ASSET ALLOCATION: PERFORMANCE OF MEAN-VARIANCE STRATEGY FOR SMALL SAMPLE OF CANADIAN ASSETS

Анотація

This project estimates performance of two out of sample Mean-Variance Strategy portfolios. Weights are rebalanced monthly. Five years rolling window portfolio uses stock return information of 5 preceding years (information set). Similarly recursive portfolio uses all information of preceding years to construct weights. Expected returns and covariance are assumed to be equal to mean of these parameters over available information set. Performance of these portfolios is compared with equal initial weights and every period equal weight portfolios. In order to do so we employed a data from November 1992 to November 2013 for 10 Canadian stocks and Treasury Bills. Ex post Sharpe ratios and cumulative returns indicated that Mean-Variance out of sample portfolios delivered lower results comparable to naïve portfolio strategies over last 15 years. We explain it by low correlation between expected and realized returns and hence by investing in wrong stocks in a wrong time. And we provide numerical and graphical evidence.