Skip to content

Risk Academy, led by the Nexus Consortium · Risk Science and Methods · Department of Risk Management

Models that are wrong or misused, and the controls that manage them

ماڈل کا خطرہ اور توثیق

A model is a decision tool with an error term, and model risk is what happens when the error is ignored.

Map model risk and validation on a real placeSee the courses

Illustrative: a joint probability density whose dependence shifts over time

01دائرہ کار

What the discipline covers

The subtopics its courses and labs teach.

کورسز
In preparation
سطحیں
L3 Advanced
تجربہ گاہیں
In development
  • Model risk governance
  • Model validation
  • Model monitoring and drift
  • Explainability and fairness
  • Third-party models
  • Generative AI in finance
  • Human-in-the-loop controls
  • Model inventories

02نقشے

Build a twin of any place

Search any place or open the world; the layers and generators follow model risk and validation on the planet map, the globe and in 3D.

The central Apennines: L'Aquila, Amatrice and Norcia: a flat plan of the faculty twin. Terrain 24 to 2,805 m; 3 reference earthquakes, M6.2 to M6.6; Earthquakes M4.5 and above, past week: read live when the 3D twin opens; USGS ShakeMap intensity contours: read live when the 3D twin opens. Contours every 500 metres from the elevation grid; reference earthquakes as marks.
The central Apennines: L'Aquila, Amatrice and Norcia, Italy. Flat plan of the same data: relief, the three reference epicentres with magnitude and depth, and the numbers below; the sampler's figure shows one precomputed field with its parameters. The 3D twin opens on devices that can draw it, once its data and code have been checked.
  • Terrain 24 to 2,805 m
  • 3 reference earthquakes, M6.2 to M6.6
  • Earthquakes M4.5 and above, past week: read live when the 3D twin opens
  • USGS ShakeMap intensity contours: read live when the 3D twin opens
Credits and dates per layer are listed under the stage. Relief in the twin is exaggerated ×1.5 for legibility.

Drag to move, right drag or Shift with a drag to turn and tilt, Ctrl with the wheel to zoom. Keyboard: arrows move, Shift with arrows turn and tilt, + and – zoom, 0 resets, V steps through the views, Escape clears.

Flat plan of the same data: relief, the three reference epicentres with magnitude and depth, and the numbers below; the sampler's figure shows one precomputed field with its parameters. The 3D twin opens on devices that can draw it, once its data and code have been checked.

The boundaries, names and designations shown do not imply any judgement on the legal status of any territory, or any endorsement or acceptance of such boundaries. Countries: Natural Earth (public domain).

03The subject

The subject, methods and measures and debates and bodies

The discipline

The field and its scope

For model validators, model owners, data scientists and risk committees who must decide whether a model is fit for the decision it supports and keep it that way in use.

Model risk and validation study the losses and harms that follow when decisions rest on models that are wrong, misused or no longer fit the world, and the controls that reduce that risk: inventories, tiering by materiality, independent validation, performance monitoring, change control and clear limits on use. The field took shape in banking after pricing and credit model failures and was codified in the Federal Reserve's SR 11-7 (2011) and the Bank of England's SS1/23 (2023); its methods now apply to insurance, catastrophe, climate, epidemiological and machine learning models, including generative systems whose outputs are hard to test against a ground truth.

The discipline is organised around the model life cycle. Development documents purpose, data, assumptions and limitations; validation tests conceptual soundness, implementation and outcomes with back-testing, benchmarking and sensitivity analysis; monitoring tracks drift in inputs and performance; governance records the inventory, the tiering and who may approve what. Explainability and fairness testing enter where models affect people. The standing questions are what validation can show for models of rare events, how much independence is enough, and how to govern third-party and foundation models whose internals are not available to the user.

Measures

Quantities and units

QuantityUnit or scaleWhat to know
Back-test exceedance rateCount of exceptions per period against the expected count at the model's confidence levelFor a quantile forecast such as value at risk, the number of times loss exceeded the forecast. Tests such as Kupiec's are weak for small samples: a year of daily data cannot reject a 99 per cent model whose true rate is 1.5 per cent.
Discrimination (AUC, Gini) and calibration (expected versus observed rate)Dimensionless, 0.5 to 1 for AUC; ratio or calibration slope for calibrationStandard performance statistics for classification models. Both are estimated from a sample with a confidence interval and both degrade when the population shifts; discrimination can stay high while calibration fails.
Population stability index and feature drift statisticsDimensionless index; or a test statistic per featureMeasures of how far the input distribution has moved from the development sample. Thresholds in use are conventions, not derived from loss consequences.
Model materiality tierOrdinal tier from exposure, complexity and useDrives the depth and frequency of validation. It is a governance judgement; the uncertainty lies in whether exposure through a model is known at all, especially for models embedded in other models.
Sensitivity of output to assumptionsChange in output per unit change in a parameter, or a Sobol index share of varianceShows which assumptions matter. Local sensitivities miss interactions; global sensitivity analysis needs many model runs that slow models cannot afford.
Validation finding count by severity and ageCounts; days openThe operational state of the control. A finding backlog with long ages signals that validation has no teeth; the count alone says nothing about the risk the findings carry.

طریقے

Methods and their limits

Each method is listed with what it assumes and where it fails.

طریقہAssumesFails when
Independent validation (conceptual soundness, implementation testing, outcomes analysis)That a team outside the model's development line can read the documentation, reproduce key results, test the code and judge the assumptions against the use.When validators lack the domain knowledge or the data to reproduce, when the model is a black box from a vendor, or when validation becomes a document review that never runs the model.
Back-testing and out-of-sample testingThat past outcomes are a fair test of the model's future use, that enough independent observations exist, and that the test is set before the results are seen.For rare events where the test has no power, when the data were used in development, or when the environment changed so that past performance no longer bears on the future.
Benchmarking and challenger modelsThat a simpler or alternative model, run on the same data, exposes where the production model's assumptions drive its output.When the benchmark shares the production model's data and blind spots, or when agreement between models is read as validation against reality.
Monitoring for drift and performance decayThat input distributions and outcomes are observed in use with little delay, and that thresholds trigger review or retraining.When outcomes arrive years later (credit default, climate), when the outcome is itself shaped by the model's decisions, or when thresholds are set without reference to the harm from a stale model.
Explainability and fairness testingThat feature attributions (for example Shapley values) and group performance comparisons reveal whether a model relies on acceptable factors and treats groups comparably.When attributions are unstable across methods, when fairness metrics conflict and no rule chooses between them, or when the protected attribute is not recorded and proxies cannot be traced.
Model inventory with tiering and use restrictionsThat every model can be found, described, assigned an owner and a tier, and that use beyond the documented purpose is detectable.When spreadsheets, embedded rules and third-party components are not counted as models, or when the inventory is a list without links to validation evidence.

Debates

What the field debates

Should high-stakes models be interpretable by design, or is post-hoc explanation of complex models enough?
Rudin (2019) argues that interpretable models perform comparably on most tabular problems and that post-hoc explanations are approximations of an approximation. The other side points to tasks (images, text, weather) where complex models are far more accurate, to supervisory acceptance of explanation methods, and holds that the requirement is reliable testing and monitoring rather than a particular model class.
What does validation mean for models of rare events that cannot be back-tested?
One view, following Oreskes and co-authors (1994), holds that such models can be confirmed in part but never validated, so the control is transparency of assumptions and sensitivity. The other holds that component-level testing, benchmarking and historical event reconstruction give adequate evidence, and that refusing the word validation leaves decision makers without a usable verdict.
Can generative and foundation models be governed under existing model risk frameworks?
Supervisors and SS1/23 hold that the principles apply: define the use, test for it, monitor and restrict. Critics reply that the user cannot inspect training data or weights, that outputs are open-ended so outcome testing is partial, and that NIST's AI Risk Management Framework and ISO/IEC 42001 add the missing elements of use-case governance and incident reporting; the evidence base of documented failures is still young.

Bodies and data

Bodies that set practice, and the data they publish

Bodies and standards are named so learners can find and apply their published work; the Academy teaches their use and links to the publisher, and no affiliation is implied.

BodyRole in the field
Board of Governors of the Federal Reserve System and the Office of the Comptroller of the CurrencyIssued SR 11-7 / OCC 2011-12, the supervisory guidance on model risk management that defined the field for banks.
Bank of England Prudential Regulation AuthorityIssued supervisory statement SS1/23 on model risk management principles for banks, in force from May 2024, extending the scope to all models including those using artificial intelligence.
National Institute of Standards and Technology (NIST)Publishes the AI Risk Management Framework (AI 100-1) and its generative AI profile, used as a reference for governing machine learning models beyond finance.
ISO/IEC JTC 1/SC 42 Artificial intelligenceMaintains ISO/IEC 42001 (AI management systems), ISO/IEC 23894 (AI risk management guidance) and related standards on data quality and bias.
European Banking Authority and European Central BankSet model requirements for internal ratings and run the targeted review of internal models that tests how banks' models are validated.
DatasetSteward
AI Incident DatabaseResponsible AI Collaborative
OECD AI Incidents MonitorOrganisation for Economic Co-operation and Development
Supervisory stress test scenarios and resultsBoard of Governors of the Federal Reserve System

04طریقہ

From a model's claim to a back-test verdict: expected exceptions, observed exceptions and the zone the count falls in

A model that claims a confidence level makes a testable promise about how often outcomes will exceed its figure; counting those exceptions over a test window and comparing the count with the expected distribution gives the verdict, and the verdict feeds the model's tier and its permitted use.

Sources: Board of Governors of the Federal Reserve System, SR 11-7: Guidance on Model Risk Management (2011); Basel Committee on Banking Supervision, Supervisory framework for the use of backtesting in conjunction with the internal models approach (1996); Bank of England PRA, SS1/23 Model risk management principles for banks (2023); Box, Science and statistics, Journal of the American Statistical Association (1976); Kupiec, Techniques for verifying the accuracy of risk measurement models, Journal of Derivatives (1995).

05At a glance

At a glance, with 5 related topics

3D گراف

فیکلٹی برائے سائنسِ خطرہ اور طریقۂ کار

265 nodes1,691 relations9 views

Explore the 3D graph: Faculty of Risk Science and Methods

06کنسول

How Model risk and validation is taught, as a process you can run

Press Play to run this process from this page's own data.

ماڈل کا خطرہ اور توثیق

How Model risk and validation is taught

5 subjects in scope, 3 key terms, courses in development, 1 lab, in the Risk Management department of Risk Science and Methods: from the scope to the method and its terms, the courses that teach it, the labs on the twin, the assessment and the record.

  1. 01 Scope

    1. 01

      Model risk governance

      Start · the discipline

      Model risk governance

    2. 02

      Model validation

      Start · the discipline

      Model validation

    3. 03

      Model monitoring and…

      Start · the discipline

      Model monitoring and drift

    4. 04

      Explainability and…

      Start · the discipline

      Explainability and fairness

    5. 05

      Third-party models

      Start · the discipline

      Third-party models

  2. 02 Method and terms

    1. 06

      From a model's claim to…

      Model · the learner

      A model that claims a confidence level makes a testable promise about how often outcomes will exceed its figure; counting those exceptions…

    2. 07

      Model risk

      Record · the discipline

      The potential for loss or poor decisions from errors in models or their misuse.

    3. 09

      توثیق

      Record · the discipline

      Independent review that a model is sound and fit for its purpose.

    4. 11

      Back-testing

      Record · the discipline

      Comparing model predictions with what happened.

  3. 03 Courses

    1. 08

      RAT-217 Model Risk…

      Record · the learner

      Model Risk Management and Validation

  4. 04 Labs

    1. 10

      جڑواں ماڈل

      Exercise · the learner, on the faculty twin

      The method runs on the faculty twin and the admitted map layers.

  5. 05 Assess

    1. 12

      Module checks

      Gate · the course

      Each module ends with a short check against its outcomes.

    2. 13

      Assessed work

      Decision · the learner

      The assessed work that closes a course: a lab record, a case memo or an assignment, reviewed against the rubric.

  6. 06 Record

    1. 14

      تکمیل کا ریکارڈ

      Outcome · the learner

      The record stacks into the discipline achievement.

    2. 15

      Discipline achievement

      Record · the Academy

      The discipline achievement follows the discipline rule, read from the records.

Model risk and validation · Risk Management An outline of how the work is designed to run; it does not report the status of any decision or record.

07تعلیم

Learning, practice and connections

Achievements

Achievements in this discipline

  1. Course achievements

    One for each course with this as its primary discipline, earned when you pass the course's assessed activities.

  2. Discipline: Model risk and validation

    Complete 3 assessed courses in model risk and validation, at least 2 with it as their primary discipline and at least 1 at Intermediate (L2) or above, totalling at least 45 measured learning hours. In development.

  3. Department: Risk Management

    Earn two discipline achievements in the Department of Risk Management. In development.

  4. Programme: Nexus-Based Risk Management flagship pathway

    Complete every core course of the Nexus-Based Risk Management flagship pathway under assessed completion, ending with its capstone. In development.

An achievement is a record of the learning completed and assessed here. How recognition works.

راستے

Pathways that teach it

Where this is practised

Where the subject is practised

  • The Global Risks Alliance (GRA)AI and Financial Model RiskCredit and pricing models, Explainability and fairness, Generative AI in finance, Human in the loop controls, Model monitoring and drift, Model risk governance, Stress and scenario models, Third party models

Field platforms of the ecosystem where the subject is studied and applied.

Practised in

The three Foundry systems where it is practised

Foundry system pages describe what each system and its rooms are designed to do; the courses teach the methods these systems apply.

Across the ecosystem

Across the ecosystem

The science, platforms and standards this discipline draws on and contributes to across the ecosystem.

Foundations in the STEM Commons

Standards for the field

Pages of the ecosystem, cited so learners can follow the field into research, platforms, standards and national practice.

08حصہ لیں

Take part in Risk Science and Methods

اپنی شناخت سے شروع کریں۔

Routes

  • راستہ

    قومی ادارے

    Ministries, statistics offices and agencies: a national desk and chapter, and a seat in the working groups. National chapters

  • راستہ

    Sectors and professions

    Operators, regulators, banks, insurers and professional bodies whose work touches risk science and methods. For organisations

  • راستہ

    تعلیم

    Courses and practice, with a record of every course completed. Free to study. Open learning

  • راستہ

    معیارات

    Draft, review and adopt the methods and data standards with the working groups. Working groups

  • راستہ

    Communities and research

    Universities, civil society and individual experts: projects, data and review. Write to the Academy

Open calls

Open callSeeksCloses
Founding call for experts to the working groups CALL-FOUNDING-EXPERTSexpert, editor, chair2026-12-15
Open nomination for the three independent expert seats of the Global Standards Council CALL-GSC-INDکونسل کا رکن2026-12-01
Nominations of interim regional members (applies only if the transitional provision is decided) CALL-GSC-REGکونسل کا رکن2026-12-15
Review of the wave-1 risk core outlines (C1, C2, C3, C4, C8, C10, C11) CALL-REVIEW-WAVE1contributor2027-01-31
Experts for the core standards of risk science: uncertainty and evidence grading, scales, structured expert judgement, tolerability, extremes and compound events, catastrophic and existential risk, ethics and equity, units and measurement (NRS-C17 to NRS-C24) CALL-CORE-SCIENCEexpert, editor, chair2027-01-31
Independent reviewers for the families of the hazard atlas, three from at least two regions for each family (NHS2-R19), priority families first CALL-ATLAS-REVIEWماہر2027-03-31
Members of the interim Frontier Safety Review Panel and red-team reviewers (biosafety and biosecurity, AI safety and security, critical-infrastructure security); applications treated in confidence CALL-FRONTIER-SAFETY-REVIEWماہر2026-11-30
Experts in ontology, metrology, software verification and conformity assessment for the requirement register, the NHS-2 machine representation, the conformance test suites and the conformity assessment scheme CALL-CONFORMANCE-ONTOLOGYexpert, editor2027-01-31

Calls past their closing date are hidden automatically. Every call, proposal and route: open areas and calls.

Cells and CERTs

  1. Gather the cell

    Bring together three to seven people with complementary skills: data, field practice, policy and communication.

  2. Register it

    Write to the team with the cell's name, members, area and region. You receive a reference and a contact. Write to the Academy

  3. Choose a charter

    Pick one focus: research and evidence, data and maps, standards review, or community preparedness.

  4. Learn together

    Follow shared study paths at Risk Academy, with practice and a record of every course completed. Hazards and resilience

  5. Lead and report

    Hold regular sessions. Publish what the cell learns through Reports. Bring findings to the working groups. Working groups

  • Specialise as a Community Emergency Response Team (CERT)

    A CERT prepares its neighbourhood for the hazards it faces. It builds a local hazard profile from this site's map. It trains in household and street preparedness. It learns first aid and light search through recognised local providers, and it runs drills. A CERT works alongside the local emergency services and follows their direction. It never replaces them and is not an emergency service.

انڈیکس

What a national team or a local community can do on The Global Risks Index (NXS-GRX) today.

  • NXS-GRX

    Your country's profile

    Open risk profiles for 195 countries, with sources and licences on every value. Open the Index

  • NXS-GRX

    Correct or reply

    Correct a value or reply for your country. The change is logged with its effect. Open the Index

  • NXS-GRX

    Offer data

    Offer a dataset. It enters through the licence gate, with your credit. Open the Index

  • NXS-GRX

    Use the open data

    Layers, profiles and tiles through the open API. Open the Index

09References

Core references and where to begin

Start with Board of Governors of the Federal Reserve System, then Bank of England Prudential Regulation Authority, then Box.

  1. Board of Governors of the Federal Reserve System, SR 11-7: Guidance on Model Risk Management (2011). federalreserve.gov
  2. Bank of England Prudential Regulation Authority, SS1/23 Model risk management principles for banks (2023). bankofengland.co.uk
  3. National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0) (2023). DOI
  4. Box, G. E. P., Science and statistics, Journal of the American Statistical Association (1976). DOI
  5. Oreskes, N., Shrader-Frechette, K. and Belitz, K., Verification, validation, and confirmation of numerical models in the earth sciences, Science (1994). DOI
  6. Rudin, C., Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead, Nature Machine Intelligence (2019). DOI
  7. Gama, J., Žliobaitė, I., Bifet, A., Pechenizkiy, M. and Bouchachia, A., A survey on concept drift adaptation, ACM Computing Surveys (2014). DOI
  8. Mitchell, M. et al., Model cards for model reporting, Proceedings of the Conference on Fairness, Accountability, and Transparency (2019). DOI
  9. International Organization for Standardization and IEC, ISO/IEC 42001:2023 Artificial intelligence management system (2023). iso.org

10سوالات

سوالات، جوابات اور کلیدی اصطلاحات

کیا یہاں تعلیم مفت ہے؟

جی ہاں۔ Risk Academy، جس کی قیادت Nexus Consortium کرتا ہے، کا ہر کورس اور تعلیمی مواد مفت ہے، اپنی رفتار سے یا کسی گروپ کے ساتھ۔ عام شرکت کی کوئی فیس نہیں؛ معاوضے والا کام، انٹرن شپس اور فیلوشپس الگ انتظامات ہیں۔

کورس مکمل کرنے سے مجھے کیا ملتا ہے؟

A completion record: the course, its version, the date and what was assessed, private by default and shared by you. It stacks into discipline, department and programme achievements. A record states what was studied and assessed and does not certify competence; it is not a licence, a professional title or an employment status.

ڈیٹا اور مقامات کہاں سے آتے ہیں؟

Every faculty teaches on a twin of a real place built from open data; each layer carries its provider, licence and date, and every result on the Academy states its source. Maps and figures are indicators for teaching, never an assessment of a place.

How do I take part beyond the courses in model risk and validation?

Form a competence cell of three to seven people, study a shared pathway, take a supervised assignment in the National Working Groups, or bring a cohort from your institution. For organisations.

کلیدی اصطلاحات

Model risk
The potential for loss or poor decisions from errors in models or their misuse.
توثیق
Independent review that a model is sound and fit for its purpose.
Back-testing
Comparing model predictions with what happened.
انحراف
Change in data or relationships that degrades a model over time.

رپورٹیںتعاون کریں

اس صفحے سے شواہد فراہم کریں۔

Reports is the Consortium's quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.

Risk Academy

Never send personal data, sensitive infrastructure detail or anything you have no right to share. A contribution is reviewed before use; sending it confers no authorship or editorial rights, and sponsorship never buys a finding.

ماڈل کا خطرہ اور توثیق

Prepare with a related course

Open courses in connected disciplines teach the methods this discipline will build on.

کورس کیٹلاگ دیکھیںBack to Department of Risk Management