Learn / Playbooks / Model validation
Model validation
A model is ready for sign-off, and the model-risk team will review every test. Prove a model works, is stable, and can be explained, before anyone relies on it.
7 steps · 8 tools (2 free) · for risk analysts at banks and NBFCs
Tools you will use
This playbook is free to read. The prompt for each step is inside its tool. See the Pro tools.
Check these first
Each tool asks the same questions and gives you a prompt to build whatever is missing.
The steps, in order
Each step names the tool to open.
Set up
Score, outcome and date fields.
Test
Does it rank good and bad apart?
Do predicted rates match actual rates?
Does it behave the same across time?
Test on a period the model never saw.
Explain and document
Reasons for a score, and checks across groups.
What the validation team asks for first.
Check before you send
- Discrimination holds up on the out-of-time sample, not only on the build sample.
- Calibration is checked by score band, not only overall.
- Every test result is saved with the date and the data used.
- Findings are written with a severity and an owner.
Warning signs
What you hand over
A validation report with test results, findings and documentation for model-risk review.
In your profile
Once you are signed in, your profile shows which steps you can run with the tools you own, and which tools you still need.