3 Presentations

• 
  03:30 PM - 03:55 PM

  A Simulation and Regression Dynamic Programming Approach to Portfolio Optimization

  • Michel Denault, presenter, GERAD - HEC Montréal
  • Jean-Guy Simonato, HEC Montréal
  • Erick Delage, GERAD, HEC Montréal

  We consider portfolio optimization with a dynamic programming approach that relies on simulation and regression. Simulations are used to model the stocks returns uncertainty. Regressions are used to approximate the value function. Wealth, an endogenous state variable, requires special care. We provide some preliminary numerical results.

• 
  03:55 PM - 04:20 PM

  Performance Evaluation of Bankruptcy Prediction Models: A Multidimensional Framework

  • Seyed Mohammad M. Mousavi, presenter, University of Edinburgh
  • Jamal Ouenniche, University of Edinburgh

  Although most studies on the modelling and analysis of corporate bankruptcy data use several performance criteria and measures to assess the performance of competing prediction models, the assessment methodology used so far remains unidimensional in nature, which leads to reporting conflicting results. In this research, we overcome this methodological issue by proposing a multidimensional assessment framework and report on numerical results on UK data.

• 
  04:20 PM - 04:45 PM

  Developing an Hybrid Approach for Market Deployment Roadmapping

  • Salman Kimiagari, presenter, TRU
  • Benoît Montreuil, Université Laval

  The paper focuses on the development of market deployment roadmaps. It proposes an hybrid approach. Geo-markets are clustered using self-organizing maps, exploiting their multi-criteria, spatial and visualization capabilities model. Within the generated clusters, an optimization model is used for deciding on the selection and timing of geo-markets deployment. The paper demonstrates the application of the approach to a business design case.