Paper: arXiv 2412.20225

Authors: Abdollah Rida

Abstract

Credit Scoring is one of the problems banks and financial institutions have to solve on a daily basis. If the state-of-the-art research in Machine and Deep Learning for finance has reached interesting results about Credit Scoring models, usage of such models in a heavily regulated context such as the one in banks has never been done so far. Our work is thus a tentative to challenge the current regulatory status-quo and introduce new BASEL 2 and 3 compliant techniques, while still answering the Federal Reserve Bank and the European Central Bank requirements. With the help of Gradient Boosting Machines (mainly XGBoost) we challenge an actual model used by BANK A for scoring through the door Auto Loan applicants. We prove that the usage of such algorithms for Credit Scoring models drastically improves performance and default capture rate. Furthermore, we leverage the power of Shapley Values to prove that these relatively simple models are not as black-box as the current regulatory system thinks they are, and we attempt to explain the model outputs and Credit Scores within the BANK A Model Design and Validation framework

Complexity vs Empirical Score

  • Math Complexity: 5.5/10
  • Empirical Rigor: 7.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper provides a highly practical and regulatory-compliant framework for using XGBoost in bank credit scoring, validated on real auto loan data from an actual institution. It balances formal statistical learning proofs with extensive empirical diagnostics like out-of-time testing, swap set analysis, and SHAP-based explainability.

Research Flowchart

  flowchart TD
    A["Research Goal: Create Basel II/III Compliant ML Credit Scoring Model"] --> B["Data Input: BANK A Auto Loan Applicant Data"]
    B --> C["Model Development: Train XGBoost Gradient Boosting Models"]
    C --> D["Explainability Process: Apply Shapley Values (SHAP)"]
    D --> E["Regulatory Validation: Evaluate via BANK A Design & Validation Framework"]
    E --> F["Outcomes: Superior Default Capture Rate & Regulatory Compliance Achieved"]

Cited References (19)

  1. Convex Optimization: Algorithms and Complexity (Foundations and Trends in Machine Learning) (2015) — link
  2. Introduction to High-Dimensional Statistics (Chapman & Hall/CRC Monographs on Statistics and Applied Probability) (2014) — link
  3. Hands-On Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems (2017) — link
  4. Foundations of Machine Learning (Adaptive Computation and Machine Learning series) (2012) — link
  5. The Elements of Statistical Learning (2001)
  6. XGBoost: A Scalable Tree Boosting System (2016) — link
  7. A unified approach to interpreting model predictions (2017) — link
  8. SMOTE: Synthetic Minority Over-sampling Technique (2011) — link
and 11 more
  1. Monotonic solutions of cooperative games (1985) — link
  2. Long Short-Term Memory (1997) — link
  3. https://github.com/ckmarkoh/GAN-tensorflow — link
  4. A deep learning approach for credit scoring using credit default swaps (2017) — link
  5. Consumer Credit - Risk Models Via Machine - Learning Algorithms (2010) — link
  6. Credit Scoring using Machine Learning Techniques (2017) — link
  7. SR 11-7: Guidance on Model Risk Management — link
  8. Federal Reserve Supervisory Assessment of Capital Planning and Positions for Large and Noncomplex Firms — link
  9. OCC 2000–16 — link
  10. OCC BULLETIN 1997-24 — link
  11. A granular computing-based approach to credit scoring modeling (2013) — link

0 of 19 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.