At our 2024 Capital Markets Summit in London, B2C2’s head of quant development, Jad Sarmo, shared invaluable insights into building a scalable, high-performance quantitative research and trading system.

Founded in 2015, B2C2 provides reliable liquidity, enabling access to crypto for global exchanges, banks, brokers, fund managers, and more. Renowned for its innovative technology, sophisticated trading algorithms, and robust risk management systems, B2C2 leads the evolving digital asset ecosystem by bringing traditional market best practices to the dynamic world of crypto.

Whether you’re a data systems leader or a quant looking to explore cutting-edge tools and strategies, Jad’s insights are not to be missed. Watch his talk below or continue reading as we outline six vital lessons for achieving quantitative excellence.

You’re only as good as your data

Jad emphasizes that data quality is the foundation of successful quantitative systems. B2C2 processes vast amounts of data from diverse and sometimes inconsistent sources, ensuring it is hydrated, normalized, and stored for low-latency, high-throughput access.

“We need a high-performance ingestion solution to capture data from all regions, encompassing both cloud and on-premises sources,” Jad explains, emphasizing kdb’s versatility in managing structured, unstructured, streaming, historical, and real-time data.

“As a vector-native database, KX enables us to efficiently record every market tick, trade, and user action in a seamless time-series format.”

Beyond regulatory compliance, high-quality data is essential for advanced capabilities like AI and machine learning. “Everyone wants to do machine learning, everyone wants to do AI,” Jad says. “But there’s no point spending all your time on [a] model if your data isn’t high quality.” B2C2 employs continuous improvement strategies, including versioning, to enhance both data and model reliability. “It’s a cycle,” Jad notes. “Improved data quality leads to better models, which in turn improves data quality.”

Get ahead in the cloud

Initially, B2C2 relied on vertical scaling to manage terabytes of time-series data. “We just added more RAM and CPUs,” Jad recalls. “But eventually, we hit AWS’s vertical scaling limits and realized we needed to adopt a horizontal scaling approach.”

By integrating AWS FSx for Lustre with kdb+, B2C2 has developed a highly efficient and scalable infrastructure that can compete with the largest firms. Cloud-based resources provide seamless scalability for both compute and storage. Jad describes it as “an incredibly efficient way to achieve virtually unlimited storage with built-in compression and redundancy, allowing kdb+ to fully optimize CPU usage for research purposes.”

Combine tools for efficient quantitative research

For Jad, combining tools like PyKX with kdb+ is key to accelerating quantitative research. PyKX bridges Python and q, offering the flexibility of Python’s machine learning libraries alongside the computational power of q.

“That’s very powerful for us because…when you mix the two, you get the best of both worlds,” Jad explains. Researchers use Python for machine learning while relying on q for intensive computation. Tasks like as-of joins in q are 200 times faster than Python, and columnar processing simplifies calculations such as time-weighted average price.

Reproducibility, version control, and shared q libraries further ensure transparency and innovation. “You need a common view of everything you’re discussing and building,” Jad emphasizes. “That’s how we’ve created a high-quality research environment.”

Use agile visualization for real-time insights

Agile visualization through KX Dashboards provides B2C2 with real-time, actionable insights. “We started building dashboards for Sales, and they loved it from day one,” Jad shares. “They could drill down by region, client type, currency, instrument—anything.”

The ability to visualize large datasets quickly offers a competitive edge. “We ended up building hundreds of dashboards across all areas because it’s so simple and fast,” Jad says. “We haven’t seen anything as powerful, especially for displaying vast amounts of data in real time.”

Be ready for market volatility

The volatility of crypto markets demands resilient systems and automation, and Jad credits kdb with enabling B2C2 to operate smoothly during major market disruptions.

“We’ve been able to automatically hedge in volatile markets,” he says, citing major market events. “As a liquidity provider, you’re market-making, hedging, and managing client flow. We’re able to handle it all very smoothly thanks to the reactivity on the plant.”

Future-proof your quantitative systems

“Our vision was to create a data infrastructure and research framework that could support the demands of crypto markets,” Jad explains. Scalability, flexibility, and performance remain central to B2C2’s approach as data volumes grow and client expectations rise.

Jad emphasizes the importance of modular systems that scale logarithmically rather than linearly. “If you multiply [your] number of clients by ten, ideally your complexity only increases by a factor of two,” he says. This approach ensures that new features and dimensions can be integrated without breaking existing functionality.

“In a nutshell, we’ve built a robust, scalable, and agile infrastructure that helps us maintain our market leader position,” Jad concludes.

Now explore how KX can enhance your own quant research and accelerate time to insight with powerful analytics, or watch our recent webinar discussing three innovative quant trading applications.