FinTree © 2024 FinTree Education Private Ltd

Rates and Returns

FinTree Fruit 1 : Interest Rate Interpretation
Interest rates can be thought of in three ways: 1. Required Rate of Return
2. Discount Rate
3. Opportunity Cost
(We use the terms “interest rate” and “discount rate” interchangeably)

FinTree Fruit 2 : Interest Rate Components

• Interest rates are set by the forces of supply and demand, where investors supply funds and borrowers
Demand their use
• Nominal Interest Rate = Real risk-free interest rate + Inflation premium + Default risk premium
+ Liquidity premium + Maturity premium.

Real risk-free interest rate Inflation premium Default risk premium

Reflects the time preferences Compensates investors Compensates investors

Of individuals for
Current versus future for expected inflation for Default risk
Real consumption.

Liquidity Premium Maturity Premium

YTM Long Term Debt (-) YTM Short Term Debt

Many bonds of small
Issuers trade infrequently after
they are issued;
the interest rate on such bonds
includes a liquidity premium

FinTree Fruit 3 : Nominal vs Real Interest Rates

(1 + Nominal risk-free rate) = (1 + real risk-free rate) * (1 + inflation premium)

Example 1: Inflation is 5%, Real RFR is 3% , Nominal Rate will be (1+ 5%)*(1+3%) -1 = 8.15%
(Close to 3% + 5% = 8%, however geometric chaining makes it 8.15%)

Example : Nominal RFR is 8.15%, Real RFR is 3% , Inflation will be (1+ 8.15 %) / (1+3%) -1 = 5%
(Again close to 8.15% - 5% = 3.15%)
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 4 : Interest Rate Quote Convention

Interest rates are always quoted in annual terms, so the interest rate on a 90-day government debt security
Quoted at 4 percent is the annualized rate and not the actual interest rate earned over the 90-day period

Interest earned on $100 investment will be $100 * 4% *90/360 = $100* 1% = $1

FinTree Fruit 5 : Holding Period Return (HPR)

Example 1 : $ 100 Investment becomes $ 105 in some time (time doesn’t matter for HPR) and also gives you a
Dividend of $ 3.

HPR = (105+ 3) / 100 = 8% ; Capital Gains = 5%, Income Yield = 3%

Example 2: Let’s say returns earned in first 3 months is 4%, next 7 months is 3% and last 2 months of the year
Is 6%, what will be one year holding period return ?

= (1+4%) * (1+3%) * (1+6%) - 1 = 13.55%

FinTree Fruit L6 : Geometric

Mean Return (GM) & Arithmetic Mean Return (AM)
Example 1 : $ 100 Investment becomes $ 105 in some time (time doesn’t matter for HPR) and also gives you a
Dividend of $ 3.
Example 1 : Calculate AM and GM Returns based on following data :
HPR = (105+ 3) / 100 = 8% ; Capital Gains = 5%, Income Yield = 3%
AM = ( 6% + 8% - 9% + 7%) / 4 = 3%
Example 2: Let’s say returns earned in first 3 months is 4%, next 7 months^ (1/4)
is 3% and last 2 months of the year
GMreturn
is 6%, what will be one year holding period = (1.06
? * 1.08 * 0.91 *1.07) -1= 2.75%

Another way to think of GM is , start with $ 100, grow @ 6% = 106,

= (1+4%) * (1+3%) * (1+6%) - 1 = 13.55%
then $ 106 grows @ 8% = 114.48, then @ -9% = 104.17 and finally
at 7% = 111.47 . So in four years, $ 100 are becoming $ 111.47.
Using TVM row :

Time 0 Year 1 Year 2 Year 3 Year 4

$ 100 $ 106 $ 114.48 $ 104.17 $ 111.47

100 PV N=4 111.47 FV

CPT I/Y = 2.75%

• AM is generally biased upwards (compared to GM), i.e. AM ≥ GM.

• Bias in AM returns is particularly higher if holding period returns are a mix of both positive
and negative returns, as in this example.
• In general, the difference between the AM and GM increases with the variability within the sample;
the more disperse the observations, the greater the difference .
 FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 7 : Harmonic Mean

• Used when variable is a rate or a ratio, such as PE ratio.

• Useful measure when variables have outliers.
• Use HM if investments are made by purchasing constant quantity of stocks every month/year
(Dollar cost averaging)
• Use AM investment are made by purchasing constant amount every month/year

• AM * HM = GM2
• Unless all observations are same values, AM > GM > HM
• Harmonic Mean is calculated as:
n
1 + 1 +......+ 1
X1 X2 Xn
Example1: Calculate AM, GM and HM of PE Ratios
Stock PE Ratio
A 3
• AM = (3 + 4 + 5 + 20) / 4 = 8
B 4
C 5 • GM = (3 * 4 * 5 * 20)(1/4) = 5.88
D 20 4
• HM = = 4.8
1 1 1 1
3
+
4 + 5 + 20

• Notice , AM > GM > HM

• Notice , HM is least affected by outlier value of 20

FinTree Fruit 8 : Trimmed Mean and Winsorized Mean

• Both used to minimize the impact of outliers in a database

• Trimmed mean removes a small defined percentage of the largest and smallest values from a dataset
before calculating the mean by averaging the remaining observations.

• Winsorized mean is calculated after replacing extreme values at both ends with the values of their
Nearest observations, and then calculating the mean by averaging the remaining observations.

• In summary: HM, Winsorized and Trimmed mean are solutions to extreme outliers values

• GM is used when compounding is required

• HM is used while averaging ratios (like PE)

 FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 9 : Time Weighted Rate of Return (TWRR) & Money Weighted Rate of Return (MWRR)

• MWRR is just the IRR if investment inflows and outflows

• MWRR is more accurate measure of what investor earned during the period , however it is not comparable
across investors
• TWRR is GM of all the Holding Period Returns
• TWRR is not sensitive to investment inflows and outflows.
• Investment managers prefer TWRR as they do not control timing of inflows and outflows.

TWRR can be computed in following steps :

1. Price the portfolio immediately prior to any significant addition or withdrawal of funds.
2. Break the overall evaluation period into sub periods based on the dates of cash inflows and outflows.
3. Calculate the holding period return for each sub period.
4. Calculate Geometric Mean of HPR of sub periods.

Example 1 : Calculate MWRR and TWRR

Time Stock Price Qty Purchased Div Per Share

0 10 1
1 20 4 1
2 25 10 1
3 50 15 (Sold) 2

MWRR can be computed by calculating IRR of Net Cash Flow (Last Column)
FinTree © 2024 FinTree Education Private Ltd

= 3 (1+110%) * (1+ 30%) * (1+ 108%) - 1 = 78.40%

FinTree Fruit 10 : Non Annual Compounding Rate

Example 1 : Fill in the blanks of the following table

FinTree © 2024 FinTree Education Private Ltd

Solution with TVM row keys is as follows:

FinTree Fruit 11 : Annualizing Rates

Example 1: Convert quoted rates to Annualized Rates using I-Conversion Function of the Table
To Access Interest Conversation Function, press 2nd and 2 on TI BA II Plus Calculator
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 12 : Continuously Compounded Returns

Example 1: Fill in the blanks Solution:

FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 13 : Return Measures

• Gross Return is the return earned by an asset manager prior to deductions for management expenses,
Custodial fees, taxes etc.
• Trading expenses, such as commissions, are deducted from the computation of gross return.
• Gross Return is an appropriate measure for comparing the investment skill of asset managers.
• Net return accounts for all managerial and administrative expenses that reduce an investor’s return.
• Because individual investors are most concerned about the net return, small mutual funds with a
are at a disadvantage compared with the larger funds that can spread their largely fixed administrative
expenses over a larger asset base (economies of scale)
• Post Tax Returns = Pre tax Returns - Taxes Paid = Pre Tax Returns * (1- tax rate)
• Risk Premium can be calculated as ={ (1+ total return on an asset) / (1+ RFR) } - 1
• Leveraged Return = Returns earned on equity (own money) invested after accounting for interest paid
on borrowed money to fund that investment

Example 1 : If you buy a stock worth $ 100, by borrowing $80 and investing $20 of equity (own).
At the end of the year , the stock value is $ 120. Interest rate on borrowing is 12.5%.
Calculate Asset Return (unlevered) and Leveraged Return.

Unlevered Return
= Profit / Total Value of Asset at the time of investment
20
= 20%
100
Leveraged Return
= Profit made after interest payment / Equity Invested
20 - 80*12.5%
= 50%
20

Another way to calculate leveraged return:

Unlevered Return + Debt/Equity Ratio * (Unlevered Return - Interest Rate)
= 20 % (+) 80/20 * (20%- 12.5%)
= 20% (+) 4 * 7.5%
= 50%
FinTree © 2024 FinTree Education Private Ltd

The Time Value of Money in Finance

FinTree Fruit 1 : Valuation of a Coupon Bond

FinTree
• A bond is a fixed-income instrument that represents a loan made by an investor to a borrower
• The face value of a bond is the price that the issuer pays at the time of maturity, also referred
to as “par value”
• The coupon rate is the interest rate paid on a bond by its issuer for the term of the security.
• It’s calculated as Face Value * Par Rate
• The term yield to maturity (YTM) refers to the total return anticipated on a bond if the bond is held
until it matures.
• A bond’s market value is how much someone will pay for the bond on the free market. It is calculated as
Present Value of future cash flows of the bond discounted at YTM.

Example 1: Calculate Value of the bond based on following data

Solution
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : PV of Perpetuity and Amortization Schedule

FinTree
• A perpetual bond is a less common type of coupon bond with no stated maturity date.
• PV(Perpetual Bond) = PMT / Discount Rate
• Amortization Schedule = Schedule that displays how loan is repaid over a period of time.

Example 1: Build Amortization Scehdule for a Loan of $ 100, to be repaid over 5 years with
level payments, at interest rate of 10%.

Use TVM row to calculate annual Installment to be paid.

2nd CLR TVM, 100 PV , 10 I/Y, 5 N , CPT PMT = 26.38

FinTree Fruit 3 : Valuation of Equity (Dividend Discount Model)

• Value of Equity Share is PV of all Future Cash Flows.
• Future Cash Flows include dividends and Terminal Value
• Terminal Value is the estimated value of the stock at the end of the forecast horizon.
• Estimated dividend payments could be:
1. Constant Payment in perpetuity = Dividend Pmt / Disc Rate
2. Constant Growth in Perpetuity (Gordon Growth Model) = D1 / (Disc Rate - growth rate)
3. Different Growth rates (2 Stage and 3 Stage models) - Look example below

Example 1 : A stock is expected to pay a dividend of $ 10 in perpetuity. Expected return on the

stock is 5% (Discount rate). Determine Value of the stock

Value0 = Dividend1 / Discount Rate

V0 = $10 / 5% = $ 200
 FinTree © 2024 FinTree Education Private Ltd

Example 2 : A stock is expected to pay a dividend of $ 10 next year. Expected return on the

FinTree
stock is 10% (Discount rate). Dividends are expected to grow at 4%.

Value0 = Dividend1 / (Discount Rate- growth rate)

V0 = $10 / (10% - 4%) = $ 166.67

Example 3 : A stock paid a dividend of $ 10 last year. Expected return on the

stock is 10% (Discount rate). Dividends are expected to grow at 4%.

Value0 = Dividend1 / (Discount Rate- growth rate)

Value0 = Dividend0 * (1+ growth rate) / (Discount Rate- growth rate)
V0 = $10 * (1+ 4%) / (10% - 4%)
V0 = $10.4 / (10% - 4%) = 173.33

Example 4 : A stock paid a dividend of $ 10 last year. Dividends are expected to grow at 15% for
first 3 years, then at 10% for next two years and 4% therefter in perpetuity. Discount rate is 10%.

Time 0 Time 1 Time 2 Time 3 Time 4 Time 5

10 * (1+15%) 11.5 * (1+15%) 13.225 * (1+15%) 15.21 * (1+10%) 16.73 * (1+10%)

= 11.5 = 13.225 = 15.21 = 16.73 = 18.40
CF1 CF2 CF3 CF4
Terminal Value5 = D6/ (ke-g)
=18.40*(1+4%)/ (10%-4%)
= 19.14 / 6%
= 318.98
= CF5 = 18.40 + 318.98
=337.3782

Compute NPV, by using NPV function by inserting on TI BA II Plus Calculator.

Keep I = 10%. NPV = 253.72

FinTree Fruit 4 : Implied Growth and Discount Rates

• Gordon Growth Model = V0 = D1/ (ke - g)
• Rearranging the equation, ke - g = D1/ V0 ; ke = D1/ V0 + g
• g = ke (-) D1/ V0
FinTree © 2024 FinTree Education Private Ltd

Example 1 : A stock is trading at $ 100. It is expected to give a dividend of $10 next year.

FinTree
Cost of equity is 15%. Calculate implied growth rate.

Value0 = Dividend1 / (Discount Rate- growth rate)

$ 100 = $ 10 / ( 15% - g)
15% - g = $10 /S100
g = 15% - 10% = 5%

Example 2 : A stock is trading at $ 100. It is expected to give a dividend of $10 next year.
g = 5%. Calculate implied cost of equity.

Value0 = Dividend1 / (Discount Rate- growth rate)

$ 100 = $ 10 / ( Ke - 5%)
Ke - 5% = $10 /S100
Ke = 10% + 5% = 15%

FinTree Fruit 5 : Price Earning Ratio

• If a stock is trading at $100, it earned Earning Per Share (EPS0) last year of $10.
It’s Trailing (because we are using EPS0) PE ratio will be $100/$10 = 10.
• If a stock is trading at $100, it is expected to earn Earning Per Share (EPS0) next year of $20.
It’s Leading (because we are using EPS1) PE ratio will be $100/$20 = 5.
• Justified PE Ratio will have V0 (calculated using GGM) in the numerator, as against actual
market price.
• Justified Trailing PE Ratio = V0 / EPS0 = Payout Ratio / (Ke-g)
• Justified Leading PE Ratio = V0 / EPS1= Payout Ratio*(1+g) / (Ke-g)
• Payout Ratio = % Earnings paid out as Dividends

Example 1 : A company has EPS of $10, it pays Dividends Per Share (DPS) of $4. It has Ke= 10%
and g = 4%. Calculated justified Leading and Trailing PE Ratio.

Justified Trailing PE Ratio = Payout Ratio / (Ke-g) = 40% / (10% - 4%) = 6.67
Justified Leading PE Ratio = Payout Ratio*(1+g) / (Ke-g) = 6.93
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 6 : Cash Flow Additivity Principle

FinTree
• Under cash flow additivity, the present value of any future cash flow stream indexed at the same point
equals the sum of the present values of the cash flows
• This principle ensures that market prices reflect the condition of no arbitrage action costs.

Example 1: Based on the stream of Cash Flows presented in Investment 1 & Investment 2,
recommend which investment strategy to choose by Comparing Net Present Value (NPV).
Required rate of return = 10%

Investment 1

Time 0 Time 1 Time 2 Time 3 Time 4 Time 5

Amount Inflow Inflow Inflow Inflow Inflow

Invested $40 $40 $40 $40 $40
$ 100

Investment 2

Time 0 Time 1 Time 2 Time 3 Time 4 Time 5

Amount Inflow Inflow Inflow Inflow Inflow

Invested $10 $20 $60 $80 $42.34
$ 100

Solution: To recommend an investment strategy, compare NPV of two options and select
strategy with higher positive NPV.

For strategy 1, CF0 = -100, CF1= 40, F1 =5 , I= 10 , NPV = 51.63

For strategy 2, CF0 = -100, CF1 =10, CF2 = 20, CF3 = 60, CF4 = 80 , CF5 = 42.34
I = 10 , NPV = 51.63

Since both the investments are producing same NPV, we will be indifferent between the two.

FinTree Fruit 7 : Implied Forward Theory

• A forward rate is an interest rate applicable to a financial transaction that will take place in the future
• Let’s say one year rate is 5%, whereas two year rate is 7%. We can calculate one year forward rate
one year from now (F1,1) will be calculated as {(1+7%)2 / (1+5%)1} - 1 = 9.038%
• Interpretation of this rate:
PV = 100 FV =114.49
N = 2 , I/Y = 7

PV = 100 FV =105 FV =114.49

N = 1 , I/Y = 5 N=1, I/Y = 9.038
FinTree © 2024 FinTree Education Private Ltd

Example 1 : 1 year zero coupon bond (STRIP) is trading at 96, 2 year zero coupon bond (STRIP)

FinTree
is trading at 88. Calculate on year rate one year from now (F1,1)

STEP 1 : Calculate one year (spot) rate . PV = 96 FV = -100, N=1 , CPT I/Y = 4.17%
STEP 2: Calculate two year (spot) rate . PV = 88 FV = -100, N=2 , CPT I/Y = 6.6%
STEP 3: Calculate F1,1= (1+6.6%)2 / (1+4.17%) - 1 = 9.09%

Forward Exchange Rates - Interest Rate Parity

FinTree Fruit 8 :
• The forward exchange rate (also referred to as forward rate or forward price) is the exchange rate
at which a bank agrees to exchange one currency for another at a future date when it enters
into a forward contract with an investor
• Forward Price is calculated using Interest rate parity.
• Interest rate parity is the fundamental equation that governs the relationship between interest rates
and currency exchange rates.

Forward Exchange Rate (DC/FC) = Spot Rate * (1+ Interest Rate of DC)n
(1+ Interest Rate of FC)n

For example, Let’s say spot exchange rate = INR 50 / $

Interest rate in India = 8% pa
Interest rate in US = 3% pa
1 Year Forward Rate = 50 * (1+8%)
1

(1+3%)1
= 50 * 1.08 / 1.03 = 52.42

This forward rate of INR 52.42/ $ is a No Arbitrage Price.

Same example using continuous compounding will be as follows:

e8%*1
1 Year Forward Rate = 50 * = INR 52.56/$
e3%*1

Same example using continuous compounding, with maturity of 3 months will be as follows:

e8%*0.25
1 Year Forward Rate = 50 * = INR 50.63/$
e3%*0.25
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 9 : Option Pricing using Cash Flow Additivity

• Call options are financial contracts that give the buyer the right but not the obligation to buy a stock at

FinTree
specific price (strike price)
• A call buyer profits when the underlying asset increases in price

Let’s say, a stock price is 100. In one year the price can either go up to 150 or 90.
Let’s assume the option has a call option has a strike price of 120 (i.e.Right to buy the asset at 120).

Stock Price at
Option Price
time1 = 150 time1 = 150 - 120
= 30

Stock Price at
time0 = 100

Option Price
Stock Price at
time1 = 90
time1 = 0
Investor will
not exercise.
• Hedge Ratio can be calculated as = Option Price @ uptick - Option Price @ down-tick

Stock Price @ uptick - Stock Price @ down-tick

30 - 0
=
150 - 90
= 0.5
• To determine the price of the option, we will create a replicating portfolio. Option seller will
sell an option at Price C and buy 0.5 (hedge ratio) stock.
• Total Investment upfront will be 0.5*100 - C.
• On expiry, the portfolio will be priced either at (uptick) 0.5*150 - 30 = 45 or 0.5*90 - 0 = 45.
• Therefore , PV of the replicating portfolio today will be (assume RFR=10%) = 45/(1+10%) =40.90
• Since, amount invested today is 0.5*100 - C = 40.90; C = 4.09
FinTree © 2024 FinTree Education Private Ltd

Statistical Measures of Asset Returns

FinTree Fruit 1 : Measures of Location

FinTree
Measures of Location

Measures of Central Other Measures of

Tendency location

1. Sample Mean Quantile: General Term for a value at or

below which a stated percentage of the data
2. Median: value of the middle item of a
lies.
dataset that has been sorted into
ascending or descending order, (n+1)/th Type of Quantiles:
Observation 1. Quartiles divide the distribution into
3. Mode: Mode is the most frequently quarters (4 parts)
occurring value; single mode-unimodal 2. Quintiles into fifths (don’t confuse this
dataset; two modes - bimodal dataset with Quantiles)

3. Deciles into tenths

Dealing with outliers in dataset Interquartile Range

three options exist for dealing with extreme

values: • Difference between the third quartile and
1. Do nothing; use the data without any the first quartile
adjustment.

2. Delete all the outliers (Trimmed Mean)

3. Replace the outliers with another value

(winsorized mean)
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : A Box and Whisker Plot

• The “box” represents the lower bound of the second

FinTree
Highest Value quartile and the upper bound of the third quartile, with
the median or arithmetic average noted as a measure of
central tendency of the entire distribution

Upper Boundary for Q3 • The whiskers are the lines that run from the box and
Median are bounded by the “Fences” which represent the lowest
X Mean
and highest values of the distribution.
Lowest Boundary for Q2
• Another form of box and whisker plot typically uses 1.5
times the interquartile range for the fences.
Lowest Value
• Thus, the upper fence is 1.5 times the interquartile
range added to the upper bound of Q3, and the lower
fence is 1.5 times the interquartile range subtracted
from the lower bound of Q2.

• Observations beyond the fences (i.e., outliers) may also

be displayed.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Measures of Dispersion

Measures of Dispersion
Dispersion is the variability around the
central tendency

Absolute Dispersion Relative Dispersion

1. Range: Maximum value − Minimum 1. Coefficient of Variation

value.

2. Mean absolute deviation (MAD)

where s is the sample standard deviation
and X is the sample mean.
3. Sample Variance

4. Sample Standard Deviation

5. Target semi-deviation (target downside

deviation

where B is the target and n is the total

number of sample observations.
FinTree © 2024 FinTree Education Private Ltd

Example 1: Based on data below, calculate:

FinTree
1. Sample Mean
2. Sample SD and Sample Variance
3. Population SD and Population Variance
4. Range
5. Mean Absolute Deviation
6. Semi-deviation (mean = target)
7. Target Semi-deviation (Target = 13)
8. Coefficient of Variation

To calculate 1, 2 & 3, use STAT function of calculator.

Enter Data by pressing 2nd and 7 and access
Stats by pressing 2nd and 8. Solution is as follows:

Solution:

1,2,3 & 4 5. Mean Absolute Deviation

6. Semi-deviation
FinTree © 2024 FinTree Education Private Ltd

FinTree
7. Target Semi-deviation (Target = 13) 8. Coefficient of Variation
= Sample SD/ Mean
= 8.72 / 10
= 0.872

FinTree Fruit 4 : Measures of shape of a distribution

• Normal Distribution : symmetrical, bell-shaped
distribution
• Its mean, median, and mode are equal.
• It is completely described by two parameters—its
mean and variance (or standard deviation).
• Other distributions may require more
information than just the mean and variance to
characterize their shape

• Skewness : A distribution that is not

symmetrical is termed skewed
• A return distribution with positive skew has
frequent small losses and a few extreme gains
• Positive Skew : Mean > Median > Mode
• A return distribution with negative skew has
frequent small gains and a few extreme losses
• Negative Skew : Mean < Median < Mode
FinTree © 2024 FinTree Education Private Ltd

• Kurtosis is a measure of the combined weight of the tails of a distribution relative to the rest of the

FinTree
distribution that is, the proportion of the total probability that is outside of, say, 2.5 standard
deviations of the mean
• A distribution that has fatter tails than the normal distribution is referred to as leptokurtic or
fat-tailed
• A distribution that has thinner tails than the normal distribution is referred to as being platykurtic
or thin-tailed
• Distribution similar to the normal distribution as it concerns relative weight in the tails is called
mesokurtic
• Excess Kurtosis : Kurtosis of the distribution - 3
• Why? Because normal distribution has a kurtosis of 3
• So, if K> 3 then distribution is Leptokurtic
• If K = 3 , then distribution is platykurtic (Normal Distribution)
• If K < 3, then distribution is Mesokurtic platykurtic

Higher Density (compared to ND) in the center

Lower Density (compared to ND) in the middle

Higher Density (compared to ND) in the tails

FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5 : Correlation and Covariance

FinTree
• A scatter plot is displays potential relationships between two variables
• Pattern of the scatter plot may indicate no apparent relationship, a linear association, or a
non-linear relationship
• It provides a quick sense of the data range.
• Inspecting the scatter plot can help to spot extreme values (i.e., outliers).
• Correlation is a measure of the linear relationship between two random variables
• The first step in considering how two variables vary together, however, is constructing their
covariance. It measures how two variables in a sample move together.
• covariance is a measure of the joint variability of two random variables
• If the random variables vary in the same direction—for example, X tends to be above its mean
when Y is above its mean, and X tends to be below its mean when Y is below its mean—then their
covariance is positive
• The size of the covariance measure is difficult to interpret as it involves squared units of measure
and so depends on the magnitude of the variables

• The sample correlation coefficient is a standardized measure of how two variables in a sample
move together.
• Correlation coefficient expresses the strength of the linear relationship between the two random
variables
• Correlation (X,Y) = Covariance(X,Y)
SDx * SDy
• Correlation ranges from −1 and +1.
• A correlation of 0, termed uncorrelated, indicates an absence of any linear relationship between
the variables
• A positive correlation close to +1 indicates a strong positive linear r ship. A correlation of 1
indicates a perfect linear relationship.
• A negative correlation close to −1 indicates a strong negative (i.e., inverse) linear relationship. A
correlation of −1 indicates a perfect inverse linear relationship.
FinTree © 2024 FinTree Education Private Ltd

FinTree
y y

Perfect Positive Relationship x Perfect Negative Relationship x

Correlation = + 1 Correlation = + 1

y y

x x
Perfect Positive Relationship Perfect Negative Relationship
Correlation between 0 to + 1 Correlation between 0 to - 1

y y

x x
No Linear Relationship Non Linear Relationship
Correlation = 0 Correlation = 0
Correlation captures only linear relationships
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 6 : Limitations of Correlation Analysis

•
•
•
•
FinTree
Correlation does not capture non-linear relationship
Correlation may be quite sensitive to outliers.
correlation does not imply causation
Spurious correlation can be on the three types:
1. correlation between two variables that reflects chance relationships in a particular dataset;
2. correlation induced by a calculation that mixes each of two variables with a third variable; and
- For example, consider a cross-sectional sample of companies’ dividends and total assets. While
there may be a low correlation between these two variables, dividing each by market
capitalization may increase the correlation.
3. correlation between two variables arising not from a direct relation between them but from their
relation to a third variable.
- height may be positively correlated with the extent of a person’s vocabulary, but the underlying
relationships are between age and height and between age and vocabulary.
FinTree © 2024 FinTree Education Private Ltd

Probability Trees and Conditional Expectations

FinTree Fruit 1 : Expected Value and Variance

FinTree
• The expected value of a random variable is the probability-weighted average of the possible
outcomes of the random variable. For a random variable X, the expected value of X is denoted
E(X).

• The variance of a random variable is the expected value (the probability-weighted average) of
squared deviations from the random variable’s expected value:

• Variance and standard deviation can not be negative.

• Standard deviation is the square root of variance.

Example 1 : Calculate Expected Value, Variance and Standard Deviation

Solution:
FinTree © 2024 FinTree Education Private Ltd

• The same question can also be solved using STAT function of the calculator with following

FinTree
steps :
STEP 1 : 2nd 7 , 2nd CLR WRK
STEP 2 : X01 = 5, Y01= 20 (not 20%)
X01 = 10, Y01= 30
X01 = 20, Y01= 40
X01 = 30, Y01= 10
STEP 3: Press 2nd and 8 , Press 2nd SET multiple times till you reach 1-V Display
STEP 4: Navigate downwards, Mean will be 15 and population sigma will be 7.74596 ~ 7.75

FinTree Fruit 2 : Total Probability Rule

Example 1: If probability studying for the exam P(S) is 70%, the probability of not study-
ing for the exams P(Sc) = 30% (called complement).
If Probability of passing if a student studies P(P/S)= 80% (called conditional Probability)
and Probability of passing if a student does not study P(P/Sc) is 40%. What is the total
probability of passing ?

P (P) = P(S) * P(P/S) + P(Sc)* P(P/Sc)

= 70% *80% + 30%*40%
=68%

Example 2 : The earnings of HDFC Bank are interest rate sensitive, benefiting from a
declining interest rate environment. Suppose there is a 60% probability that HDFC Bank
will operate in a declining interest rate environment and a 40% probability that it will
operate in a stable interest rate environment. If a declining interest rate environment occurs,
the probability that EPS will be $ 3 is estimated at 25%, and the probability that EPS will be
USD 4 is estimated at 75%. In a stable interest rate environment P( EPS = 2) is 70% and
P(EPS= 1) is 30%. Calculate expected value of EPS and Standard Deviation.

Solution: The question can be simplified by building a probability tree diagram as below:
FinTree © 2024 FinTree Education Private Ltd

FinTree
Joint Probability
EPS = 4 60%*75% =45%
75%

60%
te=
t ra
i ng in
n
ecli
ty of d 25%
il i
b ab 60%*25% =15%
Pro EPS = 3
40%*70% =28%
Pro
ba b
ilit
EPS = 2
yo
f st
abl
e in t
70%
ra t
e=
40%

30%
40%*30% =12%
EPS = 1

Conditional Mean and variance can be calculated as follows:

FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Bayes Theorem

FinTree
• Bayes' Theorem is a mathematical formula that helps us update the probability of a hypothesis
based on new evidence.
• It's named after Reverend Thomas Bayes, who introduced the concept.

Example 1: If probability studying for the exam P(S) is 70%, the probability of not studying for
the exams P(Sc) = 30% (called complement). The probability of passing if a student studies is
P(P/S)= 80% (called conditional Probability) and Probability of passing if a student does not
study P(P/Sc) is 40%. If you get to know that the student eventually passed the exam, what is the
updated probability of studying? (Note: Updated Probability is also called Posterior Probability).

Joint Probability
PASS 70%*80% =56%
80%

0%
ng =7
dyi
of stu 20%
t y
ba bili 70%*20% =14%
Pro FAIL
Pro
bab PASS 30%*40% =12%
i lity
of
not
st u
40%
dyi
ng=
30%

60%
30%*60% =18%
FAIL

Unconditional Probability of passing = 56% + 12% = 68%

Probability Passing given Studies P(P/S) = 56%
Probability of Studying given passing = P (S/P) = 56% / 68% = 82.25%
FinTree © 2024 FinTree Education Private Ltd

Portfolio Mathematics
FinTree Fruit 1 : Portfolio Return and Standard Deviation

FinTree
• Expected return on the portfolio (E(Rp)) is a weighted average of the expected returns
• portfolio variance (two assets) is calculated as follows:

= (Waσa )2 + (Wbσb )2 + 2* Wa * Wb * σa*σb* Correlation(a,b)

= (Waσa )2 + (Wbσb )2 + 2* Wa * Wb * Covariance(a,b)

• Portfolio variance (three assets) is calculated as follows:

=(Waσa )2 + (Wbσb )2 + 2* Wa * Wb * Covariance(a,b)

(Waσa )2 + (Wcσc )2 + 2* Wa * Wc * Covariance(a,c)

(Wbσb )2 + (Wcσc )2 + 2* Wb * Wc * Covariance(b,c)

Example 1 : Calculate portfolio expected return and standard deviation based on following

• Expected Return = 40% * 10% (+) 60% * 20% = 4% + 12% = 16%

• Covariance A,B = 8% * 12% * 0.3 = 0.00288
• Portfolio Variance = (40% * 8%)2 + (60% * 12%)2 + 2* 40% *60%* 8%*12%*0.3 = 0.0075904
• Portfolio Standard Deviation = SQRT (0.0075904) = 8.71%
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : Joint Probability Function

FinTree
Example 1: Calculate covariance between A and B based on following joint probability function

Solution: Covariance can be calculated in following three steps:

STEP1: Convert the Joint Probability Distribution into a simple tabular structure:

STEP 2: Calculate Expected Value of Returns A E(A), Expected Value of Returns B E(B),
Expected Value of Returns A Returns B& E(AB) as follows:

STEP 3: Covariance can be calculated as :

= Expected Value of Returns A E(A) * Expected Value of Returns B E(B) (-) Expected Value
of Returns A Returns B& E(AB)
= E(A) * E(B) - E(AB)
= 15.40% * 10.30% (-) 1.81% = -0.2258%
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Mean Variance Analysis

FinTree
• Mean-variance analysis is the process of weighing risk, expressed as variance, against expected
return.
• Investors use mean-variance analysis to make investment decisions.
• Investors weigh how much risk they are willing to take on in exchange for different levels of
reward
• Mean–variance analysis (MVA) holds exactly when investors are risk averse
• MVA holds when : 1. Returns are normally distributed or 2. Investors have quadratic utility
functions
• Mean–variance analysis, however, can still be useful—that is, it can hold approximately—when
either assumption 1 or 2 is violated

FinTree Fruit 4 : Safety First Rule and Shortfall Ratio

Example 1: An investor wants to choose between following two portfolios, recommend a portfolio
based on Roy’s Safety First Ratio and Shortfall Risk

Solution:
STEP 1: Calculate Safety First Ratio and select the portfolio based on highest SFR.

SFR = Return on Portfolio - Minimum Acceptable Return

Standard Deviation of the Portfolio

SFRA = (20% - 5%) / 15% = 1

SFRB = (25%- 5%)/10% = 2
Since, B has a higher SFR, we select portfolio B for the investor.
FinTree © 2024 FinTree Education Private Ltd

FinTree
STEP 2 : Calculate Shortfall Risk by using normal distribution tables.
Shortfall risk is the probability of not earning a minimum of 5%.
Therefore, we will look for probability on the left of -1 and -2 for portfolio A and B respectively.

For Portfolio A (SFR=1), Shortfall risk is 0.15866 = 15.866%.

i.e., There is 15.86% probability that Portfolio A will earn less than 5% Return.

For Portfolio B (SFR=2), Shortfall risk is 0.02275 = 2.275%.

i.e., There is 2.27% probability that Portfolio B will earn less than 5% Return.
We select portfolio with lowest Shortfall Risk.

Note that higher SFR always leads to lower Shortfall Risk.

 FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5 : Stress Testing and Value at Risk

FinTree
• Two main ways of managing financial risk are value at risk (VaR) and stress testing/scenario
analysis.
• Stress testing and scenario analysis refer to a set of techniques for estimating losses in extremely
unfavorable combinations of events or scenarios.
• Value at risk (VaR) is a money measure of the minimum value of losses expected over a specified
time period (e.g., a day, a quarter, or a year) at a given level of probability (often 0.05 or 0.01)
FinTree © 2024 FinTree Education Private Ltd

Simulation Methods
FinTree Fruit 1 : Log-normal Distribution

FinTree
• Use Normal Distribution - Asset returns
• Log-normal Distribution - Asset Prices
• The Black–Scholes–Merton model assumes that the price of the asset underlying the option is
log-normally distributed.
• Log-normal Distribution - “it’s log is normal”
• The two most noteworthy observations about the log-normal distribution are that it is bounded
below by 0 and it is skewed to the right (it has a long right tail).

FinTree Fruit 2 : Continuously Compounded Rates of Return

• If a stock’s continuously compounded return is normally distributed, then future stock price is
log-normally distributed.
• Stock price may be described by the log-normal distribution even when continuously
compounded returns do not follow a normal distribution (Central Limit Theorem)
• A key assumption in many investment applications is that returns are independently and
identically distributed (i.i.d.)
• Independence captures the proposition that investors cannot predict future returns using past
returns.
• Identical distribution captures the assumption of stationarity, a property implying that the
mean and variance of return do not change from period to period.
• If the one-period continuously compounded returns are normally distributed, then the T
holding period continuously compounded return, r0,T, is also normally distributed with mean μT
and variance σ2T.
• This is because a linear combination of normal random variables is also a normal random
variable.
• Sigma Scaling rule: if one day σ = 2%, one year σ will be 2%* √250 (Assuming 250 trading days
in a year). Sigma can be scaled by multiplying with square root of time
• Expected Return (mean) Scaling rule : if one day return = 2%, one year return will be
2%* 250 (Assuming 250 trading days in a year). returns can be scaled by multiplying with time
FinTree © 2024 FinTree Education Private Ltd

FinTree
Example 1: Based on Following data, calculate:
1. Daily Historical return (Continuously Compounded)
2. Daily σ
3. Annual Expected Return (assume 250 trading days/year)
4. Annual σ (assume 250 trading days/year)

Solution:
STEP 1: Calculate Continuously Compounded daily returns by first calculating
Closing Price / Opening price and them pressing LN button on the calculator.
Day 1: LN (120/100) = 18.23%
Day 2: LN (140/120) = 15.41%
Day 3: LN (170/140) = 19.41%
Day 4: LN (200/170) = 16.25%

STEP 2: Insert daily returns in STAT function and calculate mean and Sample Standard
Deviation (Keep calculator on LIN mode)
Mean Return (daily) = 17.33%
Sample SD (daily) (Sx) =1.825%

STEP 3: Annualize returns and SD by multiplying with 250 and √250 respectively.
Expected Annual Return = 17.33% * 250 = 4332.5%
Annual SD = 1.825% * √250 = 28.86%

FinTree Fruit 3 : Monte Carlo Simulation

• Monte Carlo simulation is like rolling dice or flipping a coin many times to understand the
range of possible results.
• In finance, it helps assess the potential outcomes of an investment or financial strategy by
considering various uncertain factors.
FinTree © 2024 FinTree Education Private Ltd

FinTree
Process:
1. Identify Variables: Determine the key factors that affect your financial scenario, such as
investment returns, inflation rates, or interest rates.
2. Define Ranges: Specify the possible values or ranges each variable can take. For instance,
investment returns might range from -10% to +15%.
3. Run Simulations: Randomly generate values for each variable within their defined ranges and
calculate the financial outcome. Repeat this process multiple times (e.g., 1,000 simulations) to
see a range of possible results.
4. Analyze Results: Examine the outcomes of all simulations to understand the likelihood of
different scenarios. This provides a probability distribution of potential financial outcomes.

Strengths: Monte Carlo simulation can be used to price complex securities for which no analytic
expression (formula) is available, particularly American-style options or Complex Securities.

Weaknesses: Monte Carlo simulation provides only statistical estimates, not exact results.
Analytic methods, when available, provide more insight into cause-and-effect relationships than
does Monte Carlo simulation.

FinTree Fruit 4 : Bootstrapping

• The idea behind bootstrap is to mimic the process of performing random sampling from a
population to construct the sampling distribution
• The difference lies in the fact that we have no knowledge of what the population looks like,
except for a sample with size n drawn from the population.
• In bootstrap, we repeatedly draw samples from the original sample, and each re-sample is of
the same size as the original sample. Note that each item drawn is replaced for the next draw
(i.e., the identical element is put back into the group so that it can be drawn more than once).
• Both the bootstrap and the Monte Carlo simulation build on repetitive sampling.
• Bootstrapping re-samples a dataset as the true population, and infers from the sampling
statistical distribution parameter values (i.e., mean, variance, skewness, and kurtosis) for the
population. Monte Carlo simulation builds on generating random data with certain known
statistical distribution of parameter values.
• Bootstrap simulation is a complement to analytical methods.
• Analytical methods, where available, provide more insight into cause-and-effect relationships.
FinTree © 2024 FinTree Education Private Ltd

FinTree

CFA Curriculum, 2024, Quantitative Methods, pg no.186

 FinTree © 2024 FinTree Education Private Ltd

Estimation and Inference

FinTree Fruit 1 : Sampling Methods

FinTree
Sampling Methods

When an analyst chooses to sample, they must formulate a sampling plan.

A sampling plan is the set of rules used to select a sample.

Probability Sampling Non- probability

Sampling

• Probability sampling gives every member of • Non-probability sampling depends on

the population an equal chance of being factors other than probability
selected. considerations, such as a sampler’s
• Hence it can create a sample that is judgment or the convenience to access data.
representative of the population. • Consequently, there is a significant risk that
• Probability sampling can yield more non-probability sampling might generate a
accuracy and reliability compared with non-representative sample
non-probability sampling.

Simple Random Stratified Random Convenience Convenience

Cluster sampling Sampling
Sampling Sampling Sampling

Systematic Sampling
FinTree © 2024 FinTree Education Private Ltd

FinTree
• A simple random sample is a subset of a larger population created in such a way that each
element of the population has an equal probability of being selected to the subset.
• Simple random sampling is particularly useful when data in the population is homogeneous
• Systematic sampling can be used when we cannot code (or even identify) all the members of a
population. With systematic sampling, we select every kth member until we have a sample of
the desired size. The sample that results from this procedure should be approximately random.
• Sampling error is the difference between the observed value of a statistic and the quantity it is
intended to estimate as a result of using subsets of the population.
• Sampling distribution of a statistic is the distribution of all the distinct possible values that the
statistic can assume when computed from samples of the same size randomly drawn from the
same population.

• Stratified random sampling population is divided into subpopulations (strata)

• Simple random samples are then drawn from each stratum in sizes proportional to the relative
size of each stratum in the population
• In contrast to simple random sampling, stratified random sampling guarantees that population
subdivisions of interest are represented in the sample.
• Another advantage is that estimates of parameters produced from stratified sampling have
greater precision—that is, smaller variance or dispersion—than estimates obtained from simple
random sampling.
• Bond indexing (NIFTY 50 equivalent of bond markets) is one area in which stratified sampling
is frequently applied.

• Cluster sampling also requires the division or classification of the population into
subpopulation groups, called clusters
• Then certain clusters are chosen as a whole using simple random sampling
• If all the members in each sampled cluster are sampled, this sample plan is referred to as
one-stage cluster sampling
• If a subsample is randomly selected from each selected cluster, then the plan is referred as
two-stage cluster sampling.
FinTree © 2024 FinTree Education Private Ltd

FinTree
• A major difference between cluster and stratified random samples is that in cluster sampling, a
whole cluster is regarded as a sampling unit and only sampled clusters are included.
• In stratified random sampling, however, all the strata are included and only specific elements
within each stratum are then selected as sampling units.
• Cluster sampling is commonly used for broad market surveys, and the most popular version
identifies clusters based on geographic parameters
• Cluster sampling usually yields lower accuracy because a sample from a cluster might be less
representative of the entire population.
• Its major advantage, however, is offering the most time-efficient and cost-efficient probability
sampling plan for analyzing a vast population.

• Convenience Sampling: an element is selected from the population based on whether or not it is
accessible to a researcher or on how easy it is for a researcher to access the element.
• Samples are not necessarily representative of the entire population
• Level of sampling accuracy could be limited
• Advantage: data can be collected quickly at a low cost ( time-efficient and cost-effective)

• Judgmental Sampling: involves selectively handpicking elements from the population based on
a researcher’s knowledge and professional judgment
• Affected by the bias of the researcher
• Might lead to skewed results
• For example, when auditing financial statements, seasoned auditors can apply their sound
judgment to select accounts or transactions that can provide sufficient audit coverage
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : The Central Limit Theorem

FinTree
• A sampling distribution (drawn from any population with mean = µ, variance = σ2) will have
following 3 properties (Large Sample Size)
1. Sample Mean (X) will approach Population Mean (µ)
2. Sample Standard Deviation (Sx) will approach σ/√n
3. Sampling Distribution will approach Normal Distribution.
• When sample size n is greater than or equal to 30, we can assume that the sample mean is
approximately normally distributed

FinTree Fruit 3 : Standard Error of the Sample Mean

• If we know σ, the population standard deviation, standard error is calculated as σ/√n

• If we do not know σ, standard error is calculated as Sx/√n
• Note that although the standard error is the standard deviation of the sampling distribution
• If we want to find out how precise the estimate of a population parameter from sampled data is
relative to its true value, standard error is the statistic to use

FinTree Fruit 4 : Standard Error of the Sample Mean

• Re-sampling (Bootstrapping): we repeatedly draw samples from the original sample, and each
re-sample is of the same size as the original sample
• Each item drawn is replaced for the next draw
• It is also called it is often called model-free re-sampling or non-parametric re-sampling
• Standard Error of re-sampled distribution is calculated as follows:

Sx = Standard Error of the Sample Mean

B = the number of re-samples drawn from the original sample
θb = the mean of a re-sample
θ = the mean across all the re-sample means
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5 : Jackknife

FinTree
• Unlike bootstrap, which repeatedly draws samples with replacement, jackknife samples are
selected by taking the original observed data sample and leaving out one observation at a time
from the set (and not replacing it).
• According to its computation procedure, we can conclude that jackknife produces similar
results for every run, whereas bootstrap usually gives different results because bootstrap
re-samples are randomly drawn.
• For a sample of size n, jackknife usually requires n repetitions, whereas with bootstrap, we are
left to determine how many repetitions are appropriate.
FinTree © 2024 FinTree Education Private Ltd

Hypothesis Testing
FinTree Fruit 1 : Hypothesis Basics

FinTree
• Hypothesis testing is part of the branch of statistics known as statistical inference
• A hypothesis is a statement about one or more populations that we test using sample statistics
• Six Step standard approach to hypothesis testing is as below:

STEP 1 : State the hypotheses

Step 2: Identify the appropriate test statistic

Step 3: Specify the level of significance

Step 4: State the decision rule

Step 5: Collect data and calculate the test statistic

Step 6: Make a decision

• For each hypothesis test, we always state two hypotheses: the null hypothesis (or null),
designated H , and the alternative hypothesis, designated Ha
• The null hypothesis is what we want to reject.
• The null and alternative hypotheses are stated in terms of population parameters, and we use
sample statistics to test these hypotheses
• The null and alternative hypotheses must be mutually exclusive and collectively exhaustive; in
other words, all possible values are contained in either the null or the alternative hypothesis
 FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : Type I and Type II Error

FinTree
• If we are performing a hypothesis test at 95% level of confidence, the remaining 5% is called
significance level.
• Type I Error : False Positive, i.e. Rejecting a True Null
• Type II Error: False Negative, i.e. Failing to reject a False Null
• Type I Error Probability = Significance Level
• (1- Significance Level) = Confidence Level, therefore
• (1- Prob. of Type I Error) = Confidence Level
• (1- Prob. of Type II Error) = Power of test

Null is True Null is False

Not Rejecting
Not Rejecting - Correct
TYPE II Error
Decision

Rejecting
Rejecting -Correct
TYPE I Error Decision

• Controlling the probabilities of the two types of errors involves a trade-off.

• All else equal, if we decrease the probability of a Type I error by specifying a smaller significance
Level, we increase the probability of making a Type II error.
• The only way to reduce the probabilities of both types of errors simultaneously is
to increase the sample size, n.

• The critical value or values we choose are based on the level of significance and the probability
distribution associated with the test statistic (selected from probability tables).
• If calculates value (test statistic) > critical value, we reject the null hypothesis
• When we reject null hypothesis, we say the result is statistically significant.
• p-value: smallest level of significance at which the null hypothesis can be rejected
 FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3: Overview of Hypothesis Tests

FinTree
There are total 5 types of Hypothesis tests in this Learning Module

Mean Related Variance Related

Tests Tests

Test of a Test of a Test of a

Single Mean Differences of Mean Mean of Differences
t- Test (DOM) (MOD)
(Independent Sample) (Dependent Sample)
t- Test Paired Comparison Test
t- Test

Test of a Test of a
Single Variance Differences
Chi-Square Test in variances
(DOV)
F- Test
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 4: Test of a Single Mean

FinTree
Example 1: Suppose you are analyzing FinTree Equity Fund, During the past 24 months, it has
achieved a mean monthly return of 1.40%, with a sample standard deviation of monthly returns
of 3.80 percent. Given its level of market risk and according to a pricing model, this mutual fund
was expected to have earned a 1.20 percent mean monthly return during that time period.

Assuming returns are normally distributed, are the actual results consistent with a population
mean monthly return of 1.20 percent?

Formulate and test a hypothesis that the fund's performance was different than the mean return
of 1.1 percent inferred from the pricing model. Use a 5 percent level of significance.

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 24-1 = 23

H: μ = 1.2% (NULL hypothesis)
Look up 5% two tailed
t- distribution value
Ha: μ ≠ 1.2% (Alternate Hypothesis) = 1.4 (-) 1.2
(3.8/√24)
0.2578
= 0.2578 -2.069 +2.069
Since Test Statistic (0.2578) falls between
Critical value range of -2.069 and +2.069.
We FAIL TO REJECT null hypothesis.
Which means, Sample mean of 1.4 is not
Statistically Significantly different from 1.2
One Tailed Probabilities (Hypothesized Value)

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
23 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5: Test concerning differences between means with independent samples (DOM)

FinTree
• We often want to know whether a mean value—for example, a mean return differs for two
groups. Is an observed difference due to chance or to different underlying values for the
mean?
• We test this by drawing a sample from each group.
• Samples have to be from populations that are approximately normally distributed and that
the samples are also independent of each other.
• Our focus in discussing the test of the difference of means is using the assumption that the
population variances are equal (Pooled Variance Method)

Example 1: Suppose we want to test whether the returns of the FinTree High Yield Index, shown
below, are different for two different time periods, Period 1 and Period 2 (independent Sample)

Period 1 Period 2

Mean 1.2% 1.5%

Standard Deviation 5% 6%

Sample Size 400 days 500 days

Note that these periods are of different lengths and the samples are independent; that is,
there is no pairing of the days for the two periods.

Is there a difference between the mean daily returns in Period 1 and in Period 2,
using a 5% level of significance?
FinTree © 2024 FinTree Education Private Ltd

FinTree
Define Hypothesis

H: μperiod1 = μperiod2 (NULL hypothesis)

Ha: μperiod1 ≠ μperiod2 (Alternate hypothesis)

Calculate Test Statistics

STEP 1: Calculate Pooled Variance

= 399 * (5%)2 + 499 * (6%)2

399 + 499

= 0.003111

STEP 2: Calculate test statistic

= (1.2% -1.5%) - 0

0.003111 * (1/400 + 1/500)

= - 0.8017

Compare with Critical Values Since Test Statistic (-0.8017) falls between
Critical value range of -1.96 and +1.96,
Degrees of Freedom = 400+500-2 = 898 We FAIL TO REJECT null hypothesis.
Look up 5% two tailed We conclude that there is insufficient evidence
t- distribution value
to indicate that the returns are different for the
two time periods
-0.8017
-1.96 +1.96
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 6: Test Concerning Differences between Means with Dependent Samples (paired comparison test)

FinTree
• The test of paired comparisons is more powerful than the test of the difference in the means
(pooled variance) because by using the common element (such as the same periods or
companies), we eliminate the variation between the samples that could be caused by
something other than what we are testing.

Suppose we want to compare the returns of the FinTree High Yield Index with those of the
FinTree BBB Index. We collect data over the same 1200 days for both indexes and calculate their
means and standard deviations as shown below

FinTree HY FinTree BBB Difference

Mean 1.2% 1.5% 0.3%

This is the SD
Standard Deviation 5% 6% 4.0% of differences
of return over
sample period
(given in the q)

Using 5% level of significance , is the mean of differences is different from zero ?

FinTree © 2024 FinTree Education Private Ltd

FinTree
Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 1200-1 = 1199

H: μdifference = 0 (NULL hypothesis)
Look up 5% two tailed
t- distribution value
Ha: μdifference ≠ 0 = 0.3%
(Alternate Hypothesis)
(4%/√1200)
2.598

= 2.5980 -1.96 +1.96

Since Test Statistic (0.2578) falls between
Critical value range of -2.069 and +2.069.
We FAIL TO REJECT null hypothesis.
Which means, Sample mean of 1.4 is not
Statistically Significantly different from 1.2
(Hypothesized Value)

One Tailed Probabilities

Since we need two tailed 5%,
we will look at one tailed 2.5%
one tailed column with ∞
degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 7: Test of a Single Variance

FinTree
• Example 1: FinTree Equity Fund, a small cap growth fund that has been in existence for only
24 months. During this period, FinTree Equity achieved a mean monthly return of 2% and a
standard deviation of monthly returns of 3.50%
• Using a 5 percent level of significance, test whether the standard deviation of returns is less
than 4 percent. Recall that the standard deviation is the square root of the variance, hence a
standard deviation of 4 percent or 0.04, is a variance of 0.0016.

Solution:

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 24-1 = 23

H: δ2 ≥ 16 (NULL hypothesis)
Look up 95% one tailed (right tailed)
chi-squared distribution value
Ha: δ < 16
2 (Alternate Hypothesis)
= 23 * 3.962
This is generally a one tailed test 42
22.54
Rejection Area FT
= 22.54 13.091

Since Test Statistic (22.58) falls on right of

Critical value 13.091
We FAIL TO REJECT null hypothesis.
Which means, There is insufficient evidence
to indicate that the variance is less than 16
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 8: Test of a Single Variance

FinTree
• You are investigating whether the population variance of returns on a FinTree Broad Market t
index changed after a change in market regulation. The first 121 weeks occurred before the
regulation change, and the second 61 weeks occurred after the regulation change. The
variance before the change was 8 and variance after was 3.9.
• Are the variance of returns different before the regulation change versus after the regulation
change?

Solution:

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom numerator = 121-1 = 120

H: δbefore= δafter (NULL hypothesis) Degrees of Freedom denominator = 61-1 = 60

Ha: δbefore ≠ δafter (Alternate Hypothesis) = 8

3.9
2.05

= 2.05 FT
1.47 Rejection Area

Since Test Statistic (2.05) falls on right of

Critical value 1.47
We REJECT null hypothesis.
Which means, There is sufficient evidence
to indicate that the weekly variances of
returns before the regulation change
are greater than the variances
after the regulation change.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 9: Parametric vs Non-Parametric Tests

FinTree
• Parametric Test (all tests done so far) have two important characteristics:

1. They are concerned with parameters (mean, variance)

2. They make assumptions about the distribution of the population producing the sample
(normal distribution, chi-squared, F etc)

• Nonparametric procedures are primarily used in four situations

1. When the data do not meet distributional assumptions,
2. When there are outliers, for example, we may want to use a nonparametric test of the
median, in the case of outliers, instead of a test of the mean.
3. When the data are given in ranks or use an ordinal scale, or
4. When the relevant hypotheses do not concern a parameter, for example, if the question
concerns whether a sample is random or not, we use the appropriate nonparametric test (a
“runs test”). The nonparametric runs test is used to test whether stock price changes can be
used to forecast future stock price changes—in other words, a test of the random walk
theory. Another type of question that nonparametric methods can address is whether a
sample came from a population following a particular probability distribution.

• The nonparametric test will frequently involve the conversion of observations (or a function of
observations) into ranks according to magnitude, and sometimes it will involve working with
only “greater than” or “less than” relationships (using the + and − signs to denote those
relationships).
• One must refer to specialized statistical tables to determine the rejection points of the test
statistic.
• If the assumptions of the parametric test are met, the parametric test (where available) is
generally preferred over the nonparametric test because the parametric test may have more
power, that is, a greater ability to reject a false null hypothesis.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 10: Important Non- Parametric Tests

FinTree
Tests Parametric Tests Non-Parametric Tests

Tests concerning a single

t-Test / z- Test Wilcoxon signed-rank test
mean

Tests concerning differences Mann–Whitney U test

t-Test
between means (Wilcoxon rank sum test)

Tests concerning mean

Wilcoxon signed-rank test
differences (paired t-Test
Sign test
comparisons tests)
FinTree © 2024 FinTree Education Private Ltd

Parametric and Non-Parametric Tests of Independence

FinTree Fruit 1 : Correlation Hypothesis Test

FinTree
• The parametric pairwise correlation coefficient is often referred to as Pearson correlation, the
bivariate correlation, or simply the correlation.
• Correlation is calculated as CovarianceXY/VarianceX
• Positive Covariance → Positive Correlation → Positive Slope Coefficient (beta)
• If the two variables are normally distributed, we can test to determine whether the null
hypothesis (H0: ρ = 0) should be rejected using the sample correlation, r. The formula for the
t-test is as follows:

• This test statistic is t-distributed with n − 2 degrees of freedom

• Notice that the magnitude of r needed to reject the null hypothesis decreases as sample size n
increases, for two reasons:
1. First, as n increases, the number of degrees of freedom increases and the absolute value of
the critical value of the t-statistic decreases (remember rejection happens when t stat is
greater than critical value)
2. Second, the absolute value of the numerator increases with larger n, resulting in a larger
magnitude of the calculated t-statistic (therefore t- stat will be greater than critical value)
• Notice in Example 1 vs Example 2, when sample size increases from 12 to 32, the sample
statistic increases and critical value decreases, which results into decision changing from fail
to reject (n=12) to reject (n= 32).

• As the sample sizes increase as ever-larger datasets are examined, the null hypothesis is
almost always rejected (power of test increases) and other tools of data analysis must be
applied.
FinTree © 2024 FinTree Education Private Ltd

FinTree
Example 1: Correlation between two variables is 0.5, perform a two tailed hypothesis test
assuming sample size of a) 12

Solution: a) n = 12

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 12-2 = 10

H0: ρ = 0 (NULL hypothesis)

Ha: ρ ≠ 0 (Alternate Hypothesis) = 0.5*√10

1-(0.5)2
1.8257

= 1.8257 -2.228 +2.228

Since Test Statistic (1,8257) falls between

-2.228 and + 2.228,
We FAIL TO REJECT null hypothesis.
Which Means, the correlation of 0.5 is not
significantly different than zero at 5%
level of significance.

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
10 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree
Example 2: Correlation between two variables is 0.5, perform a two tailed hypothesis test
assuming sample size of b) 32

Solution: a) n = 12

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 12-2 = 10

H0: ρ = 0 (NULL hypothesis)

Ha: ρ ≠ 0 (Alternate Hypothesis) = 0.5*√30

1-(0.5)2 3.1

= 3.1 -2.04 +2.04

Since Test Statistic (3.01) falls outside

-2.04 and + 2.04
We REJECT null hypothesis.
Which Means, the correlation of 0.5 is
significantly different from zero at 5%
level of significance.

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
30 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 2 : Spearman Rank Correlation

FinTree
• When we believe that the population departs from normality, we can use a test based on the
Spearman rank correlation coefficient.
• It is calculated based on ranks of the variables
• STEPS to calculate Spearman rank correlation:
1. Rank the observations on X from largest to smallest. Assign the number 1 to the observation
with the largest value, the number 2 to the observation with second largest value, and so on.
In case of ties, assign to each tied observation the average of the ranks that they jointly
occupy. For example, if the third and fourth largest values are tied, we assign both observa-
tions the rank of 3.5 (the average of 3 and 4).
2. Perform the same procedure for the observations on Y.
3. Calculate the difference, di , between the ranks for each pair of observations, then calculate
di2 (Square the differences in ranks).
4. Calculate Spearman rank correlation as follows

Example 1: Calculate Spearman Rank Correlation based on following data:

FinTree © 2024 FinTree Education Private Ltd

FinTree
Solution:Spearman Rank Correlation can be calculated in following steps:

STEP 1: Rank returns of fund 1 and fund 2.

STEP 2: Calculate difference of the ranks and square them.

STEP 2: Calculate Spearman Rank Correlation using following formula:

6 * 30
=1-
7*(72 - 1)

= 0.4642
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Hypothesis Testing of Spearman Rank Correlation

FinTree
The Hypothesis Test of Spearman
Rank Correlation Depends on
Sample Size.

Sample Size less than 30. Sample Size 30 or more

Specialized critical value tables t- test discussed earlier, same as
required. Pearson’s Correlation.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 4 : Contingency Table / Two way Table

FinTree
• When faced with categorical or discrete data, we cannot use the methods that we have
discussed up to this point to test whether the classifications of such data are independent.
• When classification of the data type is discrete, so we cannot use correlation to assess the
relationship between two variables.
• For example, if we have 50 fund categorized based on size (Small, Mid and Large Caps) and
style (Value, Growth), the contingency table will look as follows:

Small Mid Large TOTAL

Value 5 5 20 30

Growth 10 6 4 20

TOTAL 15 11 24 50

FinTree Fruit 5 : Hypothesis Test of Independence

• If we want to test whether a relationship exists between the size and investment type, we can
perform a test of independence using a nonparametric test statistic that is chi-square
distributed:

• m = the number of cells in the table, which is the number of groups in the first class
multiplied by the number of groups in the second class;
• Oij = the number of observations in each cell of row i and column j (i.e., observed
frequency); and
• Eij = the expected number of observations in each cell of row i and column j, assuming
independence (i.e., expected frequency).
• Degrees of freedom = (r − 1)(c − 1), where r is the number of rows and c is the number of
columns.
FinTree © 2024 FinTree Education Private Ltd

FinTree
Example 1: Perform the test of independence using contingency table created in FF 4 for 50
funds.

Solution: The test can be performed using following steps:

STEP 1: Calculate m as multiplication of number of rows and columns. As we have 2 rows and
3 columns , m = 2*3 = 6

STEP 2 : Calculate expected value for each cell using Total of Row* Total Column
Total of Table

We have total 6 cells , so we will have to perform 6 calculation as follows:

Expected Value for this cell (Value,Small) = (30*15)/ 50 = 9

In the same way, expected value for Value, Mid = (30*11)/50 = 6.6
Expected Value for Value, Large = (30*24)/50 = 14.4

Expected Value for Growth, Small = (20*15)/50 = 6

Expected Value for Growth, Mid = (20*11)/50 = 4.4
Expected Value of Growth, Large = (20*24)/50 = 9.6

STEP 3: Recreate contingency table using Expected Frequencies as follows:

Small Mid Large TOTAL

Value 9 6.6 14.4 30

Growth 6 4.4 9.6 20

TOTAL 15 11 24 50
FinTree © 2024 FinTree Education Private Ltd

FinTree
STEP 4: For each cell, calculate (Actual frequency - Expected frequency)2
Expected Frequency

Since we have 6 cells, we will have to perform 6 calculations as follows

Value, Small Cell = (5-9)2 / 9 = 1.78

Value, Mid Cell = (5-6.6)2 / 6.6 = 0.39
Value, Large Cell = (20-14.4)2 / 14.4 = 2.18

Growth, Small Cell = (10-6)2 / 6 = 2.67

Growth, Mid Cell = (4-4.4)2 / 4.4 = 0.036
Growth, Large Cell = (4-9.6)2 / 9.6 = 3.27

STEP 5: Recreate contingency table using scaled squared deviations as follows:

Small Mid Large

Value 1.78 0.39 2.18

Growth 2.67 0.036 3.27

STEP 6: Test Statistic (chi-Squared) is calculated as total of these scaled squared deviations
which will be 1.78 + 0.39 + 2.18 + 2.67 + 0.036 + 3.27 = 10.326

STEP 7: Degrees of Freedom are calculated as (no. of columns -1) * (no. of rows -1)
= (3-1)*(2-1) = 2

STEP 8: Perform the hypothesis Test at 5% Level of Significance

FinTree © 2024 FinTree Education Private Ltd

FinTree
Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 2 (STEP 7)

H0: Fund size and Investment type

are not related, so these classifications

are independent
10.3

Ha: Fund size and investment type are Fail to Reject Rejection Area
= 10.326 (STEP 6) 5.99
related, so these classifications

are not independent. Since Test Statistic (10.32) falls on the right of
Critical value of 5.99,
We REJECT null hypothesis.
Which Means, the fund size and investment
type are not independent.

Notice the difference in how we read the table for two Chi Squared tests we have learnt so far.
When we tested for Variance in one of the earlier learning module, we did a left tailed test.
Therefore for 5% level of significance, we used 95% column.

However, this test is designed to be a right tailed test. Therefore for 5% level of significance,
we will use 5% Column.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 6 : Mosaic Graphic

FinTree
• We can visualize the contingency table in a graphic referred to as a mosaic.
• In a mosaic, a grid reflects the comparison between the observed and expected frequencies.

CFA Curriculum, Volume 1 , pg no. 254

• We can visualize the contingency table in a graphic referred to as a mosaic.

• In a mosaic, a grid reflects the comparison between the observed and expected frequencies.
• The width of the rectangles in reflect the proportion of funds that are small, medium, and
large, whereas the height reflects the proportion that are value, growth, and blend
• The darker shading indicates whether the number of observations is more than expected
under the null hypothesis of independence, whereas the lighter shading indicates that the
number of observations is less than expected, with “more than” and “less than” determined
by reference to the standardized residual boxes
• The standardized residual, also referred to as a Pearson residual, is as follows:
FinTree © 2024 FinTree Education Private Ltd

Simple Linear Regression

FinTree Fruit 1 : Basics of Regression

FinTree
Basics of Regression:
• Why: Regression helps predict and understand relationships between variables.
• How: It makes a math model showing the connection between them.
• Use: Used in many fields for predictions and decision-making
• Dependent Variable: Variable you are seeking to explain (y Variable)
• Independent Variable: Variable you are using to explain changes in the dependent variable
(x variable) Error
Term
• Simple Linear Regression looks like : y = a + b*x1 + ε
Independent
Variable
Dependent Slope
Variable Intercept Coefficient

FinTree Fruit 2 : Regression Terms

• The variation of Y is often referred to as the sum of squares total (SST), or the total sum of
squares.

• The variation of X is calculated as follows:

• The goal is to fit a line to the observations on Y and X to minimize the squared deviations
from the line; this is the least squares criterion, hence, the name least squares regression.
• Because of its common use, linear regression is often referred to as ordinary least squares
(OLS) regression.
• Slope Coefficient (b) = Covariancex,y / Variancex
• Alternate Formula Slope Coefficient (b) = Correlationx,y * (SDx / SDy)
• The intercept is the value of the dependent variable if the value of the independent variable
is zero
• The slope is the change in the dependent variable for a one-unit change in the independent
variable
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Cross-Sectional versus Time-Series Regressions

FinTree
• A cross-sectional regression involves many observations of X and Y for the same time period
• Time-series data use many observations from different time periods for the same company,
asset class etc

FinTree Fruit 4: Assumptions of the simple linear regression model

1. Linearity: The relationship between the dependent variable, Y, and the independent variable,
X, is linear.
If the relationship between the independent and dependent variables is nonlinear in the
parameters, estimating that relation with a simple linear regression model will produce invalid
results: The model will be biased, because it will under and overestimate the dependent
variable
• Another implication of this assumption
is that the independent variable, X, must
not be random; that is, it is
non-stochastic. (However residuals
should be random)
• If the independent variable is random,
there would be no linear relation
between the dependent and independent
variables
CFA Curriculum, Volume 1, pg no. 274

2. Homoskedasticity: The variance of the regression residuals is the same for all Observations.
If the residuals are not homoskedastic, that is, if the variance of residuals differs across obser-
vations, then we refer to this as heteroskedasticity.
3. Independence: The observations, pairs of Ys and Xs, are independent of one another (read
observation as “error terms”).
This implies the regression residuals are uncorrelated across observations.
4. Normality: The regression residuals are normally distributed. For large sample sizes, we may
be able to drop the assumption of normality by appealing to the central limit theorem.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5 : ANOVA Table

FinTree
Sources of Sum of Degrees Mean
F Value
Variation Squares Of Freedom Square

Number of Independent
SSR/ k MSR/MSE
Explained Variables (k)
∑(Predicted Y- Mean Y) 2 Called as Mean Sum of F value is used for
Sum of squares regression (SSR) (1 for simple linear
Regression (MSR) Hypothesis Testing
Regression)

SSE/ n-k1
Error
∑(Actual Y- Predicted Y)2 n-k-1 Called as Mean Sum of
Sum of squares error (SSE)
Error (MSE)

Total ∑(Actual Y- MeanY)2

n-1
Sum of Squared Total (SST) SST = SSR + SSE

• Standard Error of Estimate (se) is also known as also known as the standard error of the regression
or the root mean square error.
• The se is a measure of the distance between the observed values of the d dent variable and those
predicted from the estimated regression, the smaller the se, the better the fit of the model.
• The se, along with the coefficient of determination and the F-statistic, is a measure of the goodness
of the fit of the estimated regression line.
• Unlike the coefficient of determination and the F-statistic, which are relative measures of fit,
the standard error of the estimate is an absolute measure of the distance of the observed dependent
variable from the regression line
• Se = √ MSE

FinTree Fruit 6 : Coefficient of Determination (R2) and F Test

• The coefficient of determination (R2), also referred to as the R-squared, is the percentage of
the variation of the dependent variable that is explained by the independent variable:
• R2 = Explained Variation/ Total Variation
• For Simple Linear Regression, it can also be calculated a square of correlation Coefficient.
• To see if our regression model is be statistically meaningful, we will need to construct an
F-distributed test statistic.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 7 : Hypothesis Test for Slope Coefficient and Intercept

FinTree
Example 1: An analyst regressed returns on a stock with returns on the index with resulting
regression equation Returns on stock = 5% + 1.2*Returns on Index + ε.

The standard error of the slope coefficient is 0.2520. Sample Size is 25. Perform a test of
significance at 95% Level of Confidence.

Solution:

Define Hypothesis Calculate Test Statistics Compare with Critical Values

t stat =
Degrees of Freedom = 25-2 = 23
H0: Slope (b1) = 0 (NULL hypothesis) (Slope Coeff. - Hypothesized Value)/ S.E.

Ha: Slope (b1) ≠ 0 = 1.2

(Alternate Hypothesis)
0.2520
4.76

= 4.76 -2.06 2.06

Since Test Statistic (4.76) falls outside the range

of -2.06 and +2.06,
We REJECT null hypothesis.
Which means, slope coefficient is significantly
different from ZERO at 5% significance.

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
23 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree
• Hypothesis Test for the intercept is done in the same way. We just replace standard error of
slope coefficient with standard error of the intercept.
• An interesting feature of simple linear regression is, that is slope coefficient between two
variable x & y, is statistically significant, we can also conclude that their correlation coefficient
is also statistically significant. Test Statistic of Slope Coefficient = Test Stat. of Correlation
Coefficient.
• Another interesting feature of simple linear regression is that the test-statistic used to test the
fit of the model (i.e., the F-distributed test statistic) is related to the calculated t-statistic used
to test whether the slope coefficient is equal to zero: t2 = F.

FinTree Fruit 7 : Indicator Variable / Dummy Variable

• Indicator variable, or dummy variable, that takes on only the values 0 or 1 as the indepen-
dent variable.
• We perform hypothesis testing in the same manner as if the independent variable were a
continuous variable

FinTree Fruit 8 : Level of Significance and p-Values

• The choice of significance level in hypothesis testing is always a matter of judgment.

• Level of Significance = Probability of Type I Error (False Positive)
• decreasing the level of significance from 0.05 to 0.01 decreases the probability of a Type I
error, but it also increases the probability of a Type II error—failing to reject the null
hypothesis when, in fact, it is false (i.e., a false negative).
• The p-value is the smallest level of significance at which the null hypothesis can be rejected.
• The smaller the p-value, the smaller the chance of making a Type I error
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 9 : Prediction Using Simple Linear Regression and Prediction Intervals

FinTree
Example 1: Let’s assume we have a regression equation as y = 10 + 2*x + ε, assume that value of
x= 3, calculate value of y.
Y = 10 + 2*3 = 16

Example 2: Let’s assume, that sample size is 22, and standard error of forecast is 3. Build a 95%
prediction interval.
Degrees of Freedom = 22-2 = 20
Critical Value from the t-distribution will be 2.086
Margin of Error will be calculated as (tc * Standard Error of forecast) = 2.086*3 = 6.258
Subtracting and adding margin of error from forecasted value: 16 ± 6.258 = 9.742 to 22.258
Interpretation: There is 95% probability that true value of Y variable (given x= 3) will be
between 9.742 to 22.258

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
20 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 10 :Comprehensive Example

FinTree
Example 1: Based on the data below
a) Build equation for simple Linear Regression
b) Generate ANOVA Table
c) Calculate R-Squared and Standard Error of Regression (Standard Error of Estimate SEE)
d) Perform two tailed hypothesis test on slope coefficient @ 95% CL
e) Perform one tailed hypothesis test on slope coefficient with hypothesized value of 1 @95%
CL
f) Perform F test @ 95% CL
g) Predict value of y , assuming forecasted value of x is 9.
h) Calculated standard error of the forecast
i) Build 95% Prediction interval around predicted interval value of y.

Solution a) Insert data in data function of the calculator (2nd 6) and go to stat function (2nd 7),
make sure the calculator is on the LIN mode (if not, press 2nd set) and look for a and b value.
a is intercept and b is slope coefficient. The regression equation should be:

Y = 2.18 + 3.05 * x + ε

b) ANOVA Table
FinTree © 2024 FinTree Education Private Ltd

FinTree
c) R Squared = Regression Sum of Squares/ Total Sum of Squares
= 159.63 / 162.8 = 98.06%

Also, Note that correlation coefficient between X and Y variable is 0.9902, the square of
correlation 0.99022 also produced R2 value of 98.06%.

Standard Error of Regression (Standard Error of Estimate SEE) is calculated as square root of
Mean Sum of Square which is 1.054 as per ANOVA Table.
Therefore, SEE = √1.054 = 1.026

d)

Define Hypothesis Calculate Test Statistics Compare with Critical Values

Degrees of Freedom = 5-2 = 3

t stat =

H0: Slope (b1) = 0 (NULL hypothesis) (Slope Coeff. - Hypothesized Value)/ S.E.

Ha: Slope (b1) ≠ 0 = 3.04

(Alternate Hypothesis)
0.2475
12.30

= 12.30 -3.182 3.182

Since Test Statistic (12.30 falls outside the range

of -3.182 and +3.182,
We REJECT null hypothesis.
Which means, slope coefficient is significantly
different from ZERO at 5% significance.

Since we need two tailed 5%,

we will look at one tailed 2.5%
one tailed column with
3 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree
e)

Define Hypothesis Calculate Test Statistics Compare with Critical Values

t stat =
Degrees of Freedom = 25-2 = 23
H0: Slope (b1) ≤ 1(NULL hypothesis) (Slope Coeff. - Hypothesized Value)/ S.E.

Ha: Slope (b1) > 1 = 3.04 - 1

(Alternate Hypothesis)
0.2520
(Think of greater than sign in alternate 8.26
hypothesis >, like an arrow pointing Rejection Area

= 8.26 2.03
towards right(→), it indicates right tailed test.
Rejection will be on the right hand side of the Since Test Statistic (8.26) falls on the right of 2.03.
We REJECT null hypothesis.
Critical Value) Which means, slope coefficient is greater
than 1 at 5% significance.

Since we need one tailed 5%,

we will look at one tailed 5%
column with
3 degrees of freedom.
FinTree © 2024 FinTree Education Private Ltd

FinTree
f)

Define Hypothesis Calculate Test Statistics Compare with Critical Values

F stat =
Degrees of Freedom Numerator = 1
H0: Slope (b1) = 0 (NULL hypothesis) MSR/ MSE Degrees of Freedom Denominator = 3

Ha: Slope (b1) ≠ 0 = 159.63

(Alternate Hypothesis)
1.05
151.42
Rejection Area

= 151.42 10.1

Since Test Statistic (151.42) falls on the right of

Notice, F stat of 151.42 can also
F critical Value of 10.1,
be calculated as square of t stat We REJECT null hypothesis.
Which means, slope coefficient is significantly
of 12.30. 12.302 = 151.42 different from ZERO at 5% significance.

Since we need one tailed 5%, we will look at 5% F Table with Degrees of
Freedom of Numerator (column headings df1) of 1 and Degrees of Freedom
of Denominator of 3 (Row Headings df2).
FinTree © 2024 FinTree Education Private Ltd

FinTree
g) Predicted Value of Y, assuming forecasted value of x is 9.

Regression Equation: Y = 2.18 + 3.05 * x + ε

Inserting 9 for x, Y = 2.18 + 3.05 * 9 = 29.60

h) Standard Error of the Forecast

to compute Standard Error of forecast, we will use following formula:

Standard Error of Estimate was calculated in sub part c to be 1.026.

Sample Size is 5
Mean of Independent variable (x) is 4.6
Forecasted Value of x is 9, therefore (x-mean of x)2 = (9 -4.6)2 = 19.36
Summation of (x-mean of x)2 is calculated as follows:

Therefore, Standard Error of Forecast is

1.026 * 1 + 1 + 19.36 = 1.5658

5

17.20
FinTree © 2024 FinTree Education Private Ltd

FinTree
i) 95% Prediction interval
Forecasted value of y =29.60 (subpart g)
Standard Error of forecast is 1.5658
t critical value for 5% Significance, with 5-2 = 3 Degrees of Freedom is 3.182 (subpart d)

Therefore , margin of error will be 3.182 * 1.5658 = 4.98

Subtracting and adding margin of error from forecasted value: 29.60 ± 4.98 = 24.61 to 34.58
Interpretation: There is 95% probability that true value of Y variable (given x= 9) will be
between 24.61 to 34.58.

FinTree Fruit 11 : Functional forms for simple linear regression

• Financial and economic data can exhibit complex relationships that may not align perfectly
with a linear model.
• To address this, we often explore modifications to the variables.
• These adjustments aim to enhance the model's ability to capture and represent the
underlying patterns in the data accurately.
• three often-used functional forms, each of which involves log transformation as follows:

Log-Lin model Lin-Log model Log-Log model

dependent variable is in independent variable is in both variables are in
logarithmic form as follows: logarithmic form as follows: logarithmic form as follows:

ln Yi = b0 + b1*Xi Yi = b0 + b1*lnXi lnYi = b0 + b1*lnXi

• model is useful in
calculating elasticities
FinTree © 2024 FinTree Education Private Ltd

Introduction to Big Data Techniques

FinTree Fruit 1 : FinTech

FinTree
FinTree Fruit 2 : Big Data
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 3 : Artificial Intelligence

FinTree Fruit 4 : Machine

FinTree
Learning
FinTree © 2024 FinTree Education Private Ltd

FinTree Fruit 5 : Data Science and Data Visualization

FinTree
FinTree Fruit 6 : Text Analytics, NLP & Algorithmic Trading