About a-team Marketing Services

A-Team Insight Blogs

Survey Highlights Challenges in Investment Research Data Amid Rising Demand for Systematic Strategies

Subscribe to our newsletter

The growing adoption of quantitative and AI/Machine Learning (ML) techniques, alongside the rise of systematic investment strategies, has elevated the importance of investment research data, according to a recent Bloomberg survey of over 150 quants, research analysts, and data scientists.

The survey, conducted during a global series of client workshops, identified data coverage, timeliness, and quality issues with historical data as the top industry challenge. Other challenges include normalising and wrangling data from multiple providers and determining which datasets to evaluate and research.

These hurdles are compounded by constraints in dataset evaluation, with two-thirds of respondents reporting that evaluating a single dataset takes at least a month. This underscores the difficulties research teams face in managing the increasing volume of data required to identify alpha-generating opportunities.

The survey also explored how firms manage research data. Half of the respondents reported using proprietary centralised solutions, compared to just 8% who outsource to third-party providers. Additionally, 62% prefer their research data to be hosted in the cloud, reflecting a growing trend towards scalable and accessible storage solutions. Meanwhile, over a third of respondents expressed a preference for traditional data delivery methods such as REST APIs, on-premises systems, and SFTP, highlighting a demand for flexibility in accessing research data.

In response to these challenges, Bloomberg has introduced new additions to its Investment Research Data suite to enhance data accessibility and usability:

Industry Specific Company KPIs and Estimates: This product offers point-in-time data on over 1,200 key performance indicators (KPIs) for a wide universe of companies, enabling detailed sector and industry research. The data is also available via Per Security, allowing for customised data universes.

Equity Pricing Point-in-Time: This dataset provides daily end-of-day composite pricing with security master data for the global public company universe, offering accurate historical pricing data.

These datasets can be integrated with Bloomberg’s existing Company Research Data products to build comprehensive company and industry knowledge graphs, facilitating more effective alpha discovery.

TradingTech Insight spoke with Angana Jacob, Global Head of Research Data, Bloomberg Enterprise Data, to discover more about how Bloomberg is addressing the challenges uncovered in the survey.

TTI: Angana, your survey highlights that data coverage, timeliness, and quality remain key challenges for quants and research analysts. How is Bloomberg addressing these specific challenges to ensure that your datasets are truly ‘AI-ready,’ and how do you see the evolution of data quality impacting alpha generation over the next few years?

AJ: Bloomberg’s high-quality Investment Research Data solutions provide comprehensive coverage across asset classes and are specifically tailored for quants, analysts and AI/ML workflows. We focus on the following key features: long point-in-time history covering several economic cycles and regimes, accurate timestamps, granular metadata and making sure the data is pre-processed and curated for analysis and backtesting.

These features are crucial in investment research as otherwise ML models can capture noise rather than genuine signals or detect false trading opportunities. Overfitting is a dangerous pitfall that quants can fall into, where their strategies perform well on historical data but fail in real-world trading. Backtesting on data that has gaps in history, inaccurate timestamps or has missing metadata can lead to detecting spurious patterns that the strategy overfits to, inflating expected returns which do not materialise. We believe working with high-quality point-in-time data, free from bias, can help gain an investment edge and we see quants increasingly focus on getting their data foundations right.

As client demand for large volumes of data and increasingly sophisticated analytics grows, they need data that is accurate, complete, timely, and aligned – we expect the focus on data quality to only rise going forward, especially with increasing industry adoption of AI. Data has to be free from inconsistencies and errors and organized in a way that AI models can interpret – that is being “AI-ready”. The data also has to be aligned with the specific task the AI model is designed to address – point-in-time historical data is essential for predictive AI models or simulations, metadata tags and annotation labels for unstructured text analysis. AI workflows rely on vast and diverse datasets, and there is increasing focus on having a unified data model to ensure that these datasets are structured and interpretable.

Additionally, data quality advancements can also help investors find new alpha sources by being able to incorporate non-traditional data such as alternative data (eg satellite imagery) or broad macroeconomic data (very noisy) that was previously considered difficult to find a signal from.

TTI: The survey found that more than half of respondents take a month or longer to evaluate a single dataset. What innovations are Bloomberg introducing to reduce this ‘time to alpha,’ and how do your Python APIs and research data products improve the data ingestion, normalization, and evaluation workflow for quants and data scientists?

AJ: Bloomberg focuses on three major areas to improve investment firms’ decision-making and their time to alpha:

Breadth and Quality of Data: Bloomberg provides a wide array of diverse datasets, and continually adds more so that clients can have the full 360 degree perspective of the asset class they are trading. In terms of quality, as mentioned above, Bloomberg’s Research Data group focus on the following key features for quants: long point-in-time history covering several economic cycles and regimes, accurate timestamps, granular metadata and making sure the data is pre-processed and curated for analysis and backtesting.

Interconnected Data: Furthermore, to remove the burden of ingesting and transforming data, Bloomberg’s Research datasets are designed to be fully mapped and interoperable. Having good data consistency and standardisation with a semantic layer, accelerates quant’s research workflow reducing operational complexity.

The technology: In addition to high-quality interconnected data with deep history, Bloomberg also offers the data integration tools to ensure it is ready to use and flexible, with customizable data delivery with minimal ingestion and ETL processes. Using Bloomberg’s Virtual Data Room (VDR) environment as one example, we also make using the cloud for hypothesis testing and experimentation extremely easy. Hosted in Python-based Jupyter Notebooks, Bloomberg’s VDR makes it faster than ever to investigate the coverage, quality and usability of Bloomberg’s Research datasets.

Our Python Unified Data Model API allows clients direct access to unified data from Bloomberg’s managed service offering, Data License Plus (DL+). It is optimized for quant research, with features such as data discoverability and interoperability and fast point-in-time timeseries retrieval.

We are probably the only provider that offers such comprehensive and high-quality data with a suite of functionality on top to empower quants in their data use. We believe our offering is very compelling, no longer do quants have to suffer from the much joked-about 80-20 rule, where most of their time is spent on data wrangling instead of actual alpha generation.

TTI: Your findings show a clear preference for cloud-based research data management, but a significant portion of respondents still want traditional access methods like REST API, on-premise, and SFTP. How is Bloomberg balancing these preferences to offer the flexibility that clients need, and do you see cloud adoption accelerating across the financial services industry?

AJ: Bloomberg has built and optimized all of our Enterprise Data services to be available and accessible for customers in the cloud, while also giving them the option of using their full legacy approaches, as well. We have always embraced a unique managed service model with a customer-first approach, allowing us to anticipate customer needs before they identify them and meet customers in the environments of their choice with our lightweight, highly flexible data and technology solutions.

Cloud adoption remains uneven across the industry, with many firms still at the beginning of their journeys. However, for the quant and research use case specifically, we’re seeing increased adoption as a cloud environment is critical to handle increasingly vast volumes of data for computationally-intensive investment processes such as backtesting, signal generation, optimizing portfolios, scenario analysis, hypothesis testing and so on. Cloud is vital for providing the scale, elasticity and cost efficiency needed for quant and research use cases. Additionally, as we see clients’ integrate more sophisticated techniques into their investment process and rapidly prototype new strategies with increasing volumes of data, cloud capabilities are necessary to support.

TTI: Bloomberg’s new Company Research Data products, including KPIs and Equity Pricing Point-in-Time, are designed to help clients build company and industry knowledge graphs. Can you share Bloomberg’s long-term vision for these knowledge graphs? How do you see them transforming investment research and unlocking new alpha-generating opportunities?

AJ: Bloomberg’s Company Research Data products allow clients to build company and industry knowledge graphs, essential to discover trading opportunities that wouldn’t be visible in siloed data. A broad range of interconnected datasets helps understand context and nuance of investment factors and reveal new cross-domain insights. For equity long-short strategies, quants need very granular data on a company in order to really understand the universe of companies they can trade in (ex. Company’s operating segments, revenue lines, geographic exposures, supply chain etc.) Additionally, they need timely information to predict earnings (ex. Credit cards and footfall information etc.). With all of this data together with a semantic layer to connect them (the knowledge graph), they get the full 360-degree perspective of a company which is essential to predict how the stock will perform.

Alongside the new Industry Specific Company KPIs and Estimates and Equity Pricing Point-in-Time products, our current product suite includes Company Financials, Estimates, Pricing and Point in Time Data, Operating Segment Fundamentals Data, Industry Specific Company KPIs and Estimates Data, Supply Chain Data and Alternative Data (Bloomberg Second Measure) products, covering a broad universe of companies and providing deep actionable insights.

Additional solutions such as Geographic Segment Fundamentals Data, Company Segments and Deep Estimates Data and Pharma Products & Brands Data products will be available in 2025. As we continue to add new datasets including in the macro space, our long-term vision is to help investors unlock insights and alpha faster by providing deep and granular insight into all factors that could drive markets.

TTI: Thank you, Angana.

Subscribe to our newsletter

Related content

WEBINAR

Recorded Webinar: Smart Trader Desktops: Placing UX at the front and centre of the trading workflow

Trading strategy is in place, the technology stack is optimised and the trading team is highly skilled – but what about the user experience? Whatever the stack, the desktop, the trading apps and their functionality, a trading platform is only as good as its user interface (UI) and user experience (UX). This webinar will review...

BLOG

Pico and BMLL Partner to Deliver Integrated Real-Time and Historical Market Data Solution

Pico, the global provider of financial markets technology services, has partnered with BMLL, the independent provider of historical data and analytics, to offer an integrated solution that combines real-time and historical market data to support research, liquidity analysis, and trading optimisation. The collaboration leverages Pico’s real-time data feeds and global infrastructure alongside BMLL’s Level 3,...

EVENT

Data Management Summit London

Now in its 15th year, the Data Management Summit (DMS) in London brings together the European capital markets enterprise data management community, to explore how data strategy is evolving to drive business outcomes and speed to market in changing times.

GUIDE

Institutional Digital Assets Handbook 2024

Despite the setback of the FTX collapse, institutional interest in digital assets has grown markedly in the past 12 months, with firms of all sizes now acknowledging participation in some form. While as recently as a year ago, institutional trading firms were taking a cautious stance toward their use, the acceptance of tokenisation, stablecoins, and...