BMLL and SIGMA AI merge historical depth with real-time flow

A new capital markets intelligence layer pairs deep order book history with live AI inference, targeting Saudi Arabia, the UK and Europe first.

By Disrupts Editorial Team Mon, 3 August 2026

A white robotic arm precisely places a silver rectangular component onto a grid of similar components on a conveyor belt in a bright, automated factory setting.

BMLL, the independent provider of historical Level 3, 2 and 1 order book data, and SIGMA AI, a real-time analytics platform for financial markets, have announced a partnership that combines granular trading history with live machine learning inference. The joint product is designed to let market participants measure live execution quality against long-run historical baselines, a capability the two firms say does not exist as a standalone product category today.

The collaboration is structured around SIGMA AI's participation in BMLL's Activate Data Credits Programme, a scheme that offers selected partners subsidised or zero-upfront access to BMLL's historical datasets during the build phase, with a defined path to commercial rollout once a use case is validated. That funding architecture matters: it lowers the barrier for specialist AI vendors to build on top of institutional-grade data without committing to full licence fees before they have a working product.

Pairing the "normal" with the "now"

The combined solution layers BMLL's historical baselines for volatility, liquidity, spread behaviour and auction dynamics beneath SIGMA AI's "Quant in the Cloud" streaming platform, which applies continuous computation, ML inference and natural language delivery to live market data. The result is a system that can flag anomalies, classify market regimes, and score execution quality by comparing what is happening right now against what the historical record says should be normal.

Andy Simpson, Founder and Chief Executive of SIGMA AI, described the strategic logic: "BMLL's strength is the definitive historical record, while SIGMA AI's strength is continuous real-time computation, ML inference, and natural language delivery on live market data. The combination will create a product category that doesn't exist today and will help market participants measure execution quality against historical baselines."

Intended users span a wide swathe of capital markets infrastructure: exchanges, buy-side institutions (both algorithmic and relationship-driven), sell-side brokers, market makers and quantitative firms. The solution will launch with a focus on Saudi Arabian, UK and European instruments before expanding to the United States.

The convergence angle: fintech infrastructure meets AI operations

The story sits at a junction that Disrupts readers operating across fintech and AI infrastructure will recognise. Capital markets data has historically been siloed: historical analytics lived in research workflows, real-time analytics lived in execution systems, and the two rarely spoke to each other at the inference layer. What BMLL and SIGMA AI are building is a persistent bridge between those two worlds, effectively creating a form of financial market situational awareness that runs continuously rather than as a post-trade audit.

The Saudi Arabia focus is notable from a macro perspective. The Kingdom's capital markets have deepened considerably since the Vision 2030 push to diversify its financial infrastructure, and international data vendors and fintech platforms have been among the earliest commercial beneficiaries of that opening. Targeting Riyadh alongside London and Frankfurt in the initial rollout is a signal that the product is being positioned for institutional adoption in emerging-market-adjacent exchanges that are increasingly demanding the same execution-intelligence tooling as their Western peers.

Broader context for cross-sector investors: the infrastructure layer enabling this kind of real-time, historically contextualised AI inference is the same stack being explored in risk surveillance, regulatory compliance and, increasingly, in macro signal generation for multi-asset portfolios. As AI inference costs fall and cloud-native data platforms mature, the arbitrage between cheap compute and expensive proprietary data narrows, which is precisely the market dynamic the BMLL Activate programme is designed to exploit. Firms that lock in premium historical data relationships now, before the inference layer becomes commoditised, are acquiring a durable structural advantage in the execution intelligence space.