Extreme Risk in Portfolios Selection: Models Comparison and Analysis

Files

Supervisors: Lassance, Nathan
Faculty: Louvain School of Management
Degree label: Master [120] : ingénieur de gestion, à finalité spécialisée
Abstract: The purpose of this thesis is to investigate the different approaches to valuation and risk minimization for heavy losses in a portfolio. We can easily acknowledge that the return of a portfolio, composed of different assets, should be normally distributed, and generally follows a Gaussian curve. This means that the probability of having a return close to the average is high and that the probability of having greater returns or losses is typically low. The models studied in this thesis, called “Extreme Models” such as Value-at-Risk, Expected Shortfall and Cornish-Fisher Value-at-Risk, help investors and portfolios manager to calculate this probability of large returns or losses. We will first explain how these models work, what their aims are, demonstrate the benefits and drawbacks and how to minimize the risk of losses for each model. Then, we will compare the results in terms of performances, weights, backtesting and other metrics. Finally, we will present the results obtained, as well as the limitations of using such models. We will draw some conclusions about the performances of each model and see what the potential use of these models are. We will foresee what perspectives are available in the future and how they might impact the results we have found in this thesis.