Wirewalk Instituteof AI
pro-250 Intermediate Available

Building AI Risk Analytics: Scenario Analysis, Monte Carlo and Key Risk Indicators

For risk analysts: build and challenge synthetic scenario,loss-model,indicator and reporting evidence with explicit scope and human decision authority.

Level
Intermediate
Study time
14 hours
Lessons
18 in 6 modules
Exam
20 questions

About this course

A planned14-hour intermediate supplied-output course for Python/statistics-literate risk professionals. Kestrel Ridge, a fictional US distributor, anchors a cumulative evidence record; a manufacturer, regional bank and insurer are separately scoped comparisons. Preserve useful quantitative notebooks, inspect computed outputs, repair input/evaluation defects and defend a continue/narrow/hold recommendation. Python execution and optional live AI deepen the work, while supplied fixtures/manual checks support the core route.

Primary-source refresh is dated November 21,2026 with access and applicability limits in the source ledger. No product ranking, legal/actuarial assurance or observed workplace benefit is claimed. The final optional practice feedback is fallible and advisory. Completion follows lessons, listed Decision Labs to any ending and the exam; it does not require an AI score or an actual independent reviewer engagement and does not certify workplace competence.

What you will be able to do

  • Analyze elicited event definitions,units and severity ranges;repair a seeded annual-loss simulation and document its12-of25-risk scope
  • Evaluate dependence-family and latent-correlation sensitivities,verify matrix/factorization conditions and preserve marginal samples without presenting an assumption as measured correlation
  • Evaluate cyber-control and recovery sensitivities with conditional/annual scope,insurance simplifications and actual disclosure/continuity authority kept separate
  • Analyze workshop,parameter and horizon-source records for unsourced claims,shared causes,changed evidence and overlap before proposing register updates
  • Evaluate mature rolling-origin KRI evidence and incident/provider extraction errors against explicit denominators and propose bounded human-review routes
  • Create a typed figure-trace record and supported board,audit-priority and insurer-stress narratives that preserve statistic,unit,period,scenario and authority
  • Evaluate numerical convergence,seeds and sensitivity separately from assumptions,predictive validity and model fitness for intended use
  • Create and revise a scoped validation recommendation with actual evidence,lineage,alternative strategy,owned gaps and a concrete independent-review request

Syllabus

6 modules · 18 lessons · about 14 hours

Module 1: The ERM Analyst's Risk Register: From 5x5 Scores to a Seeded Monte Carlo Loss Distribution

Replace the heat map with frequency and severity ranges owned by risk owners, have an LLM draft and review the simulation code, and aggregate risks with dependence.
3 lessons
  1. Why the 5x5 Heat Map Cannot Answer "How Much Could We Lose in a Bad Year?" and How to Elicit 10th/50th/90th Percentile Ranges
  2. The LLM as Coder and Reviewer: A Poisson-Lognormal Simulation in numpy with a Fixed Seed, a Loss-Exceedance Curve and Two Planted Bugs
  3. Aggregating with Dependence: Gaussian and t-Copulas, PSD Checks and Tail Sensitivity to the Correlation Assumption
  • Decision Lab · 10 minThe monthly row in the annual packChoose evidence, respond to changed sources and defend a scoped analytics recommendation.

Module 2: The Cyber-Risk and Resilience Analysts: Open FAIR for Ransomware and Recovery Time Against Impact Tolerance

Quantify a ransomware scenario as one input to the Item 106 disclosure and 8-K process, and simulate outage duration across dependencies using BIA data.
3 lessons
  1. Building an Open FAIR Ransomware Model: Scenario Statement, Sourced Factors and Before-and-After Control Sensitivity
  2. Item 106, Form 8-K and the Materiality Line: What the Model Informs, What Counsel Decides, and Red-Teaming Draft Disclosure Text
  3. Simulating Recovery Time Across Vendor, System and Site Dependencies Against an Impact Tolerance (NIST IR 8286D, SR 20-24)
  • Decision Lab · 10 minA control design is not a recovery resultChoose evidence, respond to changed sources and defend a scoped analytics recommendation.

Module 3: The Scenario Workshop Facilitator and the Emerging-Risk Analyst: Severe-but-Plausible Scenarios Without Invented Precedents

Use an LLM to widen and structure scenario workshops and horizon scanning while keeping every factual claim sourced and every scenario quantifiable.
3 lessons
  1. Running an LLM-Assisted Scenario Workshop: Templates, Candidate Scenarios and Countering Anchoring
  2. From Workshop Notes to Parameter Tables: Severe-but-Plausible Scenarios for Simulation and the ORSA
  3. A Horizon-Scanning Pipeline Mapped to the Register: Classification, Citations and an Automated Quote-Match Check
  • Decision Lab · 10 minThe owner changes the closure cardChoose evidence, respond to changed sources and defend a scoped analytics recommendation.

Module 4: The KRI Owner, the Op-Risk Analyst and the TPRM Analyst: Indicators and Extraction You Can Measure

Evaluate leading-indicator evidence, classify narratives against explicit denominators and inspect provider reports without treating extraction as assurance.
3 lessons
  1. Backtesting Candidate KRIs: Lagged Correlation, Rolling-Origin Holdouts and Thresholds Tied to Appetite
  2. Classifying Incident and Loss-Event Narratives into a Taxonomy: JSON Schema, Gold Set, Confusion Matrix and Confidence Routing
  3. Extracting SOC 2 Opinion Type, Exceptions and CUECs into Vendor Tiers and KRIs Under the Interagency TPRM Guidance
  • Decision Lab · 10 minThe quarter closes before outcomes matureChoose evidence, respond to changed sources and defend a scoped analytics recommendation.

Module 5: The Risk-Reporting Lead, the Audit Analytics Lead and the ORSA Analyst: Reporting Risk Without Inventing Numbers

Draft board packs, audit-plan risk scores and ORSA stress sections from computed outputs, with an automated checker that traces every figure.
3 lessons
  1. The Quarterly Board Risk Pack: Grounded Narrative and a Python Number-Traceability Checker
  2. Data-Driven Auditable-Entity Risk Scoring for the Audit Plan, with Backtesting and Logged CAE Overrides
  3. Drafting ORSA Stress-Test Sections from Capital-Model Outputs Under NAIC Model Act #505 and ASOP No. 56
  • Decision Lab · 10 minThe right digits in the wrong fieldChoose evidence, respond to changed sources and defend a scoped analytics recommendation.

Module 6: Capstone: The Risk-Model Owner Validates the Stack with an Independent Reviewer

Separate numerical repeatability from fitness for use, and assemble a reviewable validation recommendation with an independent-challenge request.
3 lessons
  1. Seeds, Convergence and Sensitivity: Testing Whether the 99th Percentile Is Stable
  2. Evaluating the LLM Components: Classification Accuracy, Citation Validity and Numeric-Trace Pass Rate
  3. Documentation, the Assumptions Log and the Independent Human Reviewer: Writing the Validation Report
  • Decision Lab · 10 minThe untouched test was used to choose the promptChoose evidence, respond to changed sources and defend a scoped analytics recommendation.
  • Capstone simulation · 25 minKestrel’s December validation committeeBuild a cumulative evidence matrix within a defined cost/day plan; respond to contaminated evaluation and stale scenario labels.

Exam and certificate

The exam unlocks once you have completed every lesson and played each Decision Lab to an ending. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 40 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.

Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.

The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by Lumivero, Microsoft, Anthropic, OpenAI, or Google. Product and company names are trademarks of their respective owners.

More in AI in Professional Services

All 19 courses in this track
pro-001 Free intro Available

AI in Professional Services: The One-Hour Briefing

A free, task-by-task hour on AI in law, accounting, tax, audit, recruiting and consulting. Each lesson follows one named role through one real job, such as verifying citations in a research memo, testing journal entries or checking a recruiting shortlist for adverse impact, and ends with a short exercise on synthetic data.

1 h · 9 lessons Free intro
pro-210 Intermediate Available

Building and Deploying AI in Legal Practice

Analyze legal AI outputs, current US authority, review uncertainty and confidentiality controls; carry eight worked artifacts through six module Decision Labs and one cumulative committee capstone.

25 h · 21 lessons See price at checkout