You are on page 1of 22

DESIGNING PORTFOLIO

Returns

MINIMUM RISK PORTFOLIO 2 Risky Securities


X2

X1

rp = f 1 r1 + ( 1 - f 1 )r2
Risk

p = f 1 1 + (1 - f 1 )2 2 + 2 1 2 f 1 (1 - f 1 )
2 2 2 2

Minimum Variance Portfolio : p / f 1 = 0 - Gives f 1 = 2 2 2 1 2 1 + 2 - 2 1 2


2

UTILITY FUNCTION

Standard deviation as measure of risk considers positive as well as negative deviations from expected gain. Are positive deviations risk?? Use semi variance; for symmetrical distribution 2xsemi variance=variance Mean (1st Moment),Variance (2nd Moment) completely describe a distribution. Utility function is quadratic; variance provides estimation of risk. Distant values are magnified in variance.

UTILITY OF MONEY

3rd Moment is measure of skewness


SKEWED DISTRIBUTIONS

RISK AVERSE

More likely small losses Less likely large gains

Less likely large losses More likely small gains

RISK LOVER

Even Moments measure uncertainty Odd Moments measure skewness Utility Value = b1M1- b2M22 + b3M33 - b4M44

UTILITY OF MONEY

Paul Samuelsons Fundamental Approximation Theorem of Portfolio Analysis;


The importance of moments more than variance is much smaller Variance is as important as mean for the investor welfare. Mean variance will sufficiently describe utility function. Compactness of stock returns; implying controllability of returns. Returns can be controlled by shortening the holding period. Uncertainty declines as holding period shortens. Transaction costs limit frequent rebalancing of portfolios Mean Variance Analysis will not apply to unusual events (takeovers, wars, natural calamities etc.)

Assumption: Distribution are compact

ASSUMPTION OF NORMAL DISTRIBUTION

Stock returns is NORMAL distribution and are described fully by its mean and variance; contain all information. The maximum loss can not exceed the purchase value; I.e loss can not be more than 100% but Normal distribution assigns probabilities to losses > 100%. Empirical studies reveal that even if returns from individual stock may not be normal distribution, the portfolio returns could be. (Portfolio returns of 32 stocks closely resembles and of 128 stocks is almost identical with Normal Distribution)

ASSUMPTION OF NORMAL DISTRIBUTION

Alternatively continuously compounded return is normally distributed i.e. r = eat-1, gives least value of r as 100%. In this case distribution of a will be lognormal. For short holding period r = at, and for lognormal distribution of a, r will approximate NORMAL Distribution. Mean and Standard Deviation of returns of short holding period are proportional to the mean and variance of the continuously compounded return and the time interval.

OPTIMISING PORTFOLIO OF 2 RISKY ASSETS


Minimisation of risk is not same optimisation of portfolio for all; though for some it can be. There is a trade off between risk and return. This trade off depends upon individuals characteristics; What risk one is willing to assume for what return. Utility U is a function of Increasing Return, Decreasing Risk, and Individuals Risk Profile. U = rp - 0.005 * A * p2 Where rp = f1r1 + f2r2 , f2 = 1 f1, and p2 = f1212 + f2222 + 2f1f212 Optimisation would mean maximisation of Utility Function U.

OPTIMISING 2 RISKY ASSETS Graphical View


Returns A=4(Risk Averse)

Increasing Utility A=2 (Risk Lover)

Increasing Utility Higher and Higher Increasing Risk Aversion Steeper and Steeper Risk

OPTIMISING PORTFOLIO OF 2 RISKY ASSETS


U = rp - 0.005 * A * p2 Where rp = f1r1 + f2r2 , f2 = 1 f1, and p2 = f1212 + f2222 + 2f1f212 U = f1r1+ f2r2 0.005A(f1212+f2222+2f1f212) Maximising U; putting dU/df1 = 0

r1 - r2 + 0.01 A( 2 - 1 2 ) f1 = 2 2 0.01A( 1 + 2 - 2 1 2 )
2

COMBINING RISKY AND RISK FREE ASSET


Return r1

rp rf p

U = rp - 0.005A p 2 rp = f 1 r1 - (1 - f 1 )r f p 2 = f12 12
Risk

dU = r f + f 1 ( r1 - r f ) - 0.005Af 1 2 1 2 = 0 df 1 Proportion in Risky Asset f 1 = r1 - r f 0.01A 1 2

OPTIMAL PORTFOLIO WITH RISK FREE ASSET


Reward to Variability Ratio Cp = ra-rf/a For Optimum Maximise (Maximum Slope Line) Cp = rp-rf/ a subject to f1 + f2 = 1

f1 =

( r2 - rf )

2 1

( r1 - rf ) 2 - (r2 - rf ) 1 2 + ( r1 - rf ) 2 - (r1 + r2 - 2rf ) 1 2


2

PORTFOLIO DESIGN
Sharpes Portfolio Optimisation: Single Index Model: Assumes that a single parameter instead of several considerations, will determine the desirability of security to be included in the optimum portfolio. Step I: Choose One Parameter: Excess Return to Beta Ratio (ERi) (ri rf)

SHARPE MODEL
Step II: Rank the stocks by Excess Return to Beta Ratio with highest one to be most desirable stock to be included in the portfolio Step III: Determine the cut-off rate C* such that all stocks with Excess Return to Beta Ratio greater than C* are included in the portfolio and those lower than C* are excluded. Accept in the portfolio if ERi > C* Reject if < C*

SHARPE MODEL Contd.

The value of C* is dependent upon the characteristics of the securities included in the portfolio. If any security is to be in the optimum portfolio the parameter Ci is to be calculated. As securities are listed in order of desirability to be included in the portfolio the process must begin from the top of the table of securities. As long as the parameter Ci remains in excess of C* the security will be included in the portfolio.

SHARPE MODEL Contd.

Determining the Cut off n (r rf ) i i


2 m 1 2 ei

1+

2 m

i2
2 ei

Where

m2

not associated with market i.e. Unsystematic Risk of the Stock i. i = Beta of Stock i. = Variance of the Market Index

ei2 = Variance of return on Stock i

SHARPE MODEL Contd.


Determining the proportion in each stock; the amount of money to be invested in each of the stock is determined as follows; Determine Zi for each stock and Fix the proportion in the stock i as Xi=Zi/Zi
Zi =

i i
2

i r f r C * i

SHARPE MODEL Contd.


Revising Portfolio As new securities come in there may be a need to change the portfolio to remain optimum. The cut-off rate C* will determine whether or not the security be included in the portfolio. Only those securities with Excess Return to Beta Ratio greater than C* can be included in the new portfolio.

CHARACTERISTIC LINE & SHARPE MODEL

Requirement of data and processing capabilities are extremely demanding. For an N security case total of N(N+3)/2 parameters are required (N each of expected return and standard deviations and N(N-1)/2 of the covariances). Fluctuations in the value of one security relative to another does not depend upon the characteristic of two securities alone (Covariance) but can aptly be stated in terms of relationship with an index (Beta).

CHARACTERISTIC LINE & SHARPE MODEL


Return ri = i + i*rm + ei Risk Var(ri)= Var (i)+Var (.rm)+Var (ei) Total Risk = Systematic Risk + Unsystematic Risk = 2m2 + ei2 Coefficient of Correlation i = Cov(i,m)/im Coefficient of Determination 2 = Measure of Systematic Risk = % of risk that can be explained by market.

PORTFOLIO RISK

The risk that is diversified away is the Unsystematic Risk of the stocks. Non Diversifiable risk is SYSTEMATIC Risk; The risk that is common to all stocks, external to the firm, and is uncontrollable. Therefore while devising portfolio the Systematic risk of the share as measured by its Beta assumes greater importance.

PORTFOLIO RISK

Portfolio Variance is given by

= ( f i i ) * + f e
2 p 2 i =1 2 m i =1

2 2 i i

In a portfolio the Unsystematic Risk is evened out and what remains is the Systematic Risk. Therefore, portfolio risk can be measured by Beta of each of the security in the portfolio. Risk of the portfolio is Weighted Average of the Betas of the securities forming it.

You might also like