Skip to content
Edition NigeriaGlobal
Country: Mexico

You are in the Mexico edition. Type your country to switch.

The Americas

Set up your profile

You are learning in the Mexico edition, with examples in pesos and from Mexico.

AI and machine learning applied to financial data · Lesson 6 of 10

Model risk management and explainability

Model risk is the chance of loss or harm from a model that is wrong or used wrongly. Good practice inventories models, validates them independently, monitors them in use and explains their outputs. Explanation tools show which inputs pushed a prediction up or down.

Advanced9 min read5 steps

This lesson is complete in writing. A video for it is on its way.

What you will learn

  • By the end you can define model risk and name its main sources.
  • You can describe the lifecycle of model risk management: inventory, validation, approval, monitoring and retirement.
  • You can read a simple explanation of one prediction, with each input's plus or minus contribution, and check that it adds up.
  • You can complete and critique a basic model card.

Key terms in this lesson

Model risk
The risk of loss or harm from a model that is wrong, out of date or used for something it was not designed for.
SR 11-7
Supervisory guidance on model risk management issued by US banking regulators in 2011, which set expectations for model development, independent validation and governance.
Explainability
How well people can understand why a model produced a given output, in terms of which inputs mattered and in which direction.
SHAP value
A widely used way to explain one prediction by sharing out the difference between that prediction and the model's average across the inputs, so each input gets a plus or minus contribution.
Model card
A short document that records what a model is for, the data it learned from, how it was tested, where it performs poorly, who owns it and when it was last reviewed.

Step 1 of 5

What model risk is

Opening this step…

Step 2 of 5

The model risk management lifecycle

This step opens when you finish the one above.

Step 3 of 5

Explainability: why did the model say that?

This step opens when you finish the one above.

Step 4 of 5

What explanations can and cannot tell you

This step opens when you finish the one above.

Step 5 of 5

Exercise: completing a model card

This step opens when you finish the one above.

Key takeaways

The key takeaways open when you finish every step.

Lesson quiz

5 questions. You need 85% to pass.

The quiz opens when you finish every step.

Read first

Stockertim Learn is education. It explains how investing works and never recommends buying or selling anything.

Up next

AI & securityBias and fairness in credit and …Read nowBias and fairness in credit and insurance modelsAI & security · Advanced · 15 minAI & securityGenerative AI in research and ne…Read nowGenerative AI in research and newsroomsAI & security · Advanced · 15 minAI & securityAI governance and regulation in …Read nowAI governance and regulation in financeAI & security · Advanced · 15 minAI & securityAI in trading and market surveil…Read nowAI in trading and market surveillance, and the systemic risksAI & security · Advanced · 15 minBonds & fixed incomeBonos M, Federal Treasury Certif…Read nowBonos M, Federal Treasury Certificates and United Mexican States (UMS) international bonds comparedBonds & fixed income · Beginner · 15 minProjects, consulting & analysisProject risk and stakeholder man…Read nowProject risk and stakeholder managementProjects, consulting & analysis · Intermediate · 15 minValuation & company analysisThe auditor's report and governa…Read nowThe auditor's report and governanceValuation & company analysis · Advanced · 15 minAI & securityLeakage, look-ahead bias and wal…Read nowLeakage, look-ahead bias and walk-forward validationAI & security · Advanced · 15 minAI & securityWhat AI and large language model…Read nowWhat AI and large language models are, and are notAI & security · Beginner · 15 minAI & securityRules and accountability: why an…Read nowRules and accountability: why an AI answer is not adviceAI & security · Beginner · 15 minAI & securityResearching investments with AI …Read nowResearching investments with AI tools safelyAI & security · Beginner · 15 min
English