R 66i states 3 ways to address the instability of the minimum variance frontier: 1. constrain portfolio weights (no short sales so all portfolio weights are +) 2. employ forecasting models that do better than just by using historicals 3. avoid rebaliancing until there is a significant change in the efficient frontier Can someone help me understand what exactly they mean with point 1 and how it actually deals with the issue?
you minimum variance portfolio has the lowest variace which depends on correlation between assets. If you have short sales you pretty much reverse the sign of correlations, which in turn would change variance. If you have 2 assets with correlation of -1. 50/50 will have zero variance (and minimum variance). But if you short sell one asset and go long with another your correlation is no +1 and it’s no longer zero variance.
btw are you russian?
Thank you, CFA=NOLIFE! I also checked the curriculum again, and it seems to focus on the optimization process: the model attempting to maximally exploit the differences among assets, which can result in what they call “over-fitting the data”. If no limits are placed on short sales, then you can end up with portfolios with huge short positions. And yes, I am Russian!
Cool me too.
Just as a fair warning…the entire PM section on last year’s exam was devoted to this topic. NOBODY knew what the hell was going on. I don’t know if that means they will repeat it and put it on there again, but I would certainly know it just in case.
haha, his name gives it away. count adavydov and I in as well. udachi!