LSEG has announced a collaboration with NVIDIA that enables financial institutions to analyse years of historical tick data in seconds rather than hours, using GPU-accelerated processing technology. The initiative is designed to support increasingly data-intensive quantitative research, trading analytics and AI applications, while maintaining the licensing and access controls governing institutional market data.
The collaboration combines LSEG’s Tick History data with NVIDIA cuDF, a GPU-accelerated data processing library, allowing financial institutions to process large historical datasets without creating additional full copies of the underlying data.
Benchmark testing using NVIDIA cuDF and Polars, a data processing framework, demonstrated that analysis previously requiring hours could be completed in seconds. The announcement forms part of LSEG’s Everywhere AI strategy, which aims to make its data more readily accessible within the computing environments used by financial institutions.
A central feature of the approach is the ability to bring computing capabilities closer to the data while preserving existing permissions, licensing arrangements and access controls. This could help address some of the operational and governance challenges associated with moving large volumes of historical market data between analytical environments.Scaling Historical Data Analysis
LSEG’s Tick History service contains 110 trillion rows of data covering approximately 100 million instruments, with more than 30 years of history drawn from over 580 exchanges, trading venues and contributors.
Processing datasets of this magnitude presents considerable challenges for financial institutions, particularly when quantitative researchers need to examine extended periods of market activity across multiple instruments and trading venues.
Traditional approaches can require substantial computing resources and processing time, encouraging firms to work with smaller samples or aggregated datasets. While these techniques can make analysis more manageable, they may also remove information relevant to understanding market microstructure, liquidity and execution behaviour.
GPU acceleration offers a way to process larger datasets without necessarily sacrificing that level of detail.
“Customers want to work with detailed historical data at speed, without the burden and risk of creating another full copy,” comments Ron Lefferts, Co-Head of Data & Analytics at LSEG. “This helps firms analyse deeper histories closer to the point of decision, while keeping the permissions and controls they rely on fully intact.”
The announcement comes amid growing demand for market data processing capacity. According to LSEG, traffic across its real-time network increased by 29% in the first half of 2026 compared with the equivalent period in 2025, reaching a peak of 26 million messages per second in June. Customer API requests for Tick History data increased by 48% over the same period.
Supporting Quantitative Research and AI
The collaboration is intended to support several analytical workflows, including quantitative strategy development, execution analysis, market replay and AI-enabled research.
For quantitative investment firms, faster processing could allow researchers to test strategies against longer historical periods, incorporate additional trading venues and examine a broader range of market conditions without reducing the granularity of the underlying data.
Execution analysts could use the technology to reconstruct historical order books and assess pricing, liquidity and execution quality across different market environments.
The ability to accelerate historical market replay and scenario analysis could also support risk management applications, particularly where firms need to examine how strategies or portfolios might behave under conditions resembling previous periods of market stress.
AI represents another potential application. Access to large volumes of granular historical data can support the development and evaluation of analytical models, although the computational demands involved can make comprehensive analysis expensive and time-consuming.
“AI is opening new ways to understand market behaviour across vast amounts of data,” says Ioana Boier of NVIDIA. “NVIDIA cuDF helps researchers analyse LSEG Tick History at scale without downsampling away the detail that makes the data valuable for advanced analytics and AI.”
The collaboration provides a processing foundation on which customers can apply their own analytical and AI models, rather than introducing a specific AI application or proprietary investment model.
Preparing for Longer Trading Hours
LSEG also positions the initiative against the backdrop of extended trading hours, particularly in US equities, where exchanges and market participants are preparing for increasingly continuous trading.
Historically, many institutional analytical workflows have operated around defined market sessions, with substantial data processing and reporting taking place after the market close.
As trading hours extend towards 24-hour operation, firms may have fewer opportunities to rely on overnight processing windows. Quantitative research, execution analysis and risk monitoring could increasingly need to operate alongside live trading activity.
Accelerating historical data processing could help firms incorporate more extensive historical analysis into intraday decision-making, rather than relying predominantly on calculations completed outside trading hours.
The ability to maintain existing data permissions and controls is also relevant as financial institutions expand their use of AI and distributed computing environments. Market data licensing arrangements can place restrictions on how information is accessed, processed and redistributed, making governance an important consideration when introducing new analytical infrastructure.
LSEG has not disclosed the detailed benchmark methodology, hardware configuration or specific datasets used to demonstrate the reported performance improvements. Actual processing gains will therefore depend on the workloads and computing environments involved.
Nevertheless, the collaboration illustrates how GPU acceleration is being applied to established institutional market data services, potentially allowing financial institutions to extract greater analytical value from datasets they already license.
As demand grows for more granular research, faster execution analytics and AI-ready data infrastructure, the ability to process extensive historical datasets without duplicating them could become an increasingly important consideration in institutional market data architecture.
Subscribe to our newsletter
