Description

A focused course, tailored for you

The Model Risk Validation Workbench for Regional Bank Specialists

Build a validation file that holds up to SR 11-7 review, with challenger-model evidence, ongoing performance monitoring, and a clean effective-challenge trail.

The validation memo is the first thing the examiner reads. If effective challenge cannot be seen in the file, the model gets a finding regardless of how well it performs.

$199 one-time

Tailored to your situation. Access within 24 hours. 30-day money-back.

Includes a hand-built implementation playbook delivered alongside course access, generated for your specific situation.

Why this course

Model Risk Management specialists at regional and super-regional banks live between two pressures. The first is the model inventory, which keeps growing as the bank adds credit-decisioning logic, AML transaction monitoring scores, deposit attrition forecasts, capital and liquidity components, and increasingly machine-learning challenger models. The second is OCC SR 11-7 and the matching Federal Reserve guidance, which expect documented effective challenge for every tier-one and tier-two model, with conceptual soundness, ongoing performance monitoring, and outcomes analysis that the validator owns independently from the developer. The gap most validators run into is not skill, it is structure. Challenger-model evidence is often a single benchmark with no sensitivity testing. The assumption and limitation register either does not exist or duplicates the developer's docs. Ongoing performance monitoring plans read like a quarterly calendar rather than a set of trigger thresholds tied to the use case. When the OCC examiner opens the validation file, they look for the seams between developer testing and independent validation testing, and if those seams are not visible, the model receives a Matter Requiring Attention regardless of its discriminatory power. The Workbench course is built so a specialist can produce a validation file with visible effective challenge, defensible tiering, and ongoing performance monitoring tied to thresholds the model owner monitors monthly.

What you walk away with

A validation file structure that makes effective challenge visible to the examiner in the first three pages.
A reusable challenger-model bench for PD, AML score, and forecast use cases that can be rerun on a new vintage of data.
A trigger-threshold ongoing performance monitoring plan that the model owner runs and the validator audits.
An assumption and limitation register that survives turnover in the validator role.
A model-tiering decision memo defensible to the OCC, with explicit linkage to the bank's risk appetite.

The 12 modules

Module 1. Model inventory and tiering decisions for a regional bank

Walk through how to define materiality and complexity for a regional-bank model inventory, including credit-decisioning models, AML transaction monitoring, deposit attrition forecasts, capital and liquidity component models, and emerging ML challengers. The module gives a tiering decision template that explicitly cites the bank's risk appetite statement and produces a tier rating defensible in front of an OCC examiner.

Module 2. Reading SR 11-7 and the matching Federal Reserve guidance as a validator

The supervisory letter is short, the expectations are deep. This module unpacks what SR 11-7 actually demands of independent validation, separates the obligations on the developer from those on the validator, and maps each expectation to a section of the validation report. Includes the parallel reading of the FRB matching guidance and the OCC handbook chapter.

Module 3. Conceptual soundness review for a PD model

A worked example using a logistic-regression probability-of-default model on a small-business or commercial portfolio. The module shows how to test the link between economic theory, variable selection, and the model's intended use, how to write the conceptual-soundness memo with explicit limitations, and how to design the challenge questions the developer must answer in writing.

Module 4. Building a challenger-model bench

Independent validation requires more than a single benchmark. This module shows how to stand up a small but defensible challenger bench: a parsimonious linear model, a tree-based model, and a domain-naive baseline. The output is a comparison table the examiner can read in one glance, with sensitivity analyses against vintage shift, segment shift, and missing-data treatment.

Module 5. Validation of an AML transaction monitoring score

The use case is an unsupervised or weakly-supervised AML scoring model feeding alert generation. The module walks through threshold setting, false-positive rate measurement against a SAR-validated sample, segment-level performance, and the model-output-to-alert-disposition feedback loop. Includes the specific challenges of validating an ML AML model when label noise dominates.

Module 6. Validation of a deposit attrition forecast

A balance-sheet model with implications for liquidity and net-interest-income forecasting. The module shows how to validate a forecast under regime change, how to test stability under rate shock scenarios, how to handle the lack of stationary ground truth in a deposit beta model, and how to write the limitations section so ALM and Treasury readers understand what the model can and cannot tell them.

Module 7. Assumption and limitation register

Most validation files duplicate the developer's documentation rather than building an independent register. This module gives a register template that survives turnover in the validator role, links each assumption to a monitoring trigger, and forces the validator to declare which limitations are acceptable, which require compensating controls, and which require remediation before the next use of the model.

Module 8. Ongoing performance monitoring tied to trigger thresholds

Quarterly performance reviews are calendar exercises. The module shows how to convert them into trigger-threshold monitoring: PSI on key features, KS or AUC drift on the score, segment-level break thresholds, and outcome-vs-prediction tests. Each trigger has an owner, a remediation path, and a re-validation decision rule the model risk committee approves up front.

Module 9. Outcomes analysis for credit-decisioning models

Outcomes analysis is the part the developer cannot do for you. The module shows how to design the back-test, how to handle censoring and survivorship in a small portfolio, how to interpret rank-ordering vs calibration breaks, and how to write the outcomes section so the model risk committee can decide between recalibration, redevelopment, and decommission.

Module 10. Effective challenge as a visible record

Effective challenge happens in conversation but it must live on paper. The module shows how to document the challenge log: what the validator asked, what the developer answered, what evidence was produced, and which open items remain. Includes the structure for the validation report's effective-challenge section that the OCC examiner reads as proof of independence.

Module 11. Writing the validation report the examiner reads first

Most validation reports bury the answer. This module rebuilds the report so the rating, the open findings, and the conditions on use sit in the first three pages, the supporting evidence sits in the body, and the technical appendices sit in the back. Includes a section-by-section template, the standard rating definitions, and the language patterns that survive review by Internal Audit and the OCC.

Module 12. Handling ML challengers and the rise of model risk in AI use cases

Regional banks are testing tree-based and neural challengers against legacy logistic models, and increasingly looking at generative AI for fraud-narrative drafting and adverse-action explanations. The module walks through the additional validation tests an ML challenger demands, the explainability standard the validator should require, and the new use-case categories the inventory needs to recognise as the model perimeter expands.

How this addresses your situation

Specific modules that map to what you said you are dealing with.

Module 4 (challenger bench) for the recurring finding that validations rely on a single benchmark.

Module 8 (trigger thresholds) for the recurring finding that ongoing monitoring is a calendar not a control.

Module 10 (effective challenge log) for the recurring finding that independence is not visible in the file.

Module 11 (report structure) for the recurring finding that the examiner has to hunt for the rating and the open items.

What you get with this course

Twelve written modules in the Art of Service learning environment.
A validation report template with section-by-section guidance.
A challenger-model bench worksheet for PD, AML score, and forecast use cases.
An assumption and limitation register template.
An ongoing performance monitoring plan template with trigger threshold definitions.
An effective-challenge log template.
The hand-built implementation playbook calibrated to your specific portfolio mix and current model inventory.

What you will have in hand by Day 1, Week 1, Month 1

Within 24 hours of purchase, account in the learning environment is provisioned and the implementation playbook is delivered alongside it.

Modules one through four cover inventory, tiering, supervisory expectations, and conceptual soundness in the first week of self-paced work.

Modules five through eight cover the use-case validations and the monitoring plan in the second week.

Modules nine through twelve cover outcomes analysis, effective challenge documentation, report structure, and the ML challenger expansion in the third week.

Before and after

Before

Validation reports read as a re-run of the developer's testing. Challenger evidence is thin. Ongoing performance monitoring is a quarterly calendar. Open findings are buried in the appendices. The examiner has to ask where effective challenge is.

After

Validation reports place the rating, the open findings, and the conditions on use in the first three pages. The challenger bench is reusable across vintages. Ongoing performance monitoring runs on trigger thresholds the model owner monitors monthly. Effective challenge is visible to anyone who opens the file.

What happens if you do not address this

An examiner finding on effective challenge or ongoing performance monitoring drives a Matter Requiring Attention, restricts the bank's use of the model, and lands on the validator's desk to remediate inside the supervisory cycle.

Who it is for

A Model Risk Management Specialist or Senior MRM Analyst at a regional or super-regional U.S. bank, typically owning validations for credit-decisioning, AML transaction monitoring, deposit forecasting, or capital and liquidity component models. Reports into a Director or Head of Model Risk who reports into the Chief Risk Officer. Reviewed by the OCC or the Federal Reserve and audited internally by IA. Writes between four and twelve validation reports per cycle and is accountable for the independence and rigor of each.

Who this is NOT for. Not for model developers writing first-line documentation. Not for credit officers using model outputs in underwriting. Not for AML investigators tuning alert thresholds. The course is written for the independent validator who owns the challenge.

How it arrives

Text-based course in the Art of Service learning environment, plus downloadable templates and worked examples for every module, plus the hand-built implementation playbook delivered alongside course access.

Time investment. Three weeks of self-paced reading and template work, roughly five to seven hours per week. The implementation playbook can be applied to a live validation in parallel.

Why $199 is the right number

PRMIA and GARP offer broad model-risk certifications oriented to first-line developers and to managers of the function. This course is narrower and built for the independent validator producing a defensible file for a regional or super-regional bank under OCC supervision, with worked examples on the three use cases that fill most of the validation queue.

FAQ

Is this aligned to SR 11-7 specifically, or to the broader supervisory landscape?

The spine is SR 11-7 and the matching Federal Reserve guidance. The course also maps to the OCC handbook chapter on model risk and notes the parallel expectations from the Federal Reserve for state-member banks.

Does this cover ML and generative AI validation?

Module twelve covers ML challenger validation in depth and discusses where generative AI use cases fit into the model inventory definition. The earlier modules ground the validator in the supervisory expectations the ML work must still satisfy.

Is the implementation playbook generic or built for my work?

The implementation playbook is hand-built after purchase, calibrated to the portfolio mix and model inventory you describe. It is delivered alongside course access.

What if I am at a smaller community bank rather than a regional?

The supervisory expectations scale down but do not disappear. The course material applies; the implementation playbook adjusts the depth of the challenger bench and the monitoring plan to the size of the inventory.

30-day money-back guarantee. If after a week of working through the materials this is not what you needed, reply to the receipt email and a full refund is processed. No questions, no forms.

Within 24 hours your account in the learning environment is provisioned and the tailored implementation playbook is delivered alongside it.