MSD502

Financial Analytics

Preeti Roy

Assistant Professor
IIT(ISM) Dhanbad

[email protected]

July 2024 - Nov 2024

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 1 / 69
The required reference books:

• 3rd edition by Say, R.S.

∗ Analysis of Financial Time Series

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 2 / 69
 Volatility Modelling
Unit 2

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 3 / 69
Why Volatility?
• Recent econometric techniques are used to model the attitude of the
investors not towards expected returns but towards risk.
• Require models that are capable of dealing with the volatility (variance)
of the series.
• Financial time series exhibit intermittency & volatility clustering.

• Expected value of the magnitude of the disturbance terms may be greater

at certain periods compared with others.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 4 / 69
Stylized facts favoring volatility
• Trend in the data
• High level of persistence
• Volatility not constant over time
• Series may have a random walk with drift component
• Series share co-movements
• Skewed & High Kurtosis

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 5 / 69
S& P BSE 30: Daily Stock Index, Returns and
Squared Returns

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 6 / 69
Heteroscedasticity over time

• These graphs suggest heteroskedasticity over time.

∗ Time-varying volatility of returns

∗ Volatility clustering

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 7 / 69
Volatility estimation: Types

• Realized Volatility - Daily variance estimator & Realized volatility esti-

mators
• Range-Based estimators
• Implied Volatility
• Conditional Volatility Measure

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 8 / 69
Realized Volatility - Daily variance estimator
• One can use daily variance in the month to calculate variance for the
month, called realized volatility.
• Let rtm be the monthly log return in month t
• Let rt,i be the daily log return on day i in month t .
• Suppose that daily returns are serially uncorrelated and the daily vari-
ance is constant.
• Then
n
rtm = ∑ rt,i
i =1

var [rtm ]
 
= n var rt,i

2
  ∑n rt,i − r¯t
Var rt,i = i =1
n−1
where r¯t is the mean of the daily returns
• The estimated monthly variance is thus

2
∑n rt,i − r¯t
(σtm )2 = n i =1
n−1
Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 9 / 69
Realized Volatility - realized variance estimator

• Summation of all the squared returns over a certain period (assuming

the number of trading days in a month), rather than calculating the
squared daily returns.
n
RealVart = ∑ rt,i2
i =1

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 10 / 69
Garman-Klass estimator of daily variance
• This method is uses the high, low, opening, and closing prices to
estimate daily variance
• It assumes that the price follows a random walk
• Assume that the ct be the logarithm of the closing price so

rt = ct − ct−1
h i
• Conventional estimator is based on closing price σt2 = E (ct − ct−1 )2
• Using only closing price, High Ht , low Lt , and open Ot also often are
available
• Can estimate daily variance of price (not log price) from
2 2
+0.386(ct −ot )2
2
σGK = 0.12 (Ot −C
f
t)
+ 0.88 0.5(ht −lt ) 1−f

where f is the fraction of the day that the market is closed.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 11 / 69
Range Based Estimators

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 12 / 69
Implied Volatility

• Measures the market expectations of near term volatility which con-

notes the rate and magnitude of changes in prices and a widely recog-
nized proxy for risk.
• For this the best bid-ask quotes of near and next-month Nifty options
contracts which are traded on the F&O segment of NSE are used.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 13 / 69
ARCH Process
• A stochastic process is called ARCH (Autoregressive Conditional
Heteroscedasticity) if its time-varying conditional variance is
heteroscedastic with autoregression:
rt = α0 + β1 rt−1 + εt , εt ∼ N 0, εt2


(1)

2
σt2 = α0 + α1 εt− 2
1 . . . . + αq εt−q (2)
• Equation (1) is the mean equation where regressors can be generally
added to the right hand side alongside εt .
• Equation (2) is the variance equation, which is an ARCH(q) process
where auto regression is its squared residuals has an order of q, or
has q lags..

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 14 / 69
GARCH Process
• A stochastic process is called GARCH (Generalized Autoregressive
Conditional Heteroscedasticity) if its time-varying conditional variance
is heteroscedastic with both autoregressive and moving average:
rt = εt , N 0, εt2


(3)
• σt2 = α0 + α1 εt−
2 2
1 . . . .. + αq εt−q + β1 σt−12 + . . . + βp σt−p
• = α0 + ∑qi=1 αi εt−i
2
+ ∑pj=1 βj σt−j
2
(4)
• Equation (4) is a GARCH(p, q) process where auto regression is its
squared residuals has an order of q, and the moving average
component has an order of p.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 15 / 69
GARCH Process
• GARCH model is parsimonious as compared to ARCH i.e less lags are
required.
• GARCH(1, 1) model is more simpler to implement than the ARCH
because of the lag length
• Suppose we have a GARCH(1, 1) model. Extending the variance
process backwards yields:
εt2 = α0 + α1 εt−
2 2
1 + β1 σt−1

2 2 2
= α0 + α1 εt−1 + β1 α0 + α1 εt− 2 + β1 σt−2
= .....................
α0 n−1 2
= 1−β + α1 ∑∞m=1 β1 εt−n
1
• Indeed, only the first few terms would have noteworthy influence
since

lim β n →0
n→∞ 1

• This shows how a higher-order ARCH specification can be

approximated by a GARCH(1, 1) process.
Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 16 / 69
GARCH Process
• Stationarity condition:
• The unconditional variance of GARCH would be of interest to the
property of the model.
• Applying the expectations operator to both sides of equation (4), we
have:  
E σt2 = α0 + ∑qi=1 αi E εt−i + ∑pj=1 βj E εt−j

2

2
 
Noting E σt2 = E εt− t−j is the unconditional variance of the

2

2
1 = E σ
residual, which is solved as:
α0
σ 2 = E σt2 =


1 − ∑qi=1 αi + ∑pj=1 βj
It is clear that for the process a finite variance, the following condition
must be met:
q p
∑ αi + ∑ βj < 1
i =1 j =1

In commonly used GARCH (1, 1) model, the condition is simply α1 + β1 < 1.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 17 / 69
GARCH Properties
• Many financial time series have persistent volatility, i.e the sum of α1
and β1 is close to being unity.
• A unity sum of α1 and β1 leads to so-called Integrated GARCH or
IGARCH as the process is not covariance stationary. Nevertheless, this
does not pose as serious a problem as it appears.
• Variations are necessary to adapt the standard GARCH model to the
need arising from examining the time series properties of specific
issues in finance and economics.
• There are different versions of time-varying volatility: ARCH-M &
GARCH-M and the models of asymmetry- Exponential
GARCH(EGARCH) and Threshold GARCH(TGARCH).

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 18 / 69
ARCH-M Model
• When the conditional variance enters the mean equation for an ARCH
process, then ARCH-in-Mean or simply the ARCH-M is derived:

rt = γ1 x1 + . . . . + γm xm + φ σt2 + εt

εt ∼ N 0, εt2

2
σt2 = α0 + α1 εt− 2
1 + . . . . . . . + αq εt−q

where xk , k = 1, 2, . . . m are exogenous variables which could include lagged

yt .
• In the sense of asset pricing, if rt is the return on an asset of a firm,
then xk , k = 1, . . . m would generally include on the market and possibly
other explanatory variables such as the price earning ration and the
size.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 19 / 69
EGARCH Model
• It is given by Nelson(1991) with the following specification:

q q p
ut−j ut−j
log (ht ) = γ + ∑ ζj p + ∑ ξj p + ∑ δi log (ht−i )
j =1 ht−j j =1 ht−j i =1
where ξ is asymmetric response parameter or leverage parameter.
• The sign of ξ is expected to be positive in most empirical cases so that
a negative shock increases the future volatility or uncertainty while a
positive shocks of the same eases the effect on future uncertainty.
• This is in contrast to the standrad GARCH model where shocks of the
same magnitude, positive or negative, have the same effect on future
volatility.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 20 / 69
Treshold GARCH Model
• It is known as the GJR model, named Glosten, Jagannathan and
Runkle (GJR, 1993).
• In contrast, the GJR model is much simpler than, though not as
elegant as, EGARCH.
• A general GJR model is specified as follows:
q  p
2
σt2 = α0 + ∑ αi εt− 2 2
1 + δi εt−1 + ∑ βj εt−1
i =1 j =1

there δi = 0 if εt−i > 0,So,δi is greater than zero, we conclude that

there is asymmetry.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 21 / 69
Multivariate GARCH
• A bivariate GARCH model expressed in matrices takes the form:
rt = εt , . . . . . . (1)
εt−1 | Ωt−1 ∼ N (0, Ht ) where vectors

(5)
 
rt = r1t r2t
 
εt = ε1t ε2t
 
h11t h12t
Ht =
h21t h22t

is the covariance matrix which can be designed in a numver of ways.

• Commonly used specifications of the covariance include VEC, VECH
(full parameterusation), BEKK(positive definite parameterization)
named after Baba, Engle, Kraft and Kroner (1990) and constant
correlation.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 22 / 69
 Market Models
Unit 2

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 23 / 69
Objectives before understanding CAPM
• Investor Utility function

• Capital allocation between risky and risk-free assets

• What is a portfolio and its characteristics

• Mean-variance analysis

• Understand optimal portfolio selection

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 24 / 69
Basic Elements of Investments
• The investor - Risk Aversion & Utility

• Investment Opportunity

• What is a portfolio and its characteristics

∗ Portfolio of assets

∗ Financial asset model - Reward per unit of risk

• Optimal portfolio selection - Weights

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 25 / 69
Risk Aversion & Utility
• Investor does (or should) consider expected return a desirable thing
and variance of return an undesirable thing.
• For every additional unit of risk, investors demand more and more re-
turns.
• Utility is welfare that an investor achieves for an investment decision.

• U = E (Rp ) − 12 ∗ A ∗ σp2
∗ Portfolio of assets

∗ U is the utility value, A (ranges from 0 – 10) is the risk aversion

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 26 / 69
Risk Aversion & Utility

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 27 / 69
Risk Aversion & Utility

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 28 / 69
What is a Portfolio?
• A portfolio is simply a specific combination of securities, usually
defined by portfolio weights that sum to 1.
W = W1 + W2 + W3 + .... + Wn = 1

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 29 / 69
Portfolio- Example!
• Your broker informs you that you only need to keep INR 4,00,000 in
your investment account to support the portfolio of 50 shares of TCS,
100 shares of Adani Green, and 2,000 shares of Gold ETF; in other
words, you can buy these stocks on margin. You withdraw rest 50,000
to use for other purposes. Your portfolio is summarized by the
following weights:

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 30 / 69
Portfolio- Example!
• Suppose you have Rs 10,000 to invest. One risky asset A offers you an
expected return of 4.5% p.a., and risk of 14.5% p.a. You would like to
earn an expected return that is higher than 4.5%. How is it possible,
given there exist a risk-free asset offering a return of 3% p.a.?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 31 / 69
Combination of risk-free asset with risky asset
• When we introduce one risky asset and one risk-free asset in a
market, what is the expected return and risk of the portfolio?

E (Rp ) = wE (Rs ) + (1 − w )Rf

σp = w σs
Also, the expected return can be modified and re-written by substituting
the value of w as:

E (Rp ) = Rf + w [E (Rs ) − Rf ]

[E (Rs )−Rf ]∗σp

E(Rp ) = Rf + σs

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 32 / 69
Incorporating Investor Preference in Asset
Allocation
• Objective is to maximise the utility score. We need to take the first
derivative w.r.t ’w’.

dUp 1
= E (Rs ) − Rf − A ∗ 2 ∗ w ∗ σs2
dW 2
E (Rs ) − Rf
W∗ =
A ∗ σs2
where W ∗ is the optimal weight of the risky asset (s) that maximize
the utility to the investor.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 33 / 69
Optimal Asset Allocation - Example
• You are trying to decide how to allocate your retirement savings
between Treasury bills and the stock market. The T-Bill rate is 0.09%
monthly. You expect the stock market to have a monthly return of
0.75% with a standard deviation of 3.25%.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 34 / 69
Optimal Asset Allocation - Example
• You are trying to decide how to allocate your retirement savings
between Treasury bills and the stock market. The T-Bill rate is 0.09%
monthly. You expect the stock market to have a monthly return of
0.75% with a standard deviation of 3.25%.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 35 / 69
Asset Allocation - Borrowing & Lending

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 36 / 69
Portfolio Analysis - Why needed?
• Don’t put all your eggs in one basket!
• Portfolio provides diversification for reducing the risks.
• Systematic Risk – undiversifiable
Examples: US-China trade war, Russia- Ukraine war, Arab Spring 2011,
Recession in US, Business Cycles.
• Unsystematic/Idiosyncratic Risk – diversifiable (as we go on increasing
the no. of stocks)
Example: Business Risk (increasing operating leverage), Financial Risk
(increasing financial leverage), Regulatory Risk

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 37 / 69
Portfolio Analysis - Why needed?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 38 / 69
Portfolio Analysis - Risk & Reward Assumption
• Investors like high expected returns but dislike high volatility
• Investors care only about the expected return and volatility of their
overall portfolio.
• Portfolio risk depends not only on the individual risk but also on the
interactive risk.
• How much does a stock contribute to the risk and return of a
portfolio, and how can we choose portfolio weights to optimize the
risk/reward characteristics of the overall portfolio?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 39 / 69
Portfolio Analysis - Mean-Variance Analysis
Assume investors focus only on the expected return and variance (or
standard deviation) of their portfolios: higher expected return is good,
higher variance is bad.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 40 / 69
Portfolio Analysis - Mean-Variance Analysis
Basic Properties of Portfolio Mean and Risk:

E (Rp ) = w1 E (R1 ) + w2 E (R2 ) + w3 E (R3 ) + .....wn E (Rn )

σ (Rp )2 = E (Rp − µ)2
= E[(w1 (R1 − µ1 ) + w2 (R2 − µ2 )..... + wn (Rn − µn ))2 ]

E[wi wj (Ri − µi )(Rj − µj )] = wi wj Covij

= wi wj σi σj ρij

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 41 / 69
Portfolio Analysis - Mean-Variance Analysis
• Portfolio variance is the weighted sum of all the variances and
covariances.

• There are n variances and n2 − n co-variances or unique combinations

of n C2
• Positive covariances increase portfolio variance; negative covariances
decrease portfolio variance (diversification)

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 42 / 69
Portfolio Analysis - For Two Asset Cases

E (Rp ) = w1 E (R1 ) + w2 E (R2 )

Var (Rp ) = w12 σ12 + w22 σ22 + 2w1 w2 σ1 σ2 ρ12

• What if the investor wants to minimize the portfolio risk?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 43 / 69
Portfolio Analysis - For Two Asset Cases
Put
w2 = 1 − w1
σ22 − COV12
wmin =
σ1 + σ22 − 2COV12
2

• No other portfolio can have a lower risk.

• Can the portfolio risk be reduced to zero?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 44 / 69
Portfolio Analysis - For Two Asset Cases
From January 2023 – June 2023, BPCL had an average monthly return of
0.081% and a std dev of 1.427%. Bajaj Auto had an average return of
0.291% and a std dev of 1.338%. Their correlation is -0.06. How would a
portfolio of the two stocks perform?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 45 / 69
Portfolio Analysis - For Two Asset Cases
Suppose the correlation between BPCL and Bajaj Auto. What if it equals
–1.0? -0.7? 0.0? 1.0?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 46 / 69
Choosing the optimum portfolio from the efficient
frontier?
• As a set of possible portfolios, the efficient frontier dominates all
other combinations of risky assets.
• But they all are efficient.
• We need a way to choose among efficient portfolios, and combining a
risk- free asset with portfolios on the efficient frontier.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 47 / 69
Tangency Portfolio
• If there is also a riskless asset (T-Bills), all investors should hold
exactly the same stock portfolio!
• All efficient portfolios are combinations of the riskless asset and a
unique portfolio of stocks, called the tangency portfolio.
• The tangency portfolio has the highest possible Sharpe ratio of any
portfolio.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 48 / 69
Tangency Portfolio
Optimal Combination of risky (portfolios of BPCL and Bajaj Auto) and risk
free assets:

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 49 / 69
Optimal Risky portfolio
Point of tangency: maximise the slope of CAL
Replace E (Rp ) with w1 ∗ E (R1 ) + (1 − w1 ) ∗ E (R2 )

E (r1 )σ22 − E (r2 )COV12

w∗ =
E (r1 )σ22 + E (r2 )σ12 − [E (r1 ) + E (r2 )]COV12

where,
E (r1 ) = E (R1 ) − Rf
E (r2 ) = E (R2 ) − Rf

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 50 / 69
Comprehensive Question
A security analyst has the following information about two risky assets
(Equity and Bond) and T-bills (riskless asset).
• Coefficient of correlation between equity and bond is 0.3
• Find the mean-variance portfolio weight of risky assets.
• Find the optimal risky portfolio, P and its expected return and risk.
Answer: WE = 72%, E (RP ) = 13.04%, σP = 15.74%
• Find the slope of CAL supposed by T-bills & Portfolio P. Ans: 0.51
• Find the complete optimal portfolio for Mr. Mehta having risk
aversion index, A =5. Ans: Wr isky = 65%, complete portfolio: Rf =35%,
We = 47%, Wb = 18%

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 51 / 69
CAPM
• A new investment opportunity comes along (it, too, is risky). Should
she add a bit of the new risky asset (call it asset X) to the existing
portfolio?
• The new asset should be added to the portfolio only if its addition
increases the portfolio’s Sharpe Ratio.

E (x) − rf E (p) − rf
≥
σx σp
or

E (x) − rf E (p) − rf
=ρ
σx σp
E (x) = rf + β (E (p) − rf )

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 52 / 69
CAPM
• βx measures the incremental risk that asset X will bring to the
portfolio and for which the investor must be compensated, by means
of the expected return on asset X.
• It was developed by Sharpe (1964), Lintner (1965), and Mossin (1966).
• At equilibrium, investors collectively hold exactly the supply of all
available assets. Demand for risky assets is equal to the supply.
• All risky assets as they are priced, so risk-averse investors are holding
the market portfolio.
• Assets are infinitely divisible.
• Not all risks affect equilibrium expected returns – only those risks are
considered that cannot be diversified away.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 53 / 69
Understanding CAPM
• In equilibrium all assets are priced so that the expected return on any
given asset equals the riskfree return plus a risk premium consisting
of the asset’s beta multiplied by the risk premium for the market as a
whole.
• The CAPM equation is also called the security market line (SML).
• It shows both individual securities & efficient portfolios.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 54 / 69
Understanding CAPM
• If the security lies on SML it is efficiently priced. If it lies above SML, its
underpriced and if it lies below SML, its overpriced.
• The expected return on an individual asset depends on β rather than
σ . Beta measures the stock’s nondiversifiable, or systematic, risk.
• Standard deviation, σ , is the pertinent measure of risk for diversified
portfolios, but β is the pertinent measure for individual securities.
This is not to say that an individual security’s σ does not measure risk;
rather, σ for an individual stock measures both diversifiable and
nondiversifiable (both nonsystematic and systematic) risk.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 55 / 69
Understanding CAPM
• Assume that the following securities are correctly priced according to
the CAPM. Derive SML.
R1 = 6% R2 = 12%
B1 = 0.5 B2 = 1.5
• What is the exp. Return on a security having B = 2?

• Assume that over some period a CAPM was estimated as

E (Ri ) = 0.06 + 0.10Bi
• Assume that over the same period 2 securities had the following
results:
S : ActualReturns = 15%, Bi = 0.8
Q : ActualReturns = 17%, Bi = 1.2
• What can be said about the performance of securities?

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 56 / 69
Arbitrage Pricing Theory
• What If There Are Multiple Sources of Systematic Risk?
• Let returns follow a multi-factor linear model:
E (Ri ) = Rf + β1 F1 + β2 F2 + β3 F3 + ...... + εi

• Developed by Stephen Ross (1976)

• Three key propositions:
∗ Security returns can be described by a factor model.
∗ There are sufficient securities to diversify away idiosyncratic (or
unsystematic) risk.
∗ Well functioning security markets do not allow for the persistence
of arbitrage opportunities.
∗ APT states that security returns are explained by a number of
systematic factors and not just one factor (or market factor).

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 57 / 69
Multi Factor Asset Pricing Model
• In this we have more than one factor to explain security returns.
• The standard CAPM uses only the market factor to explain security
returns.
• Why there was a need for additional risk factors beyond CAPM?
• Does the inclusion of additional risk factors in line with EMH?

E (Ri ) = Rf + β1 F1 + β2 F2 + β3 F3 + ...... + βn Fn

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 58 / 69
Basic Terminology
• Factor Models
• Common risk factors
• Zero Intercept
• Mimicking Portfolio
• Three Factor Model

Ri,t − Rf = αi + βm (Rm − Rf ) + βs SMBt + βh HMLt + εi,t

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 59 / 69
How to construct factors?
• Six portfolio formed using ME and BE/ME sorts.
• Vertical line – rank stocks on the ME (Big V/S Small)
• Horizontal line – rank stocks on BE/ME (take 30th and 30th percentiles
i.e., Big V/S Small)
• SMB = 1/3 (Small Value + Small Neutral + Small Growth) – 1/3 (Big
Value + Big Neutral + Big Growth)
• HML = ½ (Small Value + Big Value ) – ½ (Small Growth + Big Growth)

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 60 / 69
Economic Rationale behind Additional Factors
• Why Size and Value factors and what do they reveal?
• SMB and HML are not an obvious choice for risk factors.
• They may proxy yet unknown more fundamental variables.
• Firms with high BE/ME are more likely to be in financial distress.
• Firms with small sizes are more susceptible to changes in business
cycles.
• These variables may capture the sensitivity of risk factors in the
macroeconomy.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 61 / 69
Market Anomalies
Value Anomaly:

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 62 / 69
Market Anomalies
Size Anomaly:

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 63 / 69
Market Anomalies
Momentum Anomaly:

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 64 / 69
 Smart Beta Investing Strategies
Unit 2

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 65 / 69
Smart Beta Investing
• Smart beta as strategic beta, active beta, enhanced index, alternative
beta, and scientific beta.
• Focuses on weighting stocks in smart beta portfolios not on the
traditional measure of market capitalization, but by incorporating into
their weighting scheme some aspect of a security’s fundamental
value:

∗ stock’s B/M ratio,

∗ Cash flows, Dividends and buybacks, profitability,
∗ Characteristic of the security’s performance, such as a stock’s
momentum/ low volatility.

• Considered as gateway to factor investing - enable to diversify

your portfolio across factor portfolios.

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 66 / 69
Smart Beta Strategies

Source: Braun, A. P. (2018). Smart-Beta Exchange Traded Funds & Factor Investing, Case Study, Northwestern Kellogg School of Management

Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 67 / 69
Construction of Smart Beta Startegies

Fij
Wij = N
Σi =1 Fij
• wi,j is the fundamental weight for some security i for some financial
variable j, Fi,j is the value of the financial variable j (such as B/M) for
firm i and N is the number of securities in the sample.
• For more than one financial variable: the weights for each financial
variable j from above are averaged across all the financial variables
used to define the index and then re-standardized.
∗ MSCI assigns each stock in their sample what is called a Z-score.
Z-score as a measure of how many standard deviations a financial
variable is from its mean.
Fij − µ(Fj )
Zij =
σ (Fj )
• MSCI USA Enhanced Value Index uses the forward price-to-earnings
ratio, the enterprise value to operating cash flow ratio, and the
price-to-book ratio.
Preeti Roy (IIT(ISM) Dhanbad) Financial Analytics - DE course by DMS&IE July 2024 - Nov 2024 68 / 69
 Thank You

Open for Questions/ Comments