Articles
How to calculate loan loss provisions from roll rates
Build a transition matrix from monthly loan status data, test its default probabilities against actual outcomes, then calculate a provision for each delinquency bucket.
Which South African affordability rule applies to SME loans
Route each SME application by borrower type and agreement before calculating affordability. Then apply Reg 23A where it reaches the borrower, or use your own business assessment for an in-scope juristic borrower.
Breached password check: what to do at sign-up and login
Decide what the sign-up password rule should be and what a login does when the password turns out to have leaked.
Fraud checks for night loan applications need history
See whether application hour adds information on a consumer loan book, then build a referral rule from the time patterns that hold up inside score groups.
Fake employer loan applications and when concentration warrants review
Learn what employer-concentration checks can show, where a type-and-region proxy fails, and what to capture on the application before referring a case for review.
When early loan repayment merits a fraud referral
See which early repayments predicted payment difficulty on a consumer lender’s book, and how to test a referral rule before using it.
Turn a filed homeowners guideline into an insurance rules engine
Learn how to classify a filed homeowners guideline, identify what the quote form and filing leave undecided, and test the rules before they go live.
Application fraud checks for deciding which applications to refer
Learn how to test application mismatches against your own book and decide which checks should send an application to human review.
Application fraud after refusal and which cases need review
Decide which approvals after refusal deserve a client-level flag and which store patterns warrant review. The evidence separates useful client flags from store rankings that reflect product mix.
Changed documents in credit cases and what should rerun
Use the changed input to see which tasks must run again and which reviews must reopen. Then set freshness on the decision that needs current data.