Portföljoptimering som alternativ till indexfonder
Hur skulle en fond konstruerad enligt portföljoptimeringsmodeller utvecklas i jämförelse med index?
This paper investigates the possibilities to construct automatized portfolios based on optimizations strategies that could outperform comparable indexes. The study is based on time-series of Swedish stocks dating from 1986 to 2006. In the research four different portfolio optimization techniques were studied. These were: the Classical Markowitz Approach, Mean-Absolute-Deviation, Minimum-Regret and Conditional Value-at-Risk of which the three latter are based on generated scenarios. The behaviour of these models was studied for different choices of parameters such as backward time-horizon and targeted average return. The constructed portfolios were then rebalanced on equidistributed occasions in time. The results indicate that portfolio optimization models could indeed generate a better return than index. Particularly, the Minimum-regret technique on average outperforms index significantly, whereas it does not rebalance the assets in the portfolio as often as the other scenario-based models. Markowitz shows good results when rebalanced quite infrequently.