BMLL, the independent provider of historical order book data and analytics, is joining forces with Features Analytics. The deal gives the company that hunts down market manipulation and abuse an access to BMLL's layers of market data order book records spanning global equities, ETFs, futures and US equity options. The aim is to build a new class of surveillance benchmarking products.
The core idea is blunt: most firms running trade surveillance today don't really know how well their systems are performing. They set parameters, watch alerts pile up, and spend significant time chasing false positives. BMLL and Features Analytics are betting that a data-rich benchmarking layer can change that.
A New Way to Measure Whether Surveillance Actually Works
The collaboration centers on giving financial institutions the ability to measure their own surveillance stacks, not just in absolute terms, but against changing market conditions and competing solutions. Firms would be able to run like-for-like comparisons of market abuse detection rates across different systems using the same underlying order book reconstruction, producing documentation that regulators can follow and challenge if needed.
- BMLL Unveils First Execution Analytics Dataset Built With Client Input
- New BMLL Partnership Cuts ETF Spreads by 16% in First Test
- Karen King Joins BMLL as Order Book Provider Eyes Asian Expansion
That last point matters. Regulators increasingly want to see not just that a firm flagged suspicious activity, but how it reached that conclusion. Explainability has become a compliance requirement in many jurisdictions, and rebuilding the exact state of an order book at a given moment is the cleanest way to produce that kind of evidence trail.
"Market integrity and surveillance are a natural application layer on top of high-quality historical order book data," said Paul Humphrey, CEO of BMLL. "This partnership reflects our focus on enabling sophisticated workflows on top of BMLL data, now extending beyond market quality into market integrity and surveillance benchmarking use cases."
It's a challenge that extends beyond equities. LSEG recently launched a cross-venue market abuse detection platform targeting MiFID instruments and FX, combining client trade data with public feeds and news to cut false positives.
False Positives Remain a Costly Problem for Compliance Teams
Legacy surveillance systems are known for generating enormous volumes of alerts, a large proportion of which turn out to be noise. Manually reviewing those alerts ties up compliance teams and drives up operational costs - a persistent headache for banks, brokers and exchanges.
Features Analytics has positioned eyeDES as a solution specifically designed to reduce that burden, claiming its algorithms can cut false positives by more than 90% compared to rule-based systems.
"Our mission is to help financial institutions stay ahead of regulatory requirements with our unique eyeDES AI technology and tools that deliver measurable coverage and accurate detection of market abuse,” Cristina Soviany, PhD, CEO and co-founder of Features Analytics, said.
The company traces the platform's AI methodology back to cancer detection research, which required similar precision in distinguishing genuine signals from background noise.
BMLL has been steadily broadening its partnerships in recent months, from cutting ETF spreads by 16% in a joint test with Ultumus to giving Saudi institutional investors Python-native access to historical order book data via Wamid.
BMLL's Activate Program Lowers the Entry Barrier for Partners
The partnership was structured through BMLL's Activate: Data Credits Program, which lets qualified partners build and test products on BMLL's order book infrastructure with no upfront data license costs. Features Analytics received a credit allowance covering access to BMLL's Python research sandbox and Data Feed, with a clear path to full commercial deployment once the build phase wraps up.
It's a model BMLL has been refining as it broadens the ecosystem of products built on top of its data. In early 2025, BMLL and Pico launched an end-to-end solution combining real-time and historical data to help quantitative analysts move from research to production, while a partnership with Ultumus cut ETF spreads by 16% in initial tests by combining ETF index data with granular order book analytics.
More recently, BMLL launched its Trades Plus execution analytics dataset, built with direct client input to simplify transaction cost analysis and best execution workflows.
