Creating a Contemporary Risk Management System Using Python
summary
Claims a working underwriting and portfolio system built on the PyData stack that uses five years of nightly invoice feeds to forecast 30/60/90 day cash flows, detect invoice discontinuation and bankruptcy, and set limits and rates; achieves roughly a 40‑day duration with ~6–7% yield. Pipeline runs nightly with Luigi, pandas and statsmodels, calls R models via rpy2, exposes results in a simple web UI built with Spyre, and uses XGBoost for fraud and discontinuation classifiers.
AI-written summary · about
chapters (10)
- 0:00C2FO and risk management
- 5:21Risk management continuum
- 7:12Default risk with invoice forecasting
- 12:14AR discontinuation in suppliers
- 16:24Bankruptcy with fuzzy matching
- 22:04Fraud via invoice and call analysis
- 25:33Portfolio management and WFC scoring
- 37:17WFC risk management web app demo
- 40:07Key takeaways and open source
- 43:19Q&A on look-ahead bias and security
related talks
36mA Data Science Approach to Systemic Risk
31mBuilding Data-Driven Client Relationship Management in Banking with Python
23mAn Open Source Architecture for Modernizing Market Risk Management
42mRobust Automated Forecasting in Python and R
39mMachine Learning in Financial Credit Risk Assessment
28mBeyond the Balance Sheet: Transactional Data as a Window to Company Health
29mMonte Carlo Business Case Analysis with pandas
40mScaling Python - Bank Edition