ZhangLixin0714/fin-risk-models

What it solves

This project provides a library of modular models designed to detect financial risk, fraud, and regulatory non-compliance. It addresses the challenge of identifying complex patterns of financial misconduct, such as organized fraud rings, money laundering, and credit risk concentration, which are often hidden in large datasets.

How it works

The repository contains a collection of specialized models implemented as SQL logic or algorithms (sometimes in Jupyter Notebooks). Each model targets a specific risk scenario—such as flagging borrowers who share the same recipient account or detecting loans that are repaid immediately after a quarter-end to hide financial health (window dressing).

Who it’s for

This is intended for risk analysts, regulatory compliance officers, and financial engineers who need ready-to-use detection logic to monitor asset quality and identify fraudulent behavior in lending and corporate finance.

Highlights

• Fraud Detection: Identifies suspicious clusters, straw-buyer rings, and fund-pooling.
• Credit Risk Monitoring: Flags early delinquency, uncovered collateral, and abnormal provisioning changes.
• Regulatory Compliance: Detects conflicts of interest, such as employees acting on their own loan workflows.
• Diverse Scenarios: Covers a wide range of assets including SME loans, corporate trade bills, and mortgages.