| Abstract [eng] |
This thesis examines portfolio construction based on alternative risk measures using a universe of exchange-traded funds (ETFs) representing major global asset classes. The empirical analysis is conducted on daily data from 1 January 2008 to 31 December 2024 for ten ETFs spanning U.S. and international equities, emerging markets, small-cap equities, investment-grade and high-yield bonds, real estate, commodities, and gold. The objective is to compare how different definitions of risk affect portfolio weights and the resulting risk–return trade-off. Three risk measures are analysed: the standard deviation of returns, the Sharpe ratio, and maximum drawdown. Standard deviation is used to construct a minimum-variance portfolio, the Sharpe ratio is maximized to obtain a portfolio with the highest return per unit of volatility, and maximum drawdown is minimized to emphasize capital preservation. All optimized portfolios are compared to equal-weight benchmark. The methodology is implemented in Python, using the SLSQP algorithm for constrained nonlinear optimization (long-only, fully invested portfolios) and Monte Carlo simulations over a large set of randomly generated portfolios as a robustness check. |
