Embed
Email

Asset Management

Document Sample

Shared by: ewghwehws
Categories
Tags
Stats
views:
4
posted:
1/23/2012
language:
pages:
14
Asset Management



Lecture 9

Outline for today



 Black-litterman model and sensitivity in

confidence

 Treynor-Black vs Black-Litterman

 Value of active management

The Black-Litterman Model



 Step 1: Estimate the covariance matrix from

historical data

 Step 2: Determine a baseline forecast

 Step 3: Integrating the manager’s private

views

 Step 4: Developing revised (posterior)

expectations

 Step 5: Apply portfolio optimization

Sensitivity in confidence level







Confidence measured by standard deviation of view Q

Possible SD 0 0.0100 0.0173 0.0300 0.0600

Variance 0 0.0015 0.0003 0.0009 0.0036

E(RB|P) 0.0190 0.0148 0.0164 0.0152 0.0143

E(RS|P) 0.0140 0.0598 0.0424 0.0556 0.0643

Figure 27.5 Sensitivity of Black-Litterman Portfolio

Performance to Confidence Level (view is correct)

Treynor-Black vs Black-Litterman



TB BL



Maximization identical

The BL Model as Icing on the TB Cake



 Suppose that you have two portfolios—one

for the US and one for Europe

 The model would be run as two separate divisions

 Each division would compile values of alpha

relative to their own passive portfolio

 Portfolios need to be optimized separately

 Relative performance of the two markets can be

expected to add information to the independent

macro forecasts for the two economies

The BL Model as Icing on the TB Cake



 Use BL to include forecasts from comparative

economic and international finance analyses

 Replace TB alpha with BL views

 Example: assume only one stock in the active

portfolio

 Alpha, beta, E(Rm), var(Rm), var(e)

The BL Model as Icing on the TB Cake



 Use BL to include forecasts from comparative

economic and international finance analyses

 Input list for BL model

R  [ E ( RM ), E ( R A )   A E ( RM )]

A

P  (0,1  )

 A E ( RM )

PR'  Q     A  

QE  0

D A

 ( )  Var( forecasting error )

2 u (t )  a0  a1a f (t )   (t )

 2 ( D)   2 ( )   2 (e) SCL : R(t )  a    RM (t )  e(t ), t  T

The BL Model as Icing on the TB Cake



 Use BL to include forecasts from comparative

economic and international finance analyses

 Calculate the conditional expected return will give

you the same results as from the TB model:

 The realized abnormal return of time T

SCL : R(t )  a    RM (t )  e(t ), t  T

u (t )  R(t )    RM (t )  a(t )  e(t )

 The precision of record, t
u (t )  a0  a1a f (t )   (t )

 Adjust a (T )  a0  a1a f (T )

The BL Model as Icing on the TB Cake



 The BL model could be viewed as a

generalization of the TB model

 Differences?

Treynor-Black vs Black-Litterman



TB BL



Maximization identical



input Individual Views of relative

security analysis performance

target Security analysis Asset allocation

with adjustment where relative

of forecasts performance is

relevant.

Value of Active Management



 Model for estimation of potential fees

 Kane, Marcus, and Trippi (JPM, 1999)

 The percentage fee, f, that investors would be willing

to pay for active services

f  (S P  S M ) / 2 A

2 2





 Source of the power of the active portfolio

ai

 the squared information ratios

2  (ei )

 a 

S P  S M  i 1  i 

2 2 n



 (ei )  Remember f is in addition to

2 what an index fund would

1 n  ai 

f    

2 A i 1  (ei ) 

charge.

Concluding Remarks



 The gap between theory and practice has

been narrowing in recent years



 TB model is sensitive to large alpha values



 BL model relies on the “confidence” level

which is often ambiguous.



Related docs
Other docs by ewghwehws
By registering with docstoc.com you agree to our
privacy policy

You are almost ready to download!

You are almost ready to download!