Portfolio Mean-Variance and Conditional-Value-At-Risk Optimization Framework: A Case in Vietnam

Abstract

Investors have long been grounded with the notion of diversification and portfolio holding, yet are still, puzzled over the formula for a “perfect” portfolio, which could resist the market’s downfalls and thrive in favorable conditions. The conundrum is especially magnified in Vietnam, where the theoretical foundation, market performance evidence, and portfolio management’s empirical findings are in their infancy. Thus, the thesis aims at contributing to the body of literature with evidence of the two most ubiquitous portfolio construction frameworks’ performance, namely Mean-Variance and Mean-CVaR, respectively the most classic and state-of-the-art models. As a result, investors might be inspired in their decision on the portfolio formulation framework in use.

After developing the literature review, the daily returns of 14 constituents of the VN30 index throughout 10 years retrieved are employed to generate the mean-variance and mean-CVaR efficient frontiers, which serves as an in-sample performance assessment. The out-of-sample aspect is testified by putting optimal portfolios generated on 5-year periods’ returns in the context of the subsequent year with weight structure diversity and returns are the key evaluation criteria. Besides, equally-weighted portfolios are positioned in proximity to investigate the notion of optimization. effects.

In the in-sample context, the dominance of the Mean-CVaR framework in both efficient frontier and diversification could be witnessed while the out-of-sample investigations imply rarely significant preference. Overall, the 1/N rule will work the best with risk-averse investors as the extremality in weight structure of optimized portfolios, but those optimization methods are those that could facilitate investors to attain more impressive returns.