Credit Risk Optimization

Risk and portfolio management models arising in finance can be formulated and solved as optimization problems. Credit risk models are especially challenging for practical implementation due to the fact that the portfolio’s loss distribution is not known exactly. To solve such problems, special mathematical, algorithmic and implementation techniques are required. This internship project with Algorithmics Inc., a financial risk management software provider, plans to investigate solving credit risk optimization models using modern optimization algorithms. The intern will work specifically on solving large-scale linear and quadratic optimization problems where problem structure can be exploited to speed-up the solution time. Presence of integer variables in some of the optimization problems requires looking at their relaxations. The intern plans to develop algorithmic and software tools for credit risk analysis, benchmark existing techniques and design new variants specifically within Algorithmics framework, analyze and test these methodologies on real large-scale data. Special attention will be paid on multi-objective optimization for credit risk and portfolio management problems.

Faculty Supervisor:

Dr. Tamás Terlaky

Student:

Oleksandr Romanko

Partner:

Algorithmics Inc.

Discipline:

Computer science

Sector:

Finance, insurance and business

University:

McMaster University