Revolutionising financial data with large language models

Applications, advantages and future directions With the integration of large language models (LLMs), LSEG is reshaping data analytics, utilisation and management. Powered by LSEG’s extensive data, these LLMs offer transformative applications for investment professionals. Leveraging market tick data from 1996, and earlier news, transcripts and filings, LSEG ensures its LLMs are accurate and precise. This presentation explores the applications, advantages and future directions of LLMs in trading.

Find out more about fixed income trading trends in 2024

What shook the bond market last year?

“LLMs are only as strong as the data they’ve been trained on. LSEG’s extensive historical data ensures robust, accurate and precise models” Tim Anderson Director, Quantitative and Economic Data, Head of Business Solutions, LSEG

AI, data and cloud: the technology opportunities

Discover your new home for fixed income

What does this mean for fixed income in 2024?

The authors

Delivering strategic value

Unmesh Bhide Director, LSEG

Jayme Fagas Global Head of Pricing and Valuations, LSEG

Jayme Fagas has been the global head of pricing and valuations at LSEG since 2014 and is a part of the broader Pricing and Reference Services business. Prior to this role, she was global head of evaluated pricing operations at Refinitiv, then Thomson Reuters, in the enterprise content sector. Jayme brings significant expertise to LSEG having worked as a sell-side fixed income trader for more than 20 years at several major Wall Street firms. She also has years of experience in fixed income analytics and evaluations, and has been a pivotal driver in transforming the evaluated pricing market. Jayme has more than 30 years of experience within the global financial industry. As a member of the LSEG Pricing and Reference Service Leadership Team, she focuses on managing LSEG’s evaluated pricing proposition, which is delivered via DataScope Select, the strategic data delivery platform for non-streaming content at LSEG. Jayme is an active industry participant working closely with mutual funds, hedge funds, asset managers, fund administrators and custodians to provide solutions that meet the industry's regulatory and compliance needs.

Andres Gomez Head of Financial Analytics, LSEG

Andres Gomez has been with Citi for more than 30 years and first joined Salomon Brothers – which, through several mergers became Citigroup – as a member of the international operations department in 1987. In 1996, he joined Yield Book and worked in the analytics support team and was promoted to head of analytics support and consulting in 1999 and, in 2011, was named global head of client services. In 2018, Andres became part of FTSE Russell as part of the Yield Book purchase and was named head of relationship management for the information service division in 2020. He has a Bachelor's degree in business administration (finance) from Hofstra University and an MBA from the Stern Business School at New York University.

In challenging markets, confidence is vital. LSEG’s robust data and analytics provide the insights investors need. LSEG is a leading global financial markets infrastructure and data provider that brings together powerful data and analytics as a trusted source for in-depth fixed income analytics. This creates new opportunities for fixed income portfolio managers and investors. For example, they can now access the data and analytics they need in a ‘one-stop shop’. And they can engage with the same datasets through the front, middle and back offices – and across different use cases. It also means that asset managers and investors can operate with greater confidence in challenging markets, knowing they are working with high-quality, transparent fixed income pricing and reference data, indexes and analytics throughout the enterprise. Download the Fixed income at a turning point: strategies and outlook for 2024 report

Unmesh Bhide joined LSEG as director of securitised products in 2023. Prior to this, he was managing director and founder of PricingDirect at JP Morgan. He was the driving force behind the implementation of transformational initiatives introducing machine learning and other advanced technology capabilities to expand on pricing and liquidity offerings. Prior to joining JP Morgan/Bear Stearns, Unmesh worked at the Merrill Lynch Pricing Service, where he focused on developing pricing models to drive valuations. He earned a BSc in physics from the University of Bombay, an MS in physics from the Indian Institute of Technology, Bombay, and an MS in computer science from the New Jersey Institute of Technology. Unmesh is a member of the Chartered Financial Analyst (CFA) Institute and a CFA charter holder.

Disclaimer

Investors leverage LSEG's fixed income indices for benchmarking, research, and portfolio management. With a long history of providing reliable measurements across developed and emerging markets, LSEG’s fixed income indexes support precision investment strategies and research functions. Curated benchmarks with unique risk characteristics enable more refined investment approaches and analytical insights. Read more about fixed income indexes

Trusted indexes

Investors can unlock deeper insights and construct profitable investment strategies with LSEG’s Yield Book advanced analytics. The single-source solution, sophisticated modelling and unrivalled flexibility allow investors to refine analytics sets according to their needs. From regulatory alignment to tailored risk management, Yield Book enables confident decision-making, ensuring investments thrive in today's dynamic market landscape. Read more about yield book analytics

Sophisticated analytics

High-quality data

Investors can elevate their fixed income strategy with LSEG’s comprehensive pricing and reference data solutions. With daily evaluated pricing on more than 2.8 million fixed income instruments, and reference data covering upwards of 80 million instruments across all asset classes, LSEG’s valuation services empower investors to make informed decisions and mitigate risks effectively. Read more about LSEG's comprehensive pricing and reference data

High-quality data

Discover your new home for fixed income

Cloud-based solutions

The transition to a cloud-based infrastructure enables improved agility and regulatory compliance. Enhance operational efficiency and optimise fixed income processes for evolving market landscapes.

Cloud-based solutions

Enhanced data analytics

Comprehensive data analytics tools unlock deeper insights and refine investment strategies. Access diverse datasets and real-time analytics capabilities, enabling proactive decision-making in dynamic market environments.

Enhanced data analytics

AI integration

AI integration empowers investors to navigate market unpredictability and optimise portfolio performance. Seamlessly integrate AI algorithms for risk management and opportunity identification, enhancing operational agility.

AI integration

AI, data and cloud: the technology opportunities

How should investors navigate geopolitical uncertainties and potential triggers for market volatility in 2024?

While the outlook for fixed income is broadly positive, there are plenty of opportunities for volatility to return. The challenges faced last year highlighted the critical need for precise and transparent data in fixed income trading, especially in navigating geopolitical uncertainties. The adoption of data from trusted providers ensures confidence in decision-making processes and supports effective portfolio management and risk mitigation strategies.

How should investors navigate geopolitical uncertainties and potential triggers for market volatility in 2024?

Where can investors find yield opportunities amid shifting interest rates and credit spread dynamics?

The hunt for additional fixed income yield resulted in a rally of credit spreads during the last two months of 2023. Asset managers and investors seek transparency in fixed income pricing data and index methodologies to understand underlying market movements. However, investors will continue to seek yield opportunities but may face constraints in finding attractive returns. This emphasises the importance of accurate real-time data for traders to hedge positions effectively.

Where can investors find yield opportunities amid shifting interest rates and credit spread dynamics?

How will central banks' rate cuts and economic resilience affect fixed income investments in 2024?

Rate cuts are expected in 2024, reflecting a softer stance by central banks amid economic uncertainties. The global economy appears set for a soft landing, bolstering the outlook for fixed income investments. However, the challenges faced last year emphasised the critical need for precise and transparent data in fixed income trading. Investors and asset managers require accurate and trustworthy data, especially during times of high volatility, to make informed decisions. As investors move out of cash and into long-dated securities, there is a growing reliance on accurate data to support trading strategies across various asset classes.

How will central banks' rate cuts and economic resilience affect fixed income investments in 2024?

What does this mean for fixed income in 2024?

Corporate credit performance and tightening spreads

Corporate credit performed well as confidence in corporate earnings and the economic outlook strengthened. Credit spreads tightened, particularly in high-yield sectors, and investors were drawn to potential returns and yield advantages over government bonds. This tightening of credit spreads signalled increased appetite for riskier assets within the fixed income market.

Corporate credit performance and tightening spreads

Shift in investor sentiment and inflows into fixed income

Investor sentiment shifted as confidence in inflation control grew. Economic indicators suggested a soft landing rather than a recession, boosting faith in fixed income assets. Investors shifted away from cash and poured into longer-term government bonds to secure higher yields. Positive flows were seen throughout the past year, with mutual bond funds and fixed income exchange-traded funds (ETFs) witnessing significant inflows totalling $375 billion, according to Lipper.

Shift in investor sentiment and inflows into fixed income

Bond rally towards the end of the year

In 2023, the bond market defied concerns about inflation and geopolitical tensions, with a remarkable turnaround. Towards the year's end, a significant rally swept through the bond market, driven by optimism about impending rate cuts. Investors expected central banks to ease monetary policies in 2024, causing bond prices to soar. November and December were pivotal, recording the strongest two-month performance in more than three decades.

Bond rally towards the end of the year

Here’s what shook the bond market last year

The flows into bond ETFs grew by 89% – from $142.1 million in the first six months to $268.7 million for the whole year. This indicates a nearly doubling of investments in bond ETFs, highlighting a significant rise in investor interest in these instruments during the latter half of 2023.

Growth in bond ETFs

The flows into money-market mutual funds increased by $488.1 million, from $731.4 million in the first six months, to $1,219.5 million for the entire year. This substantial rise reflects a growing preference among investors for the relative safety and liquidity offered by money-market mutual funds.

Increase in money market mutual funds

The combined flows into all fixed income funds, including bond and money-market funds (ETFs and mutual funds), reached $1,606 million for the entire year. This total demonstrates the massive scale of investment in fixed income products in 2023, showing a strong overall shift towards these asset classes.

Overall combined flows

Key takeaways

Speed and accuracy LLM-powered summarisation offers rapid insights into market trends and trading activities, enhancing decision-making with real-time information.

Data quality Ensuring robust data quality forms the foundation for reliable LLM applications, enabling precise analysis and informed investment strategies.

Data efficiency Leveraging cloud-based data sharing and integration optimises efficiency by eliminating the need for manual data downloads, streamlining processes for faster insights.

Customisable LLMs Tailoring LLMs to specific use cases empowers users to address unique challenges and extract actionable insights tailored to their trading and investment needs.

Operational efficiency Artificial inteligence-driven solutions enhance operational efficiency by automating tasks such as issue resolution, freeing up valuable time for strategic decision-making and analysis.

Have your say

Submit

Which LLM-powered applications has/would have the most transformative impact on your trading and investment strategies?

Real-time trading summarisations and market insights Detection and correction of gaps or errors in historical data Seamless cloud-based data sharing and LLM integration Translation of natural language trading strategies into executable algorithms AI-driven customer support and issue resolution

Thank you!

About LSEG Data & Analytics LSEG Data & Analytics is one of the world’s largest providers of financial markets data and infrastructure. With more than 40,000 customers and 400,000 end-users across approximately 190 markets, we are an essential partner to the global financial community and redefining the future of data in financial services. We enable customers to draw crucial insights through data, feeds, analytics, AI and workflow solutions. With our unique insights seamlessly integrated into customers’ workflows, we can help them to identify opportunites and seize competitive advantage. Contact us to find out more about LSEG Quantitative Data Solutions

LLM deep dive Explore how LSEG is using LLMs to revolutionise decision-making processes in the realm of financial services.

Customisable AI applications: tailoring to needs

LLMs offer flexibility, allowing customers to train models on proprietary data for specific requirements. This customisation ensures artificial inteligence tools are precise, compliant and effective in financial applications.

Custom training: A hedge fund uses its proprietary LLM trained on LSEG’s historical market data to develop distinctive trading strategies.

Natural language processing: a business logic expert describes trading strategies in natural language, which the LLM translates into executable code, reducing the time needed to create functioning algorithms.

Historical market data

Customer LLM

Strategy development

If these stocks move within a specified percentile range over this time period – execute trades. If the volume drops to a specified level and a machine-readable news signal reaches a certain threshold – stop trading. Additionally, backtest this pattern over the past 10 years using data from the London Stock Exchange, focusing on all petroleum companies:

Automation and AI agents: enhancing efficiency

The road ahead: what’s next for LLMs in financial services?

Click anywhere to exit

AI agents powered by LLMs can automate tasks from customer support to complex financial analyses, freeing up human resources for more strategic functions.

Customer support: An artificial inteligence (AI) agent handles routine enquiries, subscription management and troubleshooting by interfacing with customer AI agents.

Customer enquiry

Initial interaction

Verification

Resolution

Feedback loop

Customer enquiry A client’s AI agent contacts the company’s AI agent to address a specific issue, such as missing access to a subscribed news package.

Initial interaction The company’s AI agent identifies the enquiry and begins the troubleshooting process by accessing relevant subscription and order records.

Verification The AI agent cross-checks the customer’s subscription details and order forms to verify the client's entitlements and current access.

Resolution On identifying the issue, the AI agent takes necessary actions, such as updating the subscription status, provisioning access and notifying the customer’s AI agent of the resolution.

Feedback loop The interaction is logged for future reference, enhancing the knowledge base and improving the AI agent’s efficiency in handling similar issues.

Market analysis: an AI agent performs real-time market assessments, summarising data from various sources into actionable insights.

Step 2: AI analyses the data to identify current trends.

Step 4: AI analyses the competitive landscape by identifying major players.

Step 5: AI checks public and private company statuses.

Step 6: AI reviews officers, directors and board members.

Step 7: AI examines filings, transcripts and news items.

AI output: Report with actionable insights to make informed decisions about investment strategies related to the EV charging infrastructure market across North America and Europe.

Step 1: AI gathers data from market research reports, government websites, industry publications and news articles.

Query: "What are the most promising markets for electric-vehicle [EV] charging-station investment in North America and Europe, considering government incentives, consumer adoption and competition?"

The future of LLMs in financial services is promising, with advancements expected in several areas:

The integration of LLMs into financial services is ushering in a new era of efficiency, accuracy and innovation. Leveraging high-quality structured data and advanced artificial inteligence (AI) capabilities, financial institutions can navigate market complexities with unprecedented agility and confidence.

LLMs offer increased operational efficiency and faster response times, allowing investment professionals to focus on high-value tasks.

Translation of business logic

Specialised mini LLMs

Expanded automation

Translation of business logic LLMs will further simplify the translation of business logic into executable trading algorithms, bridging the gap between strategic thinking and technical implementation.

Specialised mini LLMs Development of smaller, more focused LLMs tailored to specific financial tasks will enhance performance and accuracy.

Expanded automation More processes will become automated, from generating reports to executing trades, as AI agents become more sophisticated.

Step 3: AI identifies investment opportunities.

Have your say

Submit

Which LLM-powered applications has/would have the most transformative impact on your trading and investment strategies?

Thank you!

Submit

Data summarisation and analysis: cutting through complexity Analysts need to understand full days’ trading activities across multiple exchanges. LLMs revolutionise this task.

LLMs simplify data analysis by cutting through complexity.

After LLMs

Before LLMs Analysts manually extract and analyse data points, such as trading volumes, price changes and order book details, across multiple exchanges. This process could take hours or even days, making it time-consuming and prone to errors.

After LLMs In just a few seconds, an LLM generates a summarised report of the entire trading day, highlighting key metrics and anomalies. This provides swift, accurate summaries of complex data, enabling analysts to make informed decisions faster and more efficiently.

Before LLMs

Analysts need to understand full days’ trading activities across multiple exchanges. LLMs revolutionise this task. LLMs simplify data analysis by cutting through complexity.

After LLMs

Before LLMs

Enhanced data quality and accuracy: ensuring reliability

Integration and efficiency: streamlining data sharing

Data summarisation and analysis: cutting through complexity

Analysts need to understand full days’ trading activities across multiple exchanges. LLMs revolutionise data analysis.

Before LLMs Analysts manually extract and analyse trading volumes, price changes and order-book details, a time-consuming and error-prone process.

After LLMsLLMs generate trading-day summaries, highlighting key metrics and anomalies for quicker, more informed decisions.

Traders and analysts need accurate, reliable data to ensure robust trading insights for algorithmic trading. LLMs enhance data integrity by identifying and fixing gaps in historical data.

Before LLMs A one-minute anomaly in historical data could represent a missed opportunity for traders and investors.

After LLMsThe model identifies anomalies, outliers and events at nanosecond intervals, rectifies errors and provides explanations.

Cloud integration – utilising LLMs alongside cloud storage processing technology – transforms data management by eliminating costly and environmentally taxing processes.

Traditional method Each client downloads massive datasets and builds infrastructure to store and manage their own copies of the data.

With cloud integrationLSEG shares a single, centrally shared data copy via the cloud. Clients use artificial intelligence and LLM tools for analysis.

⁠In search of clean data: firms navigate data challenges as LLM adoption flourishes

Leading-edge LLM approaches: predictive analytics for merger and acquisitions targets

Large language models in finance

Demystifying the use of LLMs in finance