Black Litterman approach

[TTKDD]

Schweser book 2, Answer 13, pg 212

Dear all,

In the last para on the page it says ‘The one that does not require the analyst to start with estimated expected return is the Black Litterman approach.’

If I turn to page 171 for ‘Resampled efficient frontier (REF)’, under the second para it says ‘In response, Michaud developed a simulation approach utilising historical means, variances and covariances of assets which combined with capital market forecasts, assumes they are fair representation of their expectations’

So in my opinion the starting point of ‘Resampled efficient frontier (REF)’ is also not based on assumptions, but the answer to question 13 insists (by using word ‘ one ’).

Kindly clarify is I am missing anything.

Thank you

[S2000magician]

I agree that REF isn’t based on an analyst’s expected return.

So . . . CFA Institute blew it on this one. Color me surprised.

[TTKDD]

S2000magician,

I am trying to summarise

MVO: begins with estimation of expected return and covariance of assets

Surplus ALM: for asset sides same as MVO and inaddition estimation for the liability flows as well

In order to make the EF more stable (not volatile to inputs) other approaches like REF, BL approaches approaches have been developed.

To make the above EF more robust, MCS has been added to address the portfolip/surplus situation in the future.

So MCS can be complemented to all approaches.

Kindly confirm.

Thank you

[SWhip]

Black Litterman does not start with expected returns of the individual sectors. You reverse optimize to receive them later in the process and you can then adjust them.

Perhaps CFAI is trying to refer to the concept that you have to use reverse optimization first to uncover the expected returns, while in MVO, etc. you need expected returns at the beginning to be able to run the optimization process.

[cgottuso8190]

@SWhip

Black Litterman does not start with expected returns of the individual sectors. You reverse optimize to receive them later in the process and you can then adjust them.

Perhaps CFAI is trying to refer to the concept that you have to use reverse optimization first to uncover the expected returns, while in MVO, etc. you need expected returns at the beginning to be able to run the optimization process.

^This - the point is that BL is the only method where you don’t explicitly start with expected returns but you back into them (implied returns) based on historical standard deviations, correlations, and weights of a global index.