Future-proofing the investment operating model: Unifying data, analytics and execution in the zettabyte era

Bloomberg Professional Services

Buy-side firms are consistently faced with burgeoning volumes of data, necessitating adept management of expansive data extractions, a task fraught with intricacies and considerable costs. Arthur Orts, Bloomberg’s Head of Portfolio Analytics and Risk Sales, and Matt York, Buy-Side Product Strategy, articulate the challenges firms face while providing discerning solutions to navigate these challenges.

How has the growth of data impacted buy-side managers?

The unprecedented explosion in the volume and complexity of data, presents both opportunities and challenges for investment managers. To put this into perspective, the global datasphere is expected to grow from 64.2 zettabytes in 2020 to 175 zettabytes by 2025, with approximately 80% of this data being unstructured.

Leveraging data and analytics at scale can be a difficult undertaking, with only the largest asset managers having the resources to maintain dedicated teams capable of extracting meaningful insights. For example, Bloomberg’s MAC3 Risk model alone incorporates over 3,800 factors spanning over hundreds of thousands of companies, each relying on the daily ingestion of billions of data points. Another example is Bloomberg’s Enterprise Data business, which provides scalable enterprise-wide access to over 100 billion data points published daily via Data License consistent with the data on the Bloomberg Terminal.

However, the emergence of large language models (LLMs) and data vectorization techniques is transforming the landscape, enabling the conversion of unstructured data into structured formats that can be readily incorporated into quantitative investment strategies.

To stay competitive, buy-side firms must harness the power of this data to uncover alpha and optimize their portfolios. However, the process of transforming raw data into actionable intelligence is not without its hurdles. The data pipeline – from acquisition, to analysis, to decision-making – needs to be streamlined and efficient to keep pace with the rapidly evolving market landscape.

For most investment managers, navigating this landscape requires partnering with technology providers offering scalable, cost-effective solutions for data management and analytics. For example, Bloomberg’s Data License Plus (DL+) solution acquires, aggregates, organizes and links a customer’s Bloomberg Data License data from multiple delivery channels and multi-vendor and custom ESG data into a single Unified Data Model. By leveraging expertise and infrastructure from these providers, buy-side firms can manage the influx of data made accessible to them and efficiently derive value from their data spend to make informed decisions.

What’s an ideal setup for managers and what are the challenges to getting there?

First, a coherent operating model that unifies data management, analytics and investment processes – supported by strategic partnerships – enables investment managers to effectively navigate the data-driven era and achieve better outcomes. This non-fragmented model integrates technology, expertise and processes into a unified ecosystem, allowing managers to efficiently harness data for informed decision making.

Partnering with a single provider to unify the investment lifecycle components minimizes fragmentation, ensures consistency and enables managers to focus on core competencies while benefiting from the latest advancements in data management and analytics.

Asset managers are often challenged when they try to build in-house solutions; underestimating the resources and expertise required. Legacy systems and processes can hinder the adoption of new technologies and operating models, making it difficult to redraw it from scratch. User adoption and skills gaps also need to be factored in for a successful implementation.

Learn more about solutions for the Buy-Side here.

What does it take for firms to reach these scenarios?

Don’t break your data model. It’s as simple as that. The moment we hear “flat files”, “csv”, or “Excel”, we know that there is something that needs fixing. Those formats inherently reduce your data’s high dimensionality to a mere two dimensions, which is the worst thing that can happen. Think about how rich a data structure can be: hierarchical, relational, dimensional, metadata, semantic, temporal – its dimensions are 100 times more complex and insightful than just a column and a row. More often than not, the real insights lie within those complex structures.

To give a simple example: the Bloomberg value factor is one point of data, but it is composed of various sub-components (hierarchical), each capturing a different aspect of a company’s fundamentals. It is comparable to other investment style factors (relational), and its performance evolves over time (temporal). Understanding all of those makes managers much more informed about the state of their investment.

Beyond reliability, modernization of tech stacks and programmatic automation, this is also why API endpoints offer a significant advantage when it comes to preserving the integrity and richness of data models. When data is accessed through an API, it maintains its original structure, including the hierarchical relationships, relational connections, temporal dimensions, and metadata that provide essential context and meaning. This allows investors to explore the data in its full complexity, drilling down into specific aspects, comparing across different dimensions and uncovering valuable insights that would be lost if “flattened.”

What are the key components of a successful operating model for investment managers, and how can they ensure they are future-proof and can evolve with new technologies like AI?

A successful operating model for investment managers must be built on a unified platform that seamlessly integrates data management, analytics, decision support, execution and post-trade. This cohesive approach ensures a smooth flow of insights and actions, eliminating any breaks in the investment cycle.

But it’s not just about speed and efficiency; preserving the integrity of the data model is equally crucial. Think of it like a high-resolution image – if you flatten it or compress it too much, you lose the intricate details that make it valuable. Similarly, when you oversimplify or fragment your data model, you risk losing the rich context and multi-dimensional insights that can give you a competitive edge.

Now, as we look to the future and the growing role of AI, having a solid foundation becomes even more critical. AI models are only as good as the data they’re sitting on, whether we are talking about training, fine-tuning or RAG (retrieval-augmented generation), so a well-structured, high-quality data model is essential.

Finally, to truly harness the power of AI, an API-driven architecture is the way to go. It’s like having a modular home – you can easily add or upgrade rooms as needed without tearing down the entire structure. APIs allow you to plug in AI tools to expand their capabilities.