’s acquisition of Broadway Technology in October 2023 brought the Toc, Broadway’s proprietary software for integrating trading applications, data sources and execution logic within a low-latency architecture. The platform has since been rebranded as Bloomberg Trade EMS (TEMS).
Bloomberg has now expanded the platform into Europe, with growing interest from Asian institutions. Tsou cites one client that increased its product coverage twentyfold by consolidating curve infrastructure within the EMS.
For larger banks, the architecture allows different desks to retain their own pricing models, risk methodologies and execution strategies while sharing common infrastructure, reducing the need for separate trading platforms.
AI Moves into the Trading Workflow
Artificial intelligence is creating additional integration requirements as banks explore ways to incorporate new analytical capabilities into existing execution processes.
In automated trading, AI-driven models can adapt quoting behaviour to changing market conditions, adjusting prices or trade sizes in response to volatility and other factors. Similar capabilities are emerging in workflows where traders retain direct responsibility for execution decisions.
“Even when a trader is trading manually, we’re starting to see clients incorporate new and novel analytics into the trade workflow, through the pop-up screens they can create and customise on Bloomberg,” says Tsou. ‘When a trader sees a pop-up, instead of just seeing the prices and reference data, they can now potentially receive AI guidance from their own internal models, such as suggested price adjustments or whether a trade or client may be risky. That guidance might say: We’ve looked at your recent trades with this client and a range of other factors. Based on that, here’s the spread we’d suggest to increase your probability of winning the bid. So we’re starting to see AI signals being integrated not only into automated workflows but also into the tools that help traders make decisions on more difficult trades.”
These applications create different requirements for the underlying execution architecture. Models developed independently of the EMS may draw on proprietary trading data, client histories and internal analytics, while their outputs need to be available within the relevant trading workflow.
The distinction between fully automated execution and trader-assisted decision-making also affects latency requirements, user interaction and operational oversight. Integrating these capabilities into existing systems allows banks to introduce AI-driven functionality without necessarily changing how individual desks manage trading decisions.
Modernisation Without Wholesale Replacement
The growing ability of banks to develop their own applications is influencing how they approach trading platform modernisation. Rather than replacing entire technology stacks, firms can introduce new capabilities incrementally, varying their approach according to the requirements of individual desks and asset classes.
Some institutions are also reconsidering the traditional relationship between execution infrastructure and the trader desktop, choosing to retain responsibility for their user interfaces while sourcing the underlying execution services from an external provider.
“With the increased use of AI, people can build more than they thought they could before,” notes Tsou. “And when they build things, they’re going to need to integrate them into the electronic trading workflow. Choosing the right control plane to build around has become a theme with almost every client we’re talking to. Some clients have said: ‘We want to build all our own user interfaces, but we don’t want to build any of the backend or connectivity.’ Bloomberg can support that model – effectively, a headless EMS – with the EMS providing the underlying orchestration, connectivity and workflow while clients retain flexibility around the user experience.”
Separating the user interface from the execution infrastructure provides banks with additional flexibility in how they develop and deploy trading applications. It can also support a more gradual approach to modernisation, allowing new components to coexist with established systems while individual desks transition at different rates.
“Also, when banks modernise, they may have different strategies for different desks over different years,” says Tsou. “Perhaps this year they want to focus on one asset class, but they still need to support the others. We can play a dynamic role that shifts with them as their priorities evolve across desks and asset classes.”
As banks assess their technology priorities heading into 2027, the ability to accommodate different development and deployment strategies is becoming an important consideration. The emergence of the EMS as a control plane offers one approach to reconciling the demand for greater automation and proprietary differentiation with the practical constraints of modernising complex, established trading environments.
Subscribe to our newsletter
