Portfolio Optimization: A Robust Equity Allocation Framework Incorporating Investor Views

Abstract

This thesis proposes a framework for capital allocation in institutional equity portfolios. An original simulation model is constructed by combining a 4-factor model based on principal components with an empirical copula that models the dependency structure between assets. The model generates skewed and leptokurtic returns that closely resemble empirical data, and it is used to evaluate the performance and robustness of several prominent portfolio optimization techniques, including mean-variance, mean-CVaR, risk budgeting, and maximum diversification. The results show that both the equal-risk-contribution and maximum diversification portfolios are robust to estimation uncertainty, with the latter achieving a higher Sharpe ratio and thus being preferable. Finally, the thesis demonstrates how investor views on any distributional characteristic can be reflected in a posterior distribution using entropy pooling. This method can be integrated in the portfolio optimization process or used to stress-test optimized portfolios under various views.