What are the Pittfalls of traditional portfolio optimazation?
Extreme portfolio weights
Sensitivity of portfolio weights
Estimation errors for input values
Why is the information aggregation a pittfall in the traditional portfolio optimazation?
The Manager requires as inputs the expected returns and cthe variance-covariance information over the assets. Managers only have reliable information about the returns over a small group of assets.
Proper Variance and Covariance information are even harder to get.
Due to the correlation between assets, the return of an asset affects the return of other assets. Therefore the performance of a portfolio relies deeply on the quality of the input information
Why is the extreme weighting a pittfall in the traditional portfolio optimazation?
Not all weights are possible to implement in practice.
Why is the sensitivity weighting a pittfall in the traditional portfolio optimazation?
Small changes in the expected return have a strong impact on the portfolio weighting
State the mean varaince optimazation formula for the benchmark returns in the Black-Litterman approach
State the the solution of mean varaince optimazation formula for the benchmark returns in the Black-Litterman approach.
Which technique can we use if the portfolio weights are optimal?
In case that the weights are optimal, we can work backwards with the reverse optimazation. We solve the formula for the expected return
How does a Manager have to express his subjective views in the Black Litterman Approach?
Please describe the formula
P is number of assets*number of views
V is the vectore with our subjective views
e is the vector with error terms of the views
What are advantages of the Black-Litterman?
Portfolio weights are plausible and stable
Results are transparent
Portfolio contains subjective views
Estimation Errors are reduced
What are disadvantages of the Black-Litterman?
Almost no empirical testing
Uncertainty is difficult to implement
What are the results in the Bessler, Opfer and Wolff (2014) comparison of the “Out Of Sample BL” and the “Mean Variance Optmiazation”
BL has lower turnover -> Lower Transactions Costs
BL has a higher Sharpe Ratio
BL has a lower volatility
BL has a higher average numer of assets -> Diversification
What is the understanding of the following quote:
Assets with extreme returns tend to be most affected by estimation errors, therefore mean-standard deviation optimizers are estimation-error maximizers (Michaud, 1989).
In the Mean-Variance we want to increase or expected return. But increasing expecting return also means increasing our standard deviation -> maximizing estimation error
Assume that in the subjective view we have high correlation with with different assests. What will happen as a conclusion, if you expect a subjective higher return for one asset?
If the correlation is high and we assume a higher expected return, it will affect the other assets in a positive or negative way.
Why is the estimation errors for input values a pittfall in the traditional portfolio optimazation?
Input values can be only measured with strong uncertainty
Mean-Standarddeviation are error maximizer