Risk Data Lake: Applied risk analytics on the Moody’s Intelligent Risk Platform

Insurance risk analytics is at an inflection point, as organizations manage rapidly growing data volumes, increasing model complexity, and heightened expectations to deliver insights quickly—particularly during renewal periods or catastrophe events.

To gain scalable infrastructure for developing and deploying advanced analytical capabilities, many firms have explored, evaluated, and invested in data lake technologies.

However, many have found that traditional (or ‘vanilla’) data lakes provide scale and modern tools but often fall short of delivering the expected business value.

Across the market, a consistent pattern emerges: an assumption that centralizing data in a lake will, on its own, lead to faster, better decisions. There are high expectations that using a data lake will transform the speed and quality of decisions, but in practice, without domain context, integrated workflows, and embedded analytics, data lakes can devolve into a costly storage layer.

The result is often longer cycle times from data ingestion to insight, with limited measurable return on investment, highlighting that the primary challenge is not infrastructure; it is usability, context, and time to insight.

Introducing Moody’s Risk Data Lake: An applied data lake for insurance risk analytics

Moody’s Risk Data Lake was designed to help insurance organizations reduce time to insight and improve the usability of risk data. Rather than serving as a generic repository for catastrophe risk data, it applies insurance-specific structures and analytics to support day-to-day risk and portfolio decisions.

Risk Data Lake integrates with Moody’s Intelligent Risk Platform™ (IRP) and organizes data around risk-relevant concepts such as portfolios, policies, and locations.

This domain-aware structure helps teams move more quickly from raw data through analysis and decision-ready insight. Risk Data Lake also reduces unnecessary data movement, allowing analysts to access, query, and visualize data in one place—supporting stronger governance, lowering operational overhead, and improving performance.

Analytics for every team: From dashboards to advanced modeling

Risk Data Lake supports a range of users—from business analysts to data scientists—so organizations can standardize analytics on a single, governed foundation. For business users and analysts, this provides a low-code, drag-and-drop reporting experience with embedded dashboards.

Teams can create standardized views of exposure and loss metrics without writing any code and publish dashboards into persona-centric IRP applications such as Risk Modeler™, UnderwriteIQ™, ExposureIQ™, and TreatyIQ™ for consistent consumption across the organization.

For technical teams, Risk Data Lake supports SQL access and Python and R notebooks, enabling advanced analytics and model development within the same governed environment.

Security and entitlements configured in the IRP are applied consistently through the analytics layer. Role-level security helps ensure that users access only the data they are permitted to see—supporting governance requirements that can be difficult to enforce in general-purpose data lakes.

Client use case: Faster comparative analytics with measurable efficiency gains

Comparative analytics is central to the risk management lifecycle. Catastrophe modelers are often tasked with answering multiple types of benchmarking questions, such as how changes could apply to multiple workflows or use cases. For example:

• Model validation: How do modeled losses change when transitioning to a new model version? What are the primary drivers of differences in loss estimates between the old and the new model versions?
• Portfolio composition: How has the composition of my portfolio changed compared to the prior year? What is the quantified impact of portfolio changes on modeled losses? Which factors or drivers are contributing the most to year-over-year differences in risk?
• Custom-view-of-risk and What-if scenarios: Is our actual loss experience aligned with our view-of-risk? Should we run models with a higher sample size? How would our view of risk change?
• Market share analysis: How does my portfolio’s exposure and loss profile compare to the broader market for a specific region or peril? How do my results benchmark against Industry Exposure Databases (IED) and Industry Loss Curves (ILC)?

To answer these seemingly straightforward questions, the required underlying analytics often span multiple disconnected environments, require significant data engineering to rekey portfolios to benchmark the analysis, and significant data movement to get the analysis into a business intelligence tool to help visualize and communicate the answer.

Figure 1: Risk Data Lake within the Moody's Intelligent Risk Platform unifies data formats and inputs from a client's infrastructure, databases, and workflows.

Comparative analytics in the Risk Data Lake

The ability to easily compare or run benchmarking analysis is driven by three key capabilities of the Risk Data Lake:

1. The Risk Data Catalog catalogs the syntax of the data, allowing you to locate operational and archived data stored on the platform.
2. The unified data architecture within the platform allows for archived and operational data to be accessible, side by side, within the same tools.
3. The Risk Data Lake reporting engine enables teams to design interactive dashboards that support comparative analysis and clear visualization.

All three capabilities run inside the Intelligent Risk Platform, without requiring a business user to move a single byte of data or write a single line of code. Just set the benchmarking parameters in the highly intuitive user experience, run the models (if necessary), and choose the best visualization to communicate changes to risk stakeholders.

Let’s walk through an example. The illustration below compares two analyses and provides insights into exceedance probability (EP) and average annual loss (AAL) alongside the exposure characteristics.

Step 1: Define parameters

In the illustration above, a notional portfolio (Analysis #1) is compared with an IED-based portfolio (Analysis #2). Both portfolios were processed using a Moody’s RMS Windstorm model.

The dashboard supports multiple comparisons across any two analyses. For example, exceedance probability (EP) curves can be displayed side by side across EP types, with the option to filter to a specific EP type for detailed review. As selections change, the dashboard retrieves updated results in real time directly from the IRP data layer.

Step 2: Compare results (EP)

As the screenshot below shows, users can compare exceedance probability statistics side-by-side. Users can view all EP result types or select a specific EP type; the dashboard can also provide standard return-period losses from both analyses.

Step 3: Compare results through drill-downs (AAL) 

As shown above, the dashboard can provide results sliced and diced in different ways. For example, average annual loss can be viewed at the Admin1 (State) level as well as at the construction and occupancy levels. Also, additional filters can be introduced, such as for the Prospective Code.

The user can select Ground Up, Gross, or any other perspective and view the analytics and derive insights.

The dashboard also supports interactive drilldowns. In the example below, users can select a specific geography (North Carolina, in this example), which can reveal underlying metrics and support additional investigation.

Step 4: Define the right dashboards and summarize insights

Results can also be visualized on maps (see example below) at the individual location level or at higher geographic resolutions (for example, county). In addition to core metrics such as average annual loss (AAL), maps can display derived attributes.

For example, users can calculate loss cost by dividing ground-up AAL by total insurable value (TIV) and visualize the result at the county level. These analyses can be developed and delivered efficiently within Risk Data Lake.

This is just one example of a comparative analytics dashboard. Using Risk Data Lake’s designer experience, IRP users can create dashboards aligned to specific analytical needs and publish them to applications deployed on the Intelligent Risk Platform, including Risk Modeler, ExposureIQ, UnderwriteIQ, and TreatyIQ.

In the future, these dashboards can also be delivered to broader stakeholder groups through client-facing applications, helping extend insights beyond core modeling users.

Built for scale

Scalability is not an afterthought in Risk Data Lake—it is foundational.

Queries run directly against the data without caching or unnecessary duplication. Compute resources process data where it resides, allowing performance to scale even as data volumes grow exponentially.

We saw one customer previewing Risk Data Lake who successfully analyzed location-level output from a Moody's RMS high-definition (HD) model consisting of 4.6 billion records, joined it with exposure data, and produced exhibits to analyze convergence within the model.

The client would spend several hours exporting these results to an estimated 250 GB Results Data Module (RDM). Instead, they produced the insights in less than an hour.

The architecture ensures that whether users are building dashboards, accessing dashboards, running SQL queries, or executing advanced notebooks, performance remains consistent and reliable.

Agentic AI and the future of risk analytics

As AI capabilities continue to evolve, many organizations are exploring how to streamline analytical workflows and expand access to insights. Risk Data Lake is designed to support these emerging workflows within a governed analytics environment.

In this blog, we outlined three key dashboards for comparing losses between two different models. Although we successfully streamlined the process of comparing portfolio losses, we’re already taking this simplification further with new workflow agents developed in Moody’s Risk Labs.

Imagine if, instead of manually working through each step outlined above, we could write a prompt to automate the workflow: "Compare the results of version X and version Y of the windstorm model on portfolio Z. Create four dashboards that effectively communicate the changes in loss results."

Join us at Exceedance 2026 (Fort Lauderdale, June 1-4) to see it live.

Conclusion

Organizations seeking to advance their risk analytics capabilities and evaluate traditional data lake approaches should consider Moody’s Risk Data Lake, a purpose-built platform for risk analytics.

Risk Data Lake provides scalable, high-performance architecture, near-real-time access to data stored in Moody’s Intelligent Risk Platform, and analytical capabilities ranging from code-free dashboards to SQL query access. Users can generate insights while minimizing data movement and operational overhead.

For organizations encountering the limitations of traditional data lakes, Risk Data Lake represents a shift from infrastructure-first efforts to outcomes-focused risk analytics.

Modern risk management requires more than storage and compute—it requires analytics that are purpose-built for risk decisions. Risk Data Lake is designed to help deliver that capability at scale.