Raphael LimaPublications
← Back to the colonnade

II. Experience · 08

08

Data Scientist

Banco Itaú

Apply analytics to credit risk, IFRS 9, collections and fraud.

MAR 2014 — MAY 2015 / Banking

Role

Credit-risk modeling, Individual and corporate clients, Banking

Developed application and behavior predictive models, with data integration and data modeling, datamart maintenance and model monitoring. Partnered with Credit Risk, Model Validation, Finance, Regulatory Capital, Data and Technology teams.

Approach

Estimated the likelihood of default for individual and corporate clients with logistic regression and decision trees, and implemented and improved retail and wholesale credit-risk models.

Impact

  • Reduced the time required for model validation and back-testing by 30% with analytical automation tools.
  • Contributed to the bank's IFRS 9 readiness and implementation program.
  • Worked with PD, LGD and EAD components to support impairment estimation.

Lesson

Lasting credit results come from policy and monitoring, not from a single model.

Linked use cases

  • Banco Itaú

    Integrated credit-risk engine

    2013–2014

    ChallengeProvisioning, collections and pricing ran on disconnected risk truths: spreadsheets with manual overlays, collections only after delinquency and product price tables, causing avoidable losses and mispriced margin.

    SolutionA single PD, LGD and EAD engine, with independent validation and an audit trail, feeding expected loss, preventive collections, limit management and RAROC-based pricing.

References

← Back to the colonnade