London Stock Exchange Group (LSEG) has partnered with NVIDIA to accelerate the analysis of historical financial market data, with the companies reporting that workloads which previously took hours can now run in seconds using GPU-accelerated processing.
The collaboration focuses on LSEG Tick History, a financial data archive containing approximately 110 trillion rows covering around 100 million instruments and more than 30 years of market activity. The system provides historical data used by financial institutions for quantitative research, trading strategy testing and market analysis.
The companies are using NVIDIA’s cuDF technology alongside Polars, a framework for processing structured data. NVIDIA cuDF uses graphics processing units (GPUs) to accelerate data operations such as filtering, sorting and aggregation, which can become computationally expensive when applied to large financial datasets.
The arrangement is designed to let users analyse the historical archive without creating another complete copy of the underlying dataset. LSEG also intends to retain existing data licensing, permissions and access controls, allowing institutions to work with the information within their authorised access arrangements.
Faster processing could help quantitative researchers test trading strategies against longer periods of market activity, examine liquidity across trading venues and investigate historical market conditions. Financial institutions can also use historical market data for execution analysis and risk modelling.
The announcement follows growing demand for LSEG’s market data services. The company reported that requests to its Tick History API rose 48% in the first half of 2026 compared with the same period a year earlier. Its real-time network also reached a peak of 26 million messages per second in June 2026, according to figures included in the announcement.
The performance claim has not been accompanied by a complete public benchmark methodology. The announcement did not specify the exact workload tested, the hardware configuration or the CPU-based system used for comparison, limiting independent assessment of the reported reduction in processing time.
Discover more from Aree Blog
Subscribe now to keep reading and get access to the full archive.
