SR 11-7 is supervisory guidance issued in April 2011 by the Federal Reserve (SR 11-7) and the Office of the Comptroller of the Currency (Bulletin 2011-12), addressed to banking organisations. It defines a model broadly as any quantitative method, system or approach that produces estimates for use in decisions, and sets expectations across development, implementation, use, independent validation, governance, policies and controls. See federalreserve.gov/supervisionreg/srletters/sr1107.htm and occ.treas.gov/news-issuances/bulletins/2011/bulletin-2011-12.html.
For bank-affiliated asset managers, including bank holding company affiliates, OCC-regulated trust companies and primary-dealer affiliates, SR 11-7 applies directly through the parent's model risk management framework. Models live inside the parent bank's inventory, materiality classification and validation cycle, and findings flow to the bank's MRM committee rather than to a standalone buy-side governance body.
For non-bank asset managers, including independent investment advisers, mutual fund complexes, hedge funds and private equity firms, SR 11-7 is not legally binding. It has nevertheless become the de facto standard for buy-side model risk because (a) SEC examiners under the Investment Advisers Act and the Investment Company Act expect comparable practice for material models, (b) institutional allocators ask for SR 11-7 style artefacts during operational due diligence, and (c) prime-broker and custodian counterparties require alignment as a precondition of doing business.
A single firmwide register covering risk models, pricing and valuation models, liquidity-bucketing models used for Rule 22e-4, stress test models, fund flow forecasting and vendor or third-party models. Each entry carries a named owner, a use description and a material or non-material classification.
Materiality is tied to assets influenced, the nature of risk-altering decisions, and regulatory or financial-reporting use. Material models follow the full SR 11-7 lifecycle of development standards, independent validation and ongoing monitoring. Non-material models follow a lighter cadence with proportionate documentation.
Documented methodology, data quality testing, code review, version control and a conceptual soundness sign-off before production. Buy-side firms commonly add explicit treatment of look-ahead bias, survivorship and overfitting checks for any model that drives portfolio decisions.
A function independent of the build team validates each material model on a defined cadence, typically annually for high-materiality models and less frequently for medium materiality. Validation covers conceptual soundness, ongoing monitoring and outcomes analysis against realised performance.
Validators must be senior, qualified and empowered to challenge the build team. Findings are tracked through to remediation with deadlines and accountable owners. Effective challenge is the most frequently cited test of MRM credibility on the buy side and is the area examiners probe hardest.
Performance monitoring on production data with defined trigger thresholds and a documented response when thresholds breach. The buy-side flavour adds a portfolio-impact assessment alongside statistical metrics, so deteriorating model performance is read in terms of client outcomes, not only error rates.
Oversight by the Adviser's own board or a board committee, not the funds' boards. A Chief Risk Officer or equivalent owns the MRM policy. An annual report to the board covers inventory status, validation completion, open findings and material changes since the prior cycle.
Examines registered investment advisers under the Investment Advisers Act of 1940 and registered investment companies under the Investment Company Act of 1940. Examinations include scrutiny of material models used in valuation, liquidity classification under Rule 22e-4 and risk.
SR 11-7 and OCC Bulletin 2011-12 apply directly to banking organisations. Bank-affiliated asset managers sit inside the parent's MRM programme and supervisory cycle, with examination findings flowing through the parent.
Supervises broker-dealer affiliates of asset managers, including order-routing, best-execution and trade-surveillance models that touch the broker-dealer entity.
Supervises commodity trading advisers and commodity pool operators using systematic or algorithmic models in managed futures and derivatives strategies.
SEC examination priorities have repeatedly highlighted valuation, liquidity classification under Rule 22e-4, and material risk models as areas of focus. Common themes in published priorities and risk alerts include validation independence, documentation completeness, and the use of vendor models without sufficient internal challenge. See sec.gov.
Bank-affiliated managers face a different practical reality. The parent bank's MRM cycle drives their pace, and SR 11-7 and OCC Bulletin 2011-12 are non-negotiable. Validators sit inside the bank's second line of defence, with reporting to the parent's MRM committee. See federalreserve.gov and occ.treas.gov.
A growing trend on the buy side is the inclusion of AI and machine learning systems, particularly for portfolio construction, signal generation, factor research and ESG screening, inside the firm's model inventory. The challenge is treating these systems as models without overbuilding governance around exploratory research code that never reaches a client portfolio.
| Adjacent rule | How it interacts |
|---|---|
| Investment Advisers Act of 1940 (Section 206 fiduciary duty) | Section 206 fiduciary duty supports the SEC's expectation that material models supporting client outcomes are sound, validated and overseen. SR 11-7 supplies the operational framework that satisfies that expectation in practice, even though the Advisers Act does not name it. |
| Investment Company Act of 1940 (Rule 22e-4 liquidity risk management) | Rule 22e-4 requires open-end funds to classify portfolio investments into liquidity buckets. The classification methodology is explicitly a model, and SEC staff expect MRM rigour around it, including documented methodology, validation and ongoing monitoring. |
| EU AI Act (Regulation 2024/1689) | For managers with EU activity, AI used in services affecting EU clients may carry transparency obligations under Article 50 and broader provider duties where systems are high risk. SR 11-7 MRM artefacts substantially feed Annex IV technical documentation and risk management evidence. |
| NIST AI Risk Management Framework | The NIST AI RMF Govern, Map, Measure and Manage functions sit comfortably alongside SR 11-7 lifecycle stages. The framework is the practical reference when extending MRM to AI and machine learning systems on the buy side, and is well understood by US examiners. |
| ISO/IEC 42001 (AI management system) | ISO/IEC 42001 was published in December 2023 as the first certifiable management-system standard for AI. Buy-side firms standing up an AI policy increasingly map controls to 42001 alongside SR 11-7 MRM, particularly where group-level certification is in scope. |
“The buy-side mistake is to copy a Tier 1 bank's MRM machinery whole. The right move is to size validation rigour to materiality, and to extend the framework cleanly to AI without turning every Jupyter notebook into a production model.”
Mandatory only for bank-affiliated managers through the parent. Independent advisers and fund complexes use it as the de facto standard because examiners, allocators and counterparties expect it.
SR 11-7 defines a model as a quantitative method, system or approach that produces output for decisions. On the buy side this captures valuation models, Rule 22e-4 liquidity classification, risk models, factor models, signal-generation systems and most material screening tools.
Treat material AI and ML systems as models with the same lifecycle expectations: documented methodology, independent validation, ongoing monitoring and governance. Allow lighter governance for exploratory research code that does not influence client outcomes, with a clear materiality gate before promotion to production.
Independence is required at the function level for material models, not necessarily a large internal team. Small and mid-sized firms commonly use independent external validators or a ring-fenced internal reviewer, provided the validator is senior, qualified and empowered to exercise effective challenge.
8 to 16 weeks to a production-ready MRM artefact set: model inventory, materiality classification, validation reports for material models, ongoing monitoring plan and an audit pack mapped to SR 11-7, NIST AI RMF and ISO/IEC 42001.
SR 11-7 is 2011 US bank supervisory guidance from the Federal Reserve and OCC. Bank-affiliated asset managers sit under it through the parent; independent advisers and fund complexes have adopted it as the de facto buy-side standard. The inventory typically covers valuation, Rule 22e-4 liquidity classification, risk, factor and AI or ML signal models. Supervision runs through the SEC for advisers and funds, the Fed and OCC for bank-affiliated managers, FINRA for broker-dealer affiliates and the CFTC and NFA for CTAs and CPOs. Moweb delivers a right-sized MRM artefact set in 8 to 16 weeks.