Trusted data fundamental to AI efficacy across the capital markets

Artificial intelligence may be transforming the global capital markets at an unprecedented pace, but one of its key principles remains unchanged: the quality of any AI model is only as good as the data underpinning it. As market participants accelerate their use of AI strategies, the focus is shifting from algorithms to the quality, integrity and accessibility of the underlying data feeding them.

According to LSEG Data & Analytics, this changing dynamic is placing pricing and reference data at the very center of firms’ AI initiatives. Rather than simply supporting downstream operational processes, high-quality, trusted data is becoming the critical piece of the puzzle that is enabling AI-driven analytics, automation and decision-making across the business.

For Adrian Murray, director of product for LSEG’s pricing and reference service, the industry’s priorities are becoming increasingly clear. “To generate trust in the output, you need to have trust in the input,” he argues. “Firms want to know how the data was sourced, whether it was derived, what they’re able to use it for and whether they can trace changes over time. They need that audit trail so they can explain why an AI system produced a particular response.”

Complete visibility

Users’ increased demand for transparency and auditability extends beyond simply knowing where data originated and how it was sourced. As AI-based tools become embedded within existing production workflows, firms are increasingly demanding more granular visibility into how data has been transformed, enriched and distributed before it reaches an application or model.

Kashif Akhtar, director of reference data at LSEG, says this growing emphasis on data lineage features regularly in conversations with existing and prospective customers. “There is a much greater demand for end-to-end history—customers want to know where the data originated, whether anyone transformed it, which systems processed it, who modified it and where it is being consumed,” he says. “That creates explainability around the data and, ultimately, trust.”

The expectations to which Akhtar refers reflect a broader industry realization that governance can no longer be treated simply as a compliance exercise. Instead, lineage, auditability and transparency have become prerequisites for deploying AI responsibly at scale.

Trusted data

The growing importance of trusted data also reinforces one of LSEG’s long-standing strengths—its track record and pedigree as one of the industry’s foremost data providers. Having spent decades collecting data from thousands of sources, normalizing it and delivering it to clients worldwide, the firm’s heritage and its ability to consistently deliver high-quality data are becoming increasingly relevant.

Akhtar recalls a recent customer example illustrating this shift. A specialist provider of withholding tax services had originally built its business by scraping publicly available data from the internet. While that approach proved sufficient during its early growth, the company eventually recognized that inconsistent data quality posed an unacceptable business risk. “They realized it was time to move to a more reliable source,” Akhtar explains. “This is our bread and butter—collecting data from multiple sources, applying a consistent data model, normalizing that data and then giving customers flexibility in how they consume it.”

Changing consumption habits

Shifting customer expectations are also reshaping how firms choose to access their content.

Traditional batch files and on-premises delivery models are increasingly being superseded by cloud-native services, allowing firms to consume data more flexibly while simultaneously reducing operational complexity and accelerating deployment and consumption.

Murray believes this evolution has followed a logical progression.

Initially, he says, cloud adoption focused on delivering greater scalability, resilience and performance behind the scenes, while the next phase simplified how firms ingested data into their own environments. Now, the emphasis is moving toward ready-to-use platforms where clients can begin extracting value almost immediately.

“The evolution has gone from providing infrastructure to reducing friction,” Murray says. “Today, we’re moving toward ready-to-query databases where there’s no ingestion or ETL [extract, transform, load] required. Looking ahead, AI will increasingly enable natural language interaction with those datasets and automate workflows that customers previously needed to build themselves.”

Heterogenous journeys

That flexibility has become increasingly important because few capital markets firms follow identical cloud transformation journeys.

Large numbers continue to operate hybrid environments, combining legacy applications and infrastructure with modern cloud platforms, while others opt to consolidate data in centralized repositories to enable AI and advanced analytics.

LSEG’s response to that trend, according to Akhtar, has been to support customers accessing pricing and reference data through whichever delivery model best suits their needs. “We’ve invested heavily in offering choice and flexibility in terms of consumption,” he says. “Every customer is on a different transformation journey. We have to continue supporting traditional delivery methods while also investing in newer services because customers increasingly expect native access to content through platforms such as Snowflake and Databricks.”

If you want to take advantage of everything AI has to offer, you need the data sitting in a single location.

Kashif Akhtar

Creating opportunities

Centralizing market data within cloud-based environments has the potential to create opportunities beyond operational efficiencies. Such strategies provide firms with a consistent foundation on which to deploy AI capabilities without having to reconcile multiple workflows, formats and distribution mechanisms.

Murray believes that a unified data environment can significantly reduce engineering complexity. Rather than maintaining different ingestion mechanisms for individual data vendors, firms can establish common workflows that simplify both data management and downstream application development. “Once organizations have data in one place, they can build tools in a consistent manner instead of developing multiple ways of distributing that data internally,” he says. “It removes a great deal of complexity.”

Akhtar argues that data centralization will become increasingly important as firms deploy AI across wider parts of the business. “If you want to take advantage of everything AI has to offer, you need the data sitting in a single location,” he says. “Once external data and your own internal data are together, it’s much easier to manage those AI capabilities and it also improves the overall total cost of ownership.”

Agentic AI to the fore

The AI-based benefits accruing to firms are even more significant as they familiarize themselves with agentic AI and deploying agents in production environments. While generative AI has captured much of the industry’s attention to date, autonomous agents able to execute complex, multistep instructions are expected to deliver substantial operational efficiencies across front, middle and back offices.

For organizations consuming large volumes of pricing and reference data, Murray sees automation as the technology’s greatest opportunity, rather than answering questions. “Our clients don’t simply ask questions about their data,” he says. “They have multiple processes to make that data usable within their own workflows. Agentic AI allows much of that to be automated, removing complexity and helping clients extract much greater value from the data.”

Akhtar cites governance as another natural application for agentic AI, which could continuously monitor policy compliance, detect unauthorized access to licensed data and strengthen operational controls around security master management and reference data governance. “Today, much of that work is still performed manually,” he explains. “AI makes those processes more efficient, less expensive over time and probably more accurate because you’re reducing the potential for human error.”

He also expects AI to play an increasingly important role in monitoring the quality of data itself, where, for example, it can automatically compare content from multiple vendors, identify anomalies and validate incoming data before it enters live environments. “For all of this to work, you’re going to need more and more trusted data to power those tools,” he adds.

Increased sophistication

It all comes back to auditability, traceability and transparency.

Adrian Murray

If anything, the increasing sophistication of AI reinforces, rather than diminishes, the importance of pricing and reference data. As models begin influencing investment decisions, risk calculations and operational processes in close to real time, firms require broader instrument coverage, higher-quality content and stronger governance. “The requirement is fundamentally about greater breadth, depth and quality of data,” Akhtar says. “Customers want a strategic partner capable of providing cross-asset coverage while continuing to invest wherever additional content is needed.”

For Murray, transparency remains the common thread connecting every stage of AI adoption, providing consumers with confidence in the data they are using. Whether firms are analyzing price movements, generating analytics or deploying autonomous agents, they must always be able to explain how data has influenced the outcome of their analysis. “As clients consume pricing data, they want to understand what changes in the underlying reference data drove those price movements and how changes to the underlying terms influenced the analytics,” he says. “It all comes back to auditability, traceability and transparency.”

Data quality determines success

What is clear is that AI is already fundamentally reshaping how capital markets firms consume and act on information, although its success will depend less on the intelligence of the models themselves and more on the quality of the data feeding those models. And, as firms continue modernizing their infrastructures and embedding AI into critical workflows, high-quality pricing and reference data will continue its evolution from an operational necessity to a competitive advantage, one that will underpin the industry’s next phase of innovation.

Sponsored content