In the question, a graph is presented with 2 frontiers. Frontier A lies above frontier B. The question goes on to say that Frontier B is the true frontier with perfect knowledge of the true return parameters. It then asks if Frontier A could be: i) the conventional mean variance efficient frontier ii) the resampled efficient frontier the answer to both is no. why is that though? if the conventional mean variance efficient frontier was created with overestimated returns, couldn’t it lie above the true frontier? also, why is the true frontier always above the resampled efficient frontier? thanks.
It cant be resampled effecient frontier as this methods produces more stable effecient frontiers (using average weights ) . This frontier will lie above the conventional but still below the true efficient frontier…as the resultsstill do have elements of estimation bias… and as u said correctly that the conventional cant lie above the true due to overestimation bias… thus the answer to both is no… correct me if i m wrong
I knew the answer was NO for both b/c you can’t lie on the true efficient frontier b/c you can’t be exactly 100% sure of the true returns and risk over a given time period. However, I dont think I would have been able to answer that question off the top of my head according to the answer they provided.
Here is my thinking…if B is the true efficient Frontier, then it is perfect. Therefore, you cannot have another efficient froniter (same assets) lie above a perfect efficient frontier
ws, that is only ex post. ex ante if you are building efficient frontier you can, so problem has to be talking about ex post
^Wouldn’t perfect knowledge be more important to ex ante? I guess we all have perfect knowledge on ex post fact.
We can all be monday morning quarter backs can’t we ???
ws, that is why i am confused about this problem. If estimate returns to be 1000% and run MVO, i am sure my frontier will be better then real efficient frontier
CSK…I am not 100% sure what you are confused about. Ex-ante or ex-post? For this problem, if your perfect knowledge is the “estimated” (if perfect knowledge, it WILL be) return is 1000%, and your ran MVO, your resulting efficient frontier is B (using example’s number)…no other “efficient frontier” can get above that. I am not sure if I have addressed your questions…
why can’t the estimated frontier be above the true knowledge one? You can get this by overestimating the returns? For example, my wrong estimated frontier has 100% of return for every % of risk. Clearly this incorrect frontier would lie above the true frontier.
I just went through this problem and came up with the same exact question. From what I understand, Curve B is an idealized Efficient frontier assuming perfect knowledge (not possible in real world). Curve A is an estimate and is prone to errors. Even the conventional mean-variance EF is prone to errors as it based on “Estimates” not perfect knowledge.
Billwest, You are looking to only one side of the coin, i.e returns. You need to look at risk and reward as this is an optimization problem. An efficient frontier created with true value of expected returns will always be the most optimum in terms of risk-reward ratio. So for a given level of risk, points on a true efficient frontier will always lie above points on any other less “true” efficient frontier. So I agree with ws.