In September 2025, the Bank of England’s Prudential Regulation Authority (PRA) sent a clear signal to the financial industry through a thematic feedback letter authored by David Bailey: the honeymoon period for SS1/23 implementation is over.
“The honeymoon period for
SS1/23 implementation is over.”
– David Bailey,
Bank of England – PRA (Prudential Regulation Authority)
While firms have made progress in designing policies, the regulator noted a persistent “technology gap.” Specifically, auditors continue to highlight monitoring limitations and a failure to effectively use model operating boundaries to manage risk in volatile environments.
For firms aiming to move beyond “paper-based compliance,” technology is no longer an optional add-on. It is the foundational requirement for meeting the PRA’s 2026 expectations.
Bridging the “Technology Gap”: Moving from Manual to Automated
The PRA’s recent findings emphasize that “full alignment will take several years” for many firms due to manual bottlenecks. The regulator explicitly identified “better practice” as firms that have moved toward automated and granular monitoring.
Yields provides a modular Model Risk Management solution designed to meet the PRA’s expectations for scalable, technology-enabled oversight. Each module addresses a specific layer of the model risk lifecycle, while operating as one integrated platform.
Yields for Governance: This module acts as the “control tower,” using a workflow engine to automate the administrative burden of compliance, from model inventory management to documentation generation.
Yields for Performance: Rather than treating validation as an annual event, Yields for Performance allows institutions to standardize and automate all quantitative tests throughout the model lifecycle. By providing a standardized API-driven approach, it eliminates the need for engineers to manually recreate monitoring logic for every individual model.
Defining and Defending “Operating Boundaries”
A key highlight from the 2025 Bailey letter was the inconsistent use of model operating boundaries. Models trained in benign environments (e.g., low inflation) are often being used in volatile ones without clear thresholds for when they should be “turned off” or recalibrated.
How Yields addresses this:
Real-Time Flagging: When a model breaches its defined boundary, our governance module automatically generates a risk-prioritized alert. It guides the use, whether a data scientist or a risk manager, through the necessary follow-up tasks, ensuring that “boundary breaches” result in immediate action rather than just a footnote in a report.
Dynamic Thresholding: In Yields, you don’t just “monitor” a model; you define its safe operating zone. Our performance module allows you to set predefined statistical benchmarks (e.g., drift, accuracy, or sensitivity metrics) that act as the model’s operating boundary.
From Individual Models to Aggregate Reporting
The PRA expects Boards and Senior Management to understand aggregate model risk, the “big picture” of how inter-related models and data structures impact the firm’s safety and soundness.
How Yields addresses this:
- The Model Inventory: Yields provides a centralized, dynamic inventory that tracks not just individual models but their interdependencies. This allows firms to identify clusters of risk (e.g., a single data quality issue affecting multiple credit models).
- Actionable Dashboards: Yields translates complex quantitative results from the performance module into high-level risk scores in the governance module. This gives the Board a “single pane of glass” to see the firm-wide health of the model landscape, satisfying the Principle 2 requirement for clear model risk appetite monitoring.
Conclusion: Preparing for 2026
The PRA has signaled that in 2026, auditors will be looking for proof of progress in these specific areas: automated monitoring, data aggregation, and operating boundaries.
As the regulator noted, model risk remains elevated amidst geopolitical and economic uncertainty. Firms that rely on legacy, manual processes will find themselves “technically out of compliance” as the complexity of their AI and model landscapes continues to grow. Yields provides the technological bridge to turn these regulatory hurdles into a strategic advantage.
Author
Jos Gheerardyn has built the first FinTech platform that uses AI for real-time model testing and validation on an enterprise-wide scale. A zealous proponent of model risk governance & strategy, Jos is on a mission to empower quants, risk managers and model validators with smarter tools to turn model risk into a business driver. Prior to his current role, he has been active in quantitative finance both as a manager and as an analyst.
