BMLL and SIGMA AI have announced a partnership combining BMLL’s historical Level 3, 2 and 1 order book data with SIGMA’s “Quant in the Cloud” streaming platform, producing an analytics layer that scores live market flow against historical benchmarks. The stated use cases are real-time anomaly detection, market regime classification and execution benchmarking, with anticipated users spanning exchanges, buy- and sell-side firms, platform vendors, market makers and quantitative firms. The initial focus is Saudi Arabia, the UK and Europe, with the US to follow.
The partnership runs through BMLL Activate, the company’s data credits programme, which offers lower to no upfront licence costs during a build phase in exchange for defined access to the BMLL Data Lab and Data Feed.
Building the Real-Time Market Map
Andy Simpson, Founder and CEO of SIGMA AI, describes an architecture in which SIGMA sits between raw real-time market data, BMLL’s historical record and its own analytics. The baseline is a correlation matrix run in real time across how spread behaviour, volatility, liquidity and auction dynamics move against one another, with depth of history secondary. Simpson characterises the result as a market map, against which emerging patterns are scored to generate probability curves for volatility, spread and likelihood of execution.“What we are not talking about is price prediction, because that alpha generation disappears in a heartbeat,” Simpson tells Market & Alt Data Insight. “These are more important aspects about market microstructure, and the way that market microstructure might be formed based on that historical narrative.”
Democratizing Institutional-Grade Analytics
Anomaly detection and regime classification against historical baselines are established practice on tier-one sell-side desks and among sophisticated market makers, and several vendors offer variants. Simpson recognises that building this in-house has always been possible for firms that can afford it.
“Obviously those with very deep pockets are able to do that. The big tier-one sell side, the big tier-one market makers and so on are able to do a lot of what we’re talking about,” he says. “What you can’t do, if you’re a tier two or tier three, is gain access to that intelligence.”
Simpson points to transaction cost analysis. “Even some quite big firms, who you’d think had solved it, are looking at TCA characteristics, transaction cost analysis, at T+1,” he says. “And it’s not just 24 hours; they may be waiting 30 hours before they can get the narrative of what has happened with their brokers. That’s unacceptable. They want intraday capability, and you can’t get that waiting for TCA data to come through 30 hours after the fact.”
Partnering for Data Precision and Market Reach
SIGMA describes running a formal evaluation of alternatives before selecting BMLL, on the basis of harmonisation across venues, analytics available within the same environment, and a roadmap into futures and options. SIGMA’s first clients are expected to be in equities, with futures and options supported through its relationship with Trading Technologies.
Paul Humphrey, Chief Executive Officer of BMLL, stresses the data dependency. “AI is like a magnifying glass,” he says. “You put it over poor-quality data, it’ll look awful. You put it over good-quality data, it looks great.”
He continues: “More and more customers are seeing us as a platform, a platform they can build their business upon,” he says. “The data, and the quality of the data, is an artefact of the years and years of investment we’ve put into engineering this platform.”
That conviction predates the current wave of vendor AI launches. “I’ve been saying for the last two or three years, when I’ve been asked what my AI strategy is, that while we’re watching this pan out, we wanted to build the fuel – the data that will need to be in this environment,” Humphrey says.
Distribution, MCP, and AI Accessibility
BMLL Activate is the distribution mechanism attached to that positioning, seeding downstream applications on BMLL data rather than licensing it to end users directly. SIGMA is not the first participant to surface publicly: BMLL and Features Analytics announced a comparable arrangement in February, targeting surveillance benchmarking on the same data.
“To extract anything out of our environment in the past required skills – coding. If you were unable to code, you were unable to draw those insights from our data,” Humphrey says. “As we build MCP layers into our environment, we’re giving you access to open your own Claude in our environment. So the skill becomes knowing the question you want to ask.”
That places BMLL alongside LSEG, S&P Global, FactSet and others in exposing licensed content through Model Context Protocol servers, where the connection layer has commoditised quickly and entitlement propagation has moved to the centre of the discussion. Simpson makes the client-side case. “The investment managers we talk to have a job which is to do investment management,” he says. “It is not to become a prompt engineer. It’s not to become a data scientist. It’s not to try to stay ahead of, or understand, the various themes and advances in artificial intelligence. That’s our job.”
Prioritizing Saudi Arabia and Global Expansion
Placing Saudi Arabia ahead of the US is unusual for a capital markets analytics launch. Simpson attributes it to ecosystem structure – a maturing market whose component parts sit within reach of one another, geographically and in market structure terms, as earlier financial centres built out around clustered infrastructure.
“There is definitely appetite in Chicago. There’s definitely appetite in the UK and Europe,” he says. ” But due to the powerful ecosystem in Saudi Arabia, where everyone works together, there’s an amazing desire to advance innovation & technology – and of course we will follow that money. “
Subscribe to our newsletter
