Tradewell and Perspective – Modernizing Bond Trading with Real-Time Intelligence
Tradewell, a rising player in the corporate bond trading software industry, is revolutionizing how investors trade with its market-leading, web-based trading technology. Their goal is simple:
Enable seamless bond trading by consolidating fragmented data and providing actionable intelligence to human traders.
Tradewell has enabled their clients to trade billions in U.S. investment grade and high yield bonds. Recognized as the Futures Industry Association's 2024 Innovator of the Year, the company enhances market efficiency, liquidity, and trading outcomes. By unifying multiple real-time market data and consolidating liquidity, Tradewell streamlines corporate bond trading—providing traders with both comprehensive market intelligence and full control over their execution strategies.
In this case study, we explore how Tradewell adopted Perspective, our high-performance real-time data visualization tool, to solve critical challenges in data presentation and workflow efficiency.
The Challenge: Real-Time Data Processing in a Zero-Latency Market
Corporate bond trading remains a data-intensive industry reliant on human decision-making, unlike the fully automated equities market. In this realm real-time data models paired with data-rich, robust dashboards give traders the ability to detect and execute on market trends.
“In trading, stale data is simply not an option. We pull in data from multiple streams, clean it, enrich it, and join it in real-time before running it through our data models to generate actionable insights. By the time the data is ready, we can’t afford to waste even milliseconds on visualization—pixels need to render instantly to give traders the edge they need to execute in the market.”
– Michael Wong, CTO, Tradewell
To deliver a state-of-the-art tool for their users, Tradewell has made several key technical decisions:
- Seamless Integration of Real-Time and Historical Data: By using their open-source Python library, Beavers, Tradewell efficiently processes real-time data streams from Kafka alongside historical data stored in S3. This dual capability enables the replay of historical data for retraining and fine-tuning of models while delivering real-time insights on current data streams. Beavers handles micro-batch updates within directed acyclic graphs (DAGs), allowing the same complex DAG logic to be used in both real-time and historical modes. This provides a clean separation of the business logic and I/O operations.
- Real-Time Data Visualization: Traditional BI tools often introduce latency and lack the interactive capabilities essential for swift decision-making. Tradewell’s platform addresses this by providing near zero-latency data rendering, enabling traders to execute decisions ahead of the market without delays caused by data visualization processes.
“In our industry, real-time insights are non-negotiable. Many tools out there, including traditional BI platforms, simply can’t handle the demands of high-speed, actionable intelligence. Perspective is different—it gives us the speed, efficiency, and real-time capabilities we need to support our operations without compromising performance.”
– Michael Wong, CTO, Tradewell
Beavers: Tradewell's Real-Time Data Processing Engine
Beavers is Tradewell's open-source Python library designed for real-time stream processing. It provides a simple yet powerful framework for building complex data pipelines that can process both streaming and historical data.
https://github.com/tradewelltech/beavers
The Solution: Optimizing Client Activity Dashboards with Perspective
Tradewell leverages Perspective in its internal client support dashboards to monitor client activity and market data in real time. By integrating Perspective with Beavers, they efficiently process real-time data streams from Kafka and transform them into Arrow-based tables. This setup allows Tradewell’s users to interact with complex datasets, such as aggregated trading volumes and client activity logs, at lightning speed.
Here's how Perspective strengthens Tradewell’s tooling and technical architecture:
1. Integration with Python-Based Workflows
Tradewell's machine learning models combine data from multiple sources in real-time through Kafka and Beavers. Perspective seamlessly integrates with this Python-based solution by accepting updated Pandas and Polars DataFrames for visualization. This compatibility allows easy embedding of data visualization in the front-end application without requiring custom backend or frontend code.
2. Efficient Real-Time Data Updates
Perspective utilizes Apache Arrow to handle real-time data updates efficiently. By transmitting only the deltas—changes in the data—it minimizes bandwidth and I/O requirements. Unlike traditional tools that require full table refreshes, this approach ensures traders have instant access to current information without delays from complete dataset transfers—a crucial advantage in the fast-paced bond trading environment.
“In terms of performance, especially when it comes to updating in real time, Perspective is much more efficient at handling delta updates versus other plotting libraries that would update the entire table. That wasn’t scalable for us at all.”
– Arthur Andres, Lead Engineer
3. Simplified Development
By offering pre-built visualization components, Perspective eliminated the need for extensive frontend development. Tradewell could add new datasets by simply creating Arrow tables on top of Kafka streams, bypassing the need for additional endpoints or inefficient SQL-based database tables. Perspective views are highly interactive, eliminating the need for custom frontend and backend development solutions.
“When developing tools, we aim to avoid reinventing the wheel. We just need something that does the job, does it well, and is easy to put in place… Perspective removes the overhead of building custom visualization components from scratch.”
– Arthur Andres, Lead Engineer
4. Local-First Architecture for Scalability
Adopting a “local-first” architecture, Perspective shifts computational tasks to client-side rendering. This design reduces backend server load and eliminates scaling bottlenecks, allowing Tradewell to support a large user base without substantial infrastructure investments.
Conclusion: A Partnership Built on Innovation
Tradewell's journey exemplifies how real-time data tools can transform complex industries. Through its real-time, scalable, and interactive data visualization capabilities, Perspective enables Tradewell to focus on its core mission: empowering traders to make smarter decisions in a dynamic market.
As Tradewell continues to innovate, Perspective remains a key partner of their technical ecosystem.
