Combinepdf

MA 592: Mathematical Finance
Lecture 31
Prof. Siddhartha Pratim Chakrabarty

Department of Mathematics
Indian Institute of Technology Guwahati
Banking in 21st Century: Commercial and Investment Banking
1 Commercial banking, both retail and wholesale, primarily involves deposit
and credit activities.
2 Investment banking operations are more diverse in nature and includes
assisting in raising capital (through debt and equity), advisory services on
merger & acquisitions and restructuring, as well as other corporate
financial decisions.
3 Three key operational aspects of banking, in the context of risk
management are regulations, nature of risks and capital requirements.
Banking Regulations in United States: Some Numbers
1 Between 1984 and 2014, the number of banks reduced from 14, 483 to
5, 809.
2 Between 1984 and 2014, the percentage share of of assets of small banks
(less than $100 million) fell from 16.1% to 0.8%, but that of large banks
(more than $10 billion) rose from 34.5% to 82.9%.
3 In 2014, only 91 banks accounted for more than 80% of assets.
Banking Regulations in United States: A Brief History
1 Restrictions on interstate banking through the McFadden Act. of 1927,
amended in 1933.
2 Restrictions on interstate banking reduced by way of free entry and
reciprocal arrangement after 1970.
3 Full interstate banking through the Riegel-Neal Interstate Banking and
Branching Act. of 1994.
4 The credit crisis of 2007, led to more streamlines regulations through the
Dodd-Frank Wall Street Reform and Consumer Protection Act. of 2010.
Banking Regulations in United States: Deposit Insurance
A guarantee program introduced by banking regulators to maintain public
confidence in the banks.
1 The Federal Insurance Deposit Corporation (FDIC) was created in 1933 to
provide protection for depositors, a decision triggered by the failure of
10, 000 banks during 1930 to 1933.
2 Maximum level of protection increased from $2, 500 in 1933 to $2, 50, 000
in 2008 for per depositor, per bank.
3 The premium paid by the banks to FDIC could range from being less than
0.1% to being more than 0.35%, contingent on how safe the bank is
considered to be by the regulators.
Banking Regulations in United States: The 2008 Crisis
1 Today the three types of risk faced by banks include credit risk, market
risk and operational risk.
2 As of 2007, the five largest investment banks in the United States,
namely, Goldman Sachs (GS), Morgan Stanley (MS), Merrill Lynch (ML),
Bear Sterns (BS) and Lehman Brothers (LB), had little or no commercial
banking interests.
3 After the 2008 financial crisis, an era of investment banking came to an
end in the United States:
(A) LB went bankrupt.
(B) BS was acquired by J.P. Morgan Chase.
(C) ML was acquired by Bank of America.
(D) GS and MS today are banks with both commercial and investment
banking interests.
The Credit Crisis of 2008: The Origins
1 2002 − 2005 Low interest rates led to a bubble, driven by mortgage

lending practices.
2 2000 − 2006 Huge spurt in significantly risky subprime mortgage lending.
3 The key problems were teaser rates, prepayment penalties on subprime
loans and predatory lending.
4 In the second half of 2006, the housing market went into a slump,
resulting from decline in demand as well as foreclosures after the teaser
period.
5 At the heart of the crisis, were the Asset Backed Securities (ABS), where
subprime loans were bundled through a Special Purpose Vehicle (SPV)
and were then sold in tranches (Senior, Mezzanine and Equity).
6 ABS Collateralized Debt Obligation (CDO) were created, which further
aggravated the problem.
The Credit Crisis of 2008: The Consequences
1 Investors in tranches of ABS and ABS CDO suffered huge losses.
2 The mortgage originators faced losses and litigation.
3 Investors shied away from credit markets and instead invested in treasury
instruments.
4 Credit spreads increased and obtaining credit was getting difficult.
5 The downgrading of the ratings for ABS and ABS CDO made them
illiquid.
The Credit Crisis of 2008: What Went Wrong?
1 Mortgage originators resorted to compromised lending standards, and then
transferred the credit risk to the investors.
2 Rating agencies started rating structured products without having the
requisite experience, accompanied with the limited availability of data on
these products.
3 The products were highly complex, with the stakeholders, namely, the
rating agencies and the investors being provided with incomplete and often
inaccurate information about the credit quality of the underlying assets.
4 Reliance of the investors solely on the rating agencies, instead of doing
their due diligence.
5 Oversight in recognizing the fact that these structured products with AAA
ratings offered higher returns than AAA rated bonds.
Lecture 32

Banking Regulations (Pre-1988)
1 Before 1988, the definition of capital and level of diligence in enforcement
across nations was not uniform.
2 Banks of counties with lax regulations had advantage in global market.
3 More complicated transactions for banks resulted in great risks for banks.
4 Basel Committee was formed in 1974: Belgium, Canada, France, Germany,
Italy, Japan, Luxembourg, Holland, Sweden, Switzerland, UK and USA.
5 Regular meetings at Basel, Switzerland, resulted in The 1988 BIS Accord
(Basel-I).
The 1988 BIS Accord
First attempt at setting an international risk-based standard for capital
adequacy, based on the bank’s total credit exposure, in terms of bank’s total
“risk-weighted assets”. The credit risk exposure can be divided into three
categories as follows:
1 Arising from “on-balance-sheet” assets (excluding derivatives).
2 Arising from “off-balance-sheet” assets (excluding derivatives).
3 Arising from OTC derivatives.
The 1988 BIS Accord: The Categories
1 Arising from “on-balance-sheet” assets (excluding derivatives), where each
“on-balance-sheet” asset was assigned a risk weight reflecting its credit
risk (0% for cash, gold bullion, claims on OECD government, 20% for
claims on OECD banks and OECD public sector entities, 20% for
uninsured residential mortgages, 100% for corporate bonds, claims on
LDC debt and claims on non-OECD banks).
2 Arising from “off-balance-sheet” assets (excluding derivatives), which
includes bank’s acceptance, guarantees and loan commitments. The credit
equivalent amount is calculated by applying a conversion factor to the
principal amount of the interest.
3 Arising from OTC derivatives, such as interest rate swaps or a forward
contract, where the credit equivalent amount is calculated as
max (V , 0) + aL, where V , a and L are the current value of the derivative
to the bank, an add-on-factor and principal amount, respectively.
The 1988 BIS Accord: The Categories (Contd ...)
1 The credit equivalent of Categories 2 and 3, is multiplied by the risk
weight of the counterparty, in order to calculate the risk-weighted assets.
2 For N “on-balance-sheet” items and M “off-balance-sheet”/“OTC
derivative” items, the total risk weighted assets for the bank is:
N
X M
X
wi Li + wj∗ Cj .
i=1 j=1
3 Capital set equal to be at least 8% of the risk-weighted assets.

4 But there was no netting agreement.
5 Accordingly, netting was brought in 1995.
The 1996 Amendment
1 Outlined a standardized approach for measuring the capital charge for
market risk and assigned capital separately to debt securities, equity
securities, FX risk, commodities risk, and options.
2 Large banks preferred the Internal Ratings Based (IRB) approach, which
used Value-at-Risk (VaR) for 10-day horizon, at 99% confidence level.
3 The risk weighted average for market risk capital is given by:
12.5 × max (VaRt−1 , mc × VaRavg ) + SRC .
Here VaRt−1 is the previous day’s VaR, mc ≥ 3 is the multiplicative

factor, VaRavg is the average VaR over the past 60 days and SRC is the
Specific Risk Charge.
4 Total Capital after the 1996 Amendment:
0.08 × [Credit Risk RWA + Market Risk RWA] .

Basel-II: The Three Pillars
Basel-II, which was implemented in 2007, for internationally active banks in US
and all banks and security companies in EU, was driven by three pillars:
1 Minimum Capital Requirements set at:
0.08 × [Credit Risk RWA + Market Risk RWA] .
2 Supervisory review and consequent intervention.

3 Market discipline and consequent disclosure of more information.
Credit Risk Capital Basel-II: The Three Approaches
1 Standardized Approach: Similar to Basel-I, except for the calculation of
risk-weights.
2 Foundation Internal Ratings Based (IRB) Approach: Regulators based the
capital requirements on the VaR calculated using a 1 year horizon at
99.9% confidence level. Recognizing that the expected losses are covered
by the price of the product, the capital requirements was based on
“VaR-minus-Expected Loss”.
3 Mathematically, this VaR driven capital requirement is given by:
X
EADi × LGDi × (WCDRi − PDi ) ,
i
where EAD: Exposure at Default, LGD: Loss Given Default, WCDR:

Worst Case Default Rate and PD: Probability of Default.
4 Advanced IRB Approach: Banks provide their own EAD, LGD and PD.
Market and Operational Risk Capital Basel-II
1 For Market Risk, a VaR based approach similar to Credit Risk is used.
2 For Operational Risk, three approaches are applicable:
(A) The Basic Indicator Approach:
0.15 × Bank’s Average Annual Gross Income over last three years.
(B) The Standardized Approach:
Similar to the Basic Indicator Approach, except different factors to
different business lines.
(C) The Advanced Measurement Approach:
Internal models like IRB.
Key Changes in Basel II.5
The credit crisis led to the recognition of necessity of calculation of capital for
market risk:
1 Calculation of Stressed VaR:
max (VaRt−1 , mc × VaRavg ) + max (sVaRt−1 , ms × sVaRavg ) .
Here VaRt−1 and sVaRt−1 are the VaR and Stressed VaR (10-day 99%),
and VaRavg and sVaRavg are the average of VaR and Stressed VaR (10-day
99%) over the previous 60 days. mc , ms ≥ 3 are Multiplicative Factors.
2 Incremental Default Risk Charge: Capital Requirements making use of the
maximum of capital using “Trading Book Calculation” and “Banking
Book Calculation”.
3 The Comprehensive Risk Measure: Designed to take account of risks in
what is known as correlation book and is applicable to instruments such
as ABS and CDO, which are sensitive to the correlation between the
default risks of different assets.
Basel III: The Highlights
Basel-III regulations are currently in force, which encompasses capital
requirement for Credit Risk and Market Risk and regulates six points:
1 Capital Definitions and Requirements.
2 Capital Conservation Buffer.
3 Countercyclical Buffer.
4 Leverage Ratio.
5 Liquidity Risk.
6 Counterparty Credit Risk.
Credit Risk
1 Credit Risk arises from a default, the most common examples of which are
a lender failing to repay the loan to a bank or a bond issuer failing to
make the coupon and/or face value payment to the bond holder.
2 What is referred to as Non-Performing Asset (NPA) is a consequence of
default and qualifies under Credit Risk.
3 The first basic model for Credit Risk is the Firm-Based models, which are
modelled in the classical Black-Scholes-Merton framework where the
assets of an entity are composed of equities and debts. The advantage of
this approach is that one can adopt the results for the
Black-Scholes-Merton framework, for which the Nobel Prize in Economics
was awarded in 1997.
4 The other model is the Intensity models, which are driven by stochastic
processes, such as homogeneous and in-homogeneous Poisson process, as
well as the Gamma Ornstein-Uhlenbeck process.
WCDR and PD
1 The “Worst-Case-Default-Rate” at 99.9% confidence level, for the i-th
counterparty is given by:
" √ #
N −1 (PDi ) + ρN −1 (0.999)
WCDRi = N √ .
1−ρ
2 The default probabilities can be estimated using several approaches, such

as credit ratings, historical default probabilities, recovery rates, credit
derivatives etc.
3 Let h(t) denote the “hazard rate” at time t, that is, λ(t)∆t is the
probability of default between time t and t + ∆t, conditional on the
number of defaults between time 0 and t.
4 Then the probability of default by time t is given by:
Rt
− λ(s)ds
PD(t) = 1 − e 0 .
Market Risk
1 The path-breaking approach to the determination of Market Risk is the
Value-at-Risk (VaR), which provides an estimate of the maximum loss in
a specified best outcome (in percentage) over a specified future time
horizon.
2 If we say that the 10-day 99% VaR is an amount X , it means that 99%
chances are that the loss to the portfolio over the next 10 days, will not
exceed an amount X and it is 1% likely that it will exceed an amount of
X.
3 However certain drawbacks of VaR has resulted in the development of
Average Value-at-Risk (AVaR) as a risk measure, which provides an
estimate of the expected loss in the specified worst outcomes (in
percentage) over a specified future time horizon.
4 While VaR and AVaR has been studied in the theoretical setup of the
Black-Scholes-Merton framework, from a practical implementation point
of view, the estimation and modelling of VaR is executed through
approaches such as Historical Method, Variance-Covariance Approach and
Monte-Carlo Simulation.
5 In case of AVaR, one can adopt several approaches, including Historical
Method, Hybrid Method and Monte-Carlo Simulation.
Lecture 33

VaR
1 The mathematical formulation of VaR for the return variable X at
100(1 − α)% confidence level (α ∈ (0, 1)) for the next 1-day is given by:
VaR α (X ) = − inf [x ∈ R|α < FX (x)] .
2 A commonly used approach to estimate VaR is the historical simulation

approach. We consider the historical loss values for n days, namely vi , for
i = 0, 1, 2, . . . , n − 1. If the value today is vn , then vn+1 will be the
random variable to denote the value for tomorrow, which will take n − 1
values given by:
vi
vn+1 = vn , i = 1, 2, . . . , n − 1.
vi−1
3 The VaR will be the appropriate percentile of the values of vn+1 arranged
in descending order.
Operational Risk
1 Operational Risk encompasses a broad range of risk resulting from failure
in the internal process, people or system of a financial institution, or
resulting from external events.
2 While the high frequency losses can be modelled using parametric
distributions, the low frequency (but high intensity) losses have to be
modelled using the statistical approach of Extreme Value Theory (EVT).
3 Loss frequency can be modeled with Poisson distribution, where the
(λT )n
probability of n losses in T years is given by: e −λT .
n!
4 Suppose the number of losses in one year is N.
5 Then the loss severity for each can be generated using log-normal
distribution, with the losses being L1 , L2 , . . . , LN .
N
X
6 The total loss experience is then given by: Li .
i=1
Price Risk of Bonds and Macaulay Duration
1 The price risk of bonds depends exclusively on interest rate.
2 The approach we will use is to consider bonds as derivatives, whose
underlying variable is the interest rate.
3 A peculiarity of this underlying variable is that it is not directly traded in
the markets (unlike the underlying of plain vanilla stock options, for
example).
4 We consider the main risk measure for bonds, namely, Duration.
5 We explore the notion of Macaulay Duration for measuring interest-rate
risk, introduced in 1938.
6 Although it is based on several restrictive assumptions, its simplicity and
the fact that it captures the main factor that affects the bond price
volatility makes it the fundamental tool for bond portfolio management.
Duration: Definition and Interpretation
1 We first focus on the simplest case. Consider a risk-free pure discount
bond that will pay a nominal amount (say, 100) in exactly one period (for
example, one year) from today.
2 The one-year spot rate is r . Therefore, the price P of this bond is:
100
P= .
1+r
3 An investor who holds the bond only has to worry about one type of risk,
namely, a possible change (more explicitly, an increase) in the interest rate.
4 Therefore, the natural measure of the risk of holding the bond is the
derivative of the bond price with respect to the interest rate:
dP 100 P
=− =− .
dr (1 + r )2 1+r
5 We see that the effect on the price of the bond, of an increase in the level
of interest rate will be larger (in absolute value), the larger the price of
the bond.
Duration: Definition and Interpretation (Contd ...)
1 We can easily extend the same type of analysis to a pure discount bond
that has several periods (say, T ) to maturity. Suppose that the T -period
spot rate is r . Then the price of the bond is:
100
P= .
(1 + r )T
2 Consequently, its sensitivity to a change in the interest rate is:

dP 100 P
= −T × = −T ,
dr (1 + r )T +1 1+r
which is similar to the previous equation, with the additional factor of the
time left to maturity.
3 That is, the price of a bond is more sensitive to a change in interest rates,
if its maturity is longer.
Duration of Coupon Bonds
1 Consider a bond that pays the coupons of size Ci , at the end of each
period, for T periods (we assume that the final coupon CT includes the
principal).
2 Suppose further that the corresponding interest rates are denoted by
ri , i = 1, 2, . . . , T so that the price of the bond is:
T
X Ci
P= .
i=1
(1 + ri )i
3 We see that we cannot extend in a straightforward manner, the analysis

we performed for pure discount bonds, since the price of the bond is
determined by a collection of interest rates that change with time.
4 However, in practice, interest rate for different maturities are highly
correlated and they frequently (though not always) move in parallel, that
is, if the one-year spot rate goes up, so does the five year spot rate, for
example.
Macaulay Duration
1 The problem is, then, to find a single number that will be a good
representation of the whole term structure and whose changes will be a
good summary of changes in the term structure.
2 The obvious candidate is the yield or internal rate of return of the bond.
Recall that the yield of the bond is the number that “y ” that satisfies:
T
X Ci
P= .
i=1
(1 + y )i
3 Comparing the equations, it is clear that the yield y depends only on the
rates ri that affect the price of the bond.
4 For example, if all the rates ri go up, the yield y also goes up. Then,
similar to our analysis of risk for pure discount bonds, we consider the
derivative of the coupon bond with respect to its yield:
T T Ci
dP X Ci P X (1+y )i
=− i× = − i ,
dy i=1
(1 + y )i+1 1 + y i=1 P
where the second equality is simply the result of multiplying and dividing
by P.
Macaulay Duration (Contd ...)
1 The second factor of the equation is called Duration or Macaulay
Duration.
2 Definition: The duration of a bond is a measure of the sensitivity of the
bond to interest rate movements.
3 More explicitly, the duration D of a bond that pays coupons Ci and has
yield y and maturity T is given by:
T Ci
(1+y )i
X
D := i .
i=1
P
4 It is clear that the sensitivity of a bond to interest rate changes (measured

through the change is y ) is larger, then larger is the duration of the bond.
The holder of a bond with high duration has more interest rate exposure
than the holder of a bond with low duration.
5 The duration is the weighted average of all the time points i at which the
bond makes a payment.
1 Ci
6 Each of these numbers is assigned a weight: .
P (1 + y )i
Macaulay Duration (Contd ...)
T
X Ci
1 This actually adds up to one, since: = P.
i=1
(1 + y )i
2 The weight corresponding to time point i is the proportion of the value of
the bond that corresponds to the payment received at i.
3 This average places more weight on the time points that are more
important for the value of the bond.
4 For example, a bond that pays low coupons initially and large coupons
later on will have a higher duration than a bond that pays high coupons
early and low coupons later.
5 The duration of a pure discount bond is equal to its maturity.
Example
1 Consider a bond with a nominal value of 100 that pays a coupon of 8 at the end of each
year, has an yield of 5% and has five years left until maturity. Then:
5
X 8 100
P = + = 112.99,
i=1
(1.05)i (1.05)5
5 8 100
X (1.05)i (1.05)5
D = i +5 = 4.36.
i=1
112.99 112.99
2 Now consider the above example with 30 years left until maturity. Then:
30
X 8 100
P = + = 146.12,
i=1
(1.05)i (1.05)30
30 8 100
X (1.05)i (1.05)30
D = i + 30 = 14.82.
i=1
146.12 146.12
3 Finally consider the previous example, but with an yield of 10% instead of 5%. Then:
30
X 8 100
P = + = 81.15,
i=1
(1.1)i (1.1)30
30 8 100
X (1.1)i (1.1)30
D = i + 30 = 10.65.
i=1
81.15 81.15
Lecture 34

Hedging by Immunization: An Example
1 Suppose that the term structure is flat and the annual interest rate is 5%.
2 We have a liability with a nominal value of 100, and the payment will take
place in two years.
3 In the market there are only two pure discount bonds, a one year pure
discount bond with nominal value 100 and a four year pure discount bond
with nominal value 100.
4 We start with the present value of the liability, that is:
100
= 90.70
(1.05)2
in cash.
5 In order to guarantee that we will be able to meet the payment in two
years (approximately), we have to invest the money on a portfolio of bonds
with the duration equal to the duration of the portfolio of short positions.
6 In this case, there is only one liability with maturity in two years,
equivalent to a short position in a pure discount bond, with maturity in
two years and therefore, the duration is equal to two years.
7 We want to invest in a portfolio with duration of two, as well.
Hedging by Immunization: An Example (Contd ...)
1 There are only two pure discount securities, with maturities (and,
therefore, durations) equal to one and four, respectively.
2 We now take advantage of the property:
If a bond portfolio comprises of n bonds with weights w1 , w2 , . . . , wn and
durations of D1 , D2 , . . . , Dn , respectively, then the duration of the
portfolio is given by:
X n
D= wi Di .
i=1
3 We will invest a proportion α of the initial capital in the bond with

duration 1 and the rest in the bond with duration 4.
4 For matching duration, we solve:
2
αD1 + (1 − α)D2 = D ⇒ α + (1 − α)4 = 2 ⇒ α = .
3
2
5 We therefore invest 90.70 × = 60.47 in the one year bond, and the rest
3
90.70 − 60.47 = 30.23 in the four year bond.
1 The price of the one year bond is:
100
= 95.24,
1.05
and therefore, we buy:
60.47
= 0.63
95.24
units of the one year bond.
2 Similarly, the price of the four year bond is:
100
= 82.27,
(1.05)4
and therefore, we buy:

30.23
= 0.37
82.27
units of the four year bond.
1 Suppose that immediately after investing in such a portfolio, the interest
rates go up to 6% (and the term structure remains flat). The present
values of both the liability and the portfolio of bonds drop. The new
values of the liability and the portfolio of bonds are:
100 100 100
= 89 , 0.63 × + 0.37 × = 88.74.
(1.06)2 1.06 (1.06)4
The drop in the value of the portfolio roughly matches the drop in the
value of the liability.
2 Suppose now that, immediately after forming the portfolio of bonds, the
interest rate drops to 4% (and the term structure remains flat). In this
case, the value of liability and portfolio goes up. The new values are:
100 100 100
= 92.46 , 0.63 × + 0.37 × = 92.20.
(1.04)2 1.04 (1.04)4
Again, the increase in the value of the portfolio approximately matches

the increase in the value of the liability.
1 There are several problems with the previous analysis. Firstly, this
approach relies on the unrealistic assumptions that the term structure is
flat and that all the interest rate movements will be in parallel shifts.
2 Secondly, the duration changes as the interest rates move.
3 Example for Change in Duration: Consider the liability of the previous
Example and the portfolio of bonds formed for immunization purposes, in
case the rates go down to 4%. After the interest rate changes, the
duration of the liability will still be equal to 2. However, the duration of
the portfolio of bonds will not be equal to 2. Now, the new price of the
portfolio for bonds is 92.20. The new weights of the one year and the four
year bonds are:
100
100
0.63 × 1.04 0.37 × (1.04)4
= 0.66, = 0.34.
92.20 92.20
4 Duration of the portfolio is:
D = 0.66 × 1 + 0.34 × 4 = 2.02.
Therefore the bond is more sensitive to an increase in interest rates.

Convexity
1 Convexity is the measure of the error we are exposed to, by using the first
derivative to predict the impact of a change in interest rates, on the price
of a portfolio of bonds.
2 If we consider the second derivative of the bond price with respect to the
T
d 2P X Ci
yield, then we obtain: = i(i + 1) .
dy 2 i=1
(1 + y )i+2
3 We see that it is positive and therefore, the relationship between the price
of a bond and its yield is convex.
4 The stronger the convexity (the larger the second derivative) the larger the
error we are exposed to by using the duration for immunization purposes.
5 The convexity is defined as:
1 ∂2P
C := .
P ∂y 2
6 Using a Taylor’s Series expansion up to the second-order term, we obtain:
∂P 1 ∂2P D 1
∆P = ∆y + (∆y )2 = − P∆y + CP(∆y )2 .
∂y 2 ∂y 2 1+y 2
Value-at-Risk (VaR)
1 Ideally we look for a number (or a set of numbers) that expresses the
potential loss with a given level of confidence, enabling the risk manager
to adjudge whether the risk is acceptable or not.
2 Value-at-Risk (VaR) is a standard benchmark for measuring financial risk
a
.
a The main drawback, however, is that it does not provide information about the
potential impact of extreme and highly unlikely events occurring
Quantiles
1 Let (Ω, F, P) be a probability space and let X : Ω → R be a random
variable.
2 The cumulative distribution function FX : R → [0, 1], defined by:
FX (x) = P(X ≤ x),
which is right-continuous and non-decreasing.

Quantiles: Definition
1 For α ∈ (0, 1), the number:
q α (X ) = inf{x|α < FX (x)},
is called the upper α-quantile of X .

2 For α ∈ (0, 1), the number:
qα (X ) = inf{x|α ≤ FX (x)},
is called the lower α-quantile of X .

3 Any q ∈ [qα (X ), q α (X )] is called an α-quantile of X .
4 Properties: Let X and Y be random variables. Then:
(A) X ≥ Y ⇒ q α (X ) ≥ q α (Y ).
(B) For any b ∈ R, q α (X + b) = q α (X ) + b.
(C) For b > 0, q α (bX ) = bq α (X ).
(D) q α (−X ) = −q(1−α) (X ).
Lemma 1
If FX is continuous and strictly increasing, then q α (X ) = FX−1 (α).
Proof of Lemma 1
Since the given condition on FX ensures that it is invertible, the inverse
function α → FX−1 (α) is continuous and α < FX (x) is equivalent to
FX−1 (α) < x. This gives:
q α (X ) = inf{x|α < FX (x)} = inf{FX−1 (α) < x} = FX−1 (α).
Lemma 2
Let X be a random variable. If f : R → R is right-continuous and
non-decreasing, then:
q α (f (X )) = f (q α (X )).
Measuring Downside Risk
1 We work on a single step financial market model, in which we invest at
time t = 0 and terminate our investment at time t = T .
2 We denote by X , the discounted value of the position at time T .
3 Definition: For α ∈ (0, 1), we define the Value-at-Risk (VaR) of X at
confidence level 100(1 − α)% as:
VaR α (X ) = −q α (X ) = − inf{x|α < FX (x)}.
4 Let us observe that, since X denotes the gain from an investment,

therefore −X denotes the loss.
5 Accordingly, we can express VaR, in terms of loss, as follows:
VaR α (X ) = −q α (X ) = q1−α (−X ),

= inf{x|(1 − α) ≤ P(−X ≤ x)} = inf{x|P(x < −X ) ≤ α}.
6 This means that the probability of the loss exceeding VaR α (X ) is no

greater than α.
7 In other words, at confidence level (1 − α), our loss is no worse than
VaR α (X ).
Lecture 35

Some Properties of VaR
Let X and Y be random variables. Then:
1 X ≥ Y ⇒ VaR α (X ) ≤ VaR α (Y ).
2 For any a ∈ R, VaR α (X + a) = VaR α (X ) − a.
3 For any a ≥ 0, VaR α (aX ) = aVaR α (X ).
Theorem 1
N
X
Assume that X is a discrete random variable with P(X = xi ) = pi , pi = 1
i=1
and x1 < x2 < · · · < xN . Then:
VaR α (X ) = −xkα ,
kα−1
X
where kα ∈ N is the largest number such that pi ≤ α.
i=1
Proof of Theorem 1
1 Since X has discrete distribution and x1 < x2 < · · · < xN , we can see that:
k
X
P(X ≤ xk ) = pi .
i=1
( k
) ( k−1
)
X X
2 We shall use the fact: min k|α < pi = max k| pi ≤ α .
i=1 i=1
3 Therefore:
q α (X ) = inf {x|α < P(X ≤ x)} ,

= min {xk |α < P(X ≤ xk )} ,
( k
)
X
= min xk α < pi ,
i=1
( k−1
)
X
= max xk pi ≤ α = xkα .
i=1
4 Hence: VaR α (X ) = −q α (X ) = −xkα .

Theorem 2
1 For Z ∼ N(0, 1), VaR α (Z ) = N −1 (α) (∵ q α (X ) = FX−1 (α), if FX is
continuous and strictly increasing).
2 Suppose that the today’s price of a stock is S(0).
3 Assume that the price of the stock at time T is S(T ) = S(0)e m+σZ , with
Z ∼ N(0, 1).
4 Let X = e −rT S(T ) − S(0).
5 Now q α (Z ) = N −1 (α).
6 We observe that X = f (Z ), where f (ζ) = e −rT S(0)e m+σζ − S(0), is an
increasing function.
7 Accordingly, we get:
VaR α (X ) = −q α (f (Z )) = −f (q α (Z )) = −f (N −1 (α)),
−1
h i
= S(0) 1 − e m−rT +σN (α) .
Example on Calculating VaR Using Historical Simulation
1 Suppose that an investor owns, on September 25, 2008, a portfolio worth
$10 million consisting of investments in four stock indices:
(A) Dow Jones Industrial Average (DJIA) in the US.
(B) Financial Times Stock Exchange (FTSE) 100 in the UK.
(C) Cotation Assistée en Continu (CAC) 40 in France.
(D) Nikkei 225 in Japan.
2 The value of the investment in each index on September 25, 2008, is
shown in the following Table.
Index Portfolio value ($000s)

DJIA $4,000
FTSE 100 $3,000
CAC 40 $1,000
Nikkei 225 $2,000
Total $10,000
Table: Investment portfolio used for the VaR calculation example
Example on Calculating VaR Using Historical Simulation (Contd ...)
An extract of the 501 days of historical data (from August 7, 2006 to
September 25, 2008) is shown in the following Table.
Day Date DJIA FTSE 100 CAC 40 Nikkei 225

0 August 7, 2006 11,219.38 5,828.8 4,956.34 15,154.06
1 August 8, 2006 11,173.59 5,818.1 4,967.95 15,464.66
2 August 9, 2006 11,076.18 5,860.5 5,025.15 15,656.59
3 August 10, 2006 11,124.37 5,823.4 4,976.64 15,630.91
.. .. .. .. .. ..
. . . . . .
499 September 24, 2008 10,825.17 5,095.6 4,114.54 12,115.03
500 September 25, 2008 11,022.06 5,197.0 4,226.81 12,006.33
Table: Data on stock indices for historical simulation calculation

1 September 25, 2008 is an interesting date to choose in evaluating an
equity investment.
2 The turmoil in credit markets, which started in August 2007, was over a
year old.
3 Equity prices have been declining for several months.
4 Volatilities were increasing.
5 Lehman Brothers had filed for bankruptcy ten days earlier.
6 The next Table shows the values of the market variables on September 26,
2008, for the scenarios considered.
Scenario DJIA FTSE 100 CAC 40 Nikkei 225 Portfolio Value Loss
Number ($000s) $000s)
1 10,977.08 5,187.46 4,236.71 12,252.62 10,021.502 -21.502
2 10,925.97 5,234.87 4,275.48 12,155.54 10,023.327 -23.327
3 11,070.01 5,164.10 4,186.01 11,986.84 9,985.478 14.522
. . . . . . .
. . . . . . .
. . . . . . .
499 10,831.43 5,057.36 4,117.75 12,030.80 9,828.450 171.550
500 11,222.53 5,300.42 4,342.14 11,899.00 10,141.826 -141.826
Table: Scenarios generated for September 26, 2008, using the data in the
previous Table
1 Scenario 1 (the first row in the previos Table shows the values of market
variables on September 26, 2008, assuming that their percentage changes
between September 25 and September 26, 2008, are the same as they
were between August 7 and August 8, 2006.
2 Scenario 2 (the second row in the previous Table) shows the value of the
market variables on September 26, 2008, assuming these percentage
changes are the same as those between August 8 and August 9, 2006 and
so on.
3 In general, Scenario i assumes that the percentage changes in the indices
between September 25 and September 26, 2008 are the same as they were
between Day (i − 1) and Day i for 1 ≤ i ≤ 500.
4 The 500 rows of the previous Table are the 500 scenarios considered.
5 The DJIA was 11,022.06 on September 25, 2008. On August 8, 2006, it
was 11,173.59, down from 11,219.38 on August 7, 2006.
6 Therefore the value of the DJIA under Scenario 1 is:
11, 173.59
11, 022.06 × = 10, 977.08.
11, 219.38
1 Similarly, the values of the FTSE 100, the CAC 40 and Nikkei 225 are
5,187.46, 4,236.71 and 12,252.62, respectively.
2 Therefore the value of the portfolio under Scenario 1 is (in $000s) is:
10977.08 5187.46 4236.71 12252.62
4, 000× +3, 000× +1, 000× +2, 000× = 10, 021.502.
11022.06 5197.00 4226.81 12006.53
3 The portfolio therefore has a gain of $21,502 under Scenario 1.

4 A similar calculation is carried out for the other scenarios.
5 The losses for the 500 different scenarios are then ranked.
6 An extract from the results of doing this is shown in the next Table.
Scenario Number Loss ($000s)
494 499.395
339 359.440
329 341.366
299 251.943
487 247.571
131 241.712
.. ..
. .
Table: Losses ranked from highest to lowest for 500 scenarios
1 The worst scenario number is 494 (where indices are assumed to change
in the same way that they did at the time of the bankruptcy of Lehman
Brothers).
2 The one day 99% VaR can be estimated as the fifth-worst loss, which is
$247, 571.
3 In
√ this case the ten-day VaR would therefore be
10 × 247, 571 = 782, 889.
4 In practice, a bank’s portfolio is, of course considerably more complicated
than the ones we have considered here.
5 It is likely to consist of thousands or tens of thousands of positions.
6 Some of the bank’s position are typically in forward contracts, options,
and other derivatives.
Lecture 36

Central Limit Theorem
1 What does the Theorem say: Sum of a large number of i.i.d. random
variables will be approximately normal.
2 More precisely: Let X1 , X2 , X3 , . . . be a sequence of i.i.d. random
variables, each with mean µ and variance σ 2 .
Xn
3 Let Sn := Xi .
i=1
4 Central Limit Theorem: For large n, Sn behaves like N (nµ, nσ 2 ).

Sn − nµ
5 Equivalently: √ ∼ N (0, 1), in the sense that for any x, we have:
σ n

Sn − nµ
P √ ≤ x → N (x) as n → ∞.
σ n
Pseudo-Random Numbers
1 Models for option valuation involve “randomness”.
2 Computers being discrete: It is usually sufficient to work with
“pseudo-random” numbers.
3 Pseudo-random numbers: Collections of numbers that are produced by
deterministic algorithms, and yet seem to be random in the sense that
they possess appropriate statistical properties.
4 We may test a pseudo-random number generator by taking M samples
{ξi }M
i=1 and computing:
M
1 X
(A) Sample Mean: µM = ξi .
M i=1
M
2 1 X
(B) Sample Variance: σM = (ξi − µM )2 .
M − 1 i=1
Pseudo-Random Numbers (Contd ...)
1 Another approach to testing a random number generator is to divide the
x-axis into sub-intervals or bins of length ∆x and count the number of
samples in each sub-intervals.
2 Let:
(A) Ni : Number of samples in the bin [i∆x, (i + 1)∆x].
(B) M: Total number of samples.
3 Accordingly, we must expect:
Ni
P (i∆x ≤ X ≤ (i + 1)∆x) ≈ .
M
4 Also, we know that:
(i+1)∆x
Z
P (i∆x ≤ X ≤ (i + 1)∆x) = f (x)dx.
i∆x
Pseudo-Random Numbers (Contd ...)
1 We use the Riemann sum approximation and let xi denote the mid-point
of the sub-interval [i∆x, (i + 1)∆x]. Then:
(i+1)∆x
Z
f (x)dx = ∆xf (xi ).
i∆x
Ni
2 So we expect: f (xi ) = .
M∆x
Ni
3 Accordingly, plotting should give an approximation to the density
M∆x
function values of f (xi ).
Monte-Carlo Method
1 Recall: Valuation of an European option reduces to the computation of an
expectation.
2 We will accomplish this through the usage of pseudo-random numbers.
3 To begin with, we consider the case of a general random variable X ,
whose expected value E (X ) = a and variance Var (X ) = b 2 are not known.
4 Suppose:
(A) We are interested in computing and approximation to “a” (and
possibly “b”).
(B) We are able to take independent samples of X using a
pseudo-random number generator.
5 Let X1 , X2 , . . . , XM denote independent random variables with the same
distribution as X .
M
1 X
6 Then, we might expect: aM := Xi .
M i=1
7 It can be shown that aM is an unbiased estimate of E (X ).
Monte-Carlo Method (Contd ...)
M
2 1 X
1 To estimate Var (X ) we can use bM := (Xi − aM )2 .
M − 1 i=1
2
2 It can be shown that bM is an unbiased estimate of Var (X ).
M
X
3 Now, by the Central Limit Theorem Xi behaves like N (Ma, Mb 2 )
i=1
random variate.
b2

4 So: aM − a is approximately N 0, .
M
5 Equivalently: aM − a is approximately N (0, 1) random variable scaled by
b
√ .
M

1.96b 1.96b
6 This leads to: P a − √ ≤ aM ≤ a + √ = 0.95 a .
M M
a If Z ∼ N (0, 1), then P(µ − 1.96σ ≤ Z ≤ µ + 1.96σ) = 0.95
Monte-Carlo Method (Contd ...)

1.96b 1.96b
1 We thus have: P aM − √ ≤ a ≤ aM + √ = 0.95.
M M
2 Replacing the unknown b by the approximation bM , we see that the
1.96b 1.96b
unknown expected value a lies in the interval aM − √ , aM + √ ,
M M
with probability 0.95.
3 In other words, this gives an approximate 95% confidence interval for “a”.
4 This analysis leads us to the basic Monte-Carlo method for approximating
“a”.
5 We compute M independent samples and determine aM .
6 In order to monitor the error, we also compute the variance approximation
2
“bM ”.
Monte-Carlo for European Option Valuation
1 We are now in a position to use Monte-Carlo for option valuation.
2 We consider the payoff at time T of the form f (S(T )).
3 Then the time 0 price of the option is given by e −rT E∗ (f (S(T ))).
4 Recall that: 1 2
S(T ) = S(0)e (r − 2 σ )T +σW (T ) .
5 Consequently, the price of the option is given by:
1 2
√
e −rT f S(0)e (r − 2 σ )T +σz T , z ∼ N (0, 1).
Monte-Carlo for European Option Valuation (Contd ...)
1 Then the resulting Monte-Carlo algorithm can be summarized as:
for i = 1 : M
Compute an N (0, 1) sample ξi
√

1 2
Set Si = S(0) exp µ − σ T + σ T ξi
2
−rT
Set Vi = e f (Si )
end
M
1 X
Set aM := Vi
M i=1
M
2 1 X
Set bM := (Vi − aM )2 .
M − 1 i=1
2 The output provides an approximate option price aM and an approximate
95% confidence interval.
Lecture 37

Monte-Carlo for Greeks
1 In addition to the option value, we know that the Greeks, particularly
∂V
∆= , plays a vital role in the hedging strategy that a trader must
∂S
adopt to replicate the option.
2 A Taylor series expansion gives:
∂V (S, t) V (S + h, t) − V (S, t)
= + O(h), h → 0.
∂S h
3 Choosing a small value of h, we obtain the finite difference approximation:
∂V (S, t) V (S + h, t) − V (S, t)
≈ + O(h).
∂S h
4 This produces an approximation to Delta, at a single point, based on
option values at two points with slightly different arguments.
Monte-Carlo for Greeks (Contd ...)
1 Using the risk-neutral discounted expected payoff, we can approximate the
time-zero Delta via Monte-Carlo estimation, as follows:

E∗ (f (S(T ))|S(0) = S0 + h) − E∗ (f (S(T ))|S(0) = S0 )
e −rT .
h
2 Observation:
Error for each of the two Monte-Carlo estimation is
1
O √ and hence after dividing the difference by h, the overall error
M
1
can be expected to be √ .
h M
3 Hence: To keep a good approximation for small h, we need to make M
large (more samples will be required than the basic Monte-Carlo
approach).
1 Note: The two random variables in the two expectations are clearly
“correlated”.
2 Implication: For any particular asset path, the payoff starting from
S(0) = S0 is likely to be close to the payoff starting from S0 + h.
3 Intuitively: The corresponding sample mean errors shall be close, provided
that we use the same paths for the two simulations.
4 We thus apply Monte-Carlo to the equivalent problem:
E∗ [[f (S(T ))|S(0) = S0 + h] − [f (S(T ))|S(0) = S0 ]]

e −rT .
h
Thus we have the following Monte-Carlo algorithm for approximating Delta:
for i = 1 : M
Compute a N (0,
1) sample ξ i
√

1 2
Set Si = S0 exp µ − σ T + σ T ξi
2
√

h 1
Set Si = (S0 + h) exp µ − σ 2 T + σ T ξi
2
e −rT f (Sih ) − f (Si )
Set ∆i =
h
end
M
1 X
Set aM := ∆i
M i=1
M
2 1 X
Set bM := (∆i − aM )2 .
M − 1 i=1
Monte-Carlo for Exotic Options
1 The basic Monte-Carlo for European option can easily be extended to
handle path dependency.
2 Accordingly, an extra step is involved, in terms of the temporal grid points
T −0
tj = j∆t, for 0 ≤ j ≤ N, such that ∆t = .
N
3 Then the asset price path is given by:
√

1
Sj+1 = Sj exp r − σ 2 ∆t + σ ∆tξj , ξj ∼ N (0, 1).
2
Example of Exotic Option: Algorithm for Up-And-Out Barrier Call Option
for i = 1 : M
for j = 0 : N − 1
Compute an N (0, 1)
sample ξj
√

1
Set Sj+1 = Sj exp r − σ 2 ∆t + σ ∆tξj
2
end
Set Simax = max Sj
0≤j≤N
If Simax < B
Set Vi = exp(−rT ) max(SN − K , 0)
Else Set Vi = 0
end
M
1 X
Set aM := Vi
M i=1
M
2 1 X
Set bM := (Vi − aM )2 .
M − 1 i=1
Example of Exotic Option: Algorithm for Floating Strike Lookback Call Option
for i = 1 : M
for j = 0 : N − 1
Compute an N (0, 1)
sample ξj
√

1
2
end
Set Simin = min Sj
0≤j≤N
Set Vi = exp(−rT ) max(SN − Simin , 0)
end
M
1 X
Set aM := Vi
M i=1
M
2 1 X
Set bM := (Vi − aM )2 .
M − 1 i=1
Example of Exotic Option: Algorithm for Average Price Asian Put Option
for i = 1 : M
for j = 0 : N − 1
Compute an N (0, 1)
sample ξj
√

1
2
end
X N
Set Smeani = ∆t Sj
j=1
Set Vi = exp(−rT ) max(K − Smeani )
end
M
1 X
Set aM := Vi
M i=1
M
2 1 X
Set bM := (Vi − aM )2 .
M − 1 i=1
Some Simulation Results: Pricing of Exotic Options
1 We ran the simulations for the four different path dependent options using
MatLabTM (Code: Exotic_Options.m).
2 The number of sample paths and time intervals taken are M = 5000 and
N = 1000 respectively.
3 The parameters used were S(0) = 100, r = 6%, σ = 30%, K = 120 and
B = 250.
4 The expiration time was T = 1 and hence ∆t = T /N = 10−3 .
Option Type Price Variance 95% CI
European put 20.3997 364.4944 [19.8705,20.9289]
Up-and-out barrier call 6.8101 239.8059 [6.3808,7.2393]
Floating strike lookback call 23.7967 532.7144 [23.1570,24.4365]
Average price Asian put 17.8306 187.0266 [17.4515,18.2097]
Table: Results of Monte Carlo simulation for exotic option pricing

Lecture 38

Parametric Estimation Using Historical Data
1 Suppose: Historical asset price data is available at equally spaced time
points ti = i∆t.
2 Accordingly, let S(ti ) denoted the asset price at time ti .
3 We define the following log returns:
S(ti )
Ui := log .
S(ti−1 )
4 Recall
that{Ui } are i.i.d. normal random variates with mean
1 2
µ − σ ∆t and variance σ 2 ∆t.
2
5 Thus, collating historical asset price data and obtaining the values of
log-returns (Ui ’s) is equivalent to sampling from:

1
N µ − σ 2 ∆t, σ 2 ∆t .
2
Parametric Estimation Using Historical Data (Contd ...)
1 Suppose that the current time is tn .
2 Let the M + 1 most current asst prices be
S(tn−M ), S(tn−M+1 ), . . . , S(tn−1 ), S(tn ).
3 Using the corresponding log-return data {Un+1−i }M
i=1 , the sample mean
and the sample variance estimate become:
M
1 X
(A) aM := Un+1−i .
M i=1
M
2 1 X
(B) bM := (Un+1−i − aM )2 .
M − 1 i=1
4 Thus we may estimate µ and σ from:
bM
(A) σ 2 ∆t = bM2
⇒ σ = σ∗ = √ .
∆t
2
aM + 21 bM

1
(B) µ − σ ∆t = aM ⇒ µ = µ∗ =
2
.
2 ∆t
Accuracy of the Sample Variance Estimator
bM
1 In order to get idea about the accuracy of the estimate σ ∗ = √ , we
∆t
need to recognize the fact that we are essentially using Monte-Carlo
2
simulation to compute bM as an approximation to the expected value of
2
variable(U − E (U))
the random , where
1
U∼N µ − σ 2 ∆t, σ 2 ∆t .
2
2 Equivalently, after dividing throughout by ∆t, we are using Monte-Carlo
b2
simulation to compute σ ∗2 = M as an approximation to the expected
∆t 2
value of the random variable Ub − E (U)b , where
√

1
Ub ∼ N µ − σ2 ∆t, σ 2 .
2
Accuracy of the Sample Variance Estimator (Contd ...)
1 Hence, an approximate 95% confidence interval for σ 2 is given by:
1.96v
σ ∗2 ± √ ,
M
2
where v 2 is the variance of the random variable Ub − E (U)
b .
2 It can be shown that v 2 = 2σ 4 .

Why Variance Reduction?
1 While: The Monte-Carlo method gives a simple and flexible technique for
option valuation, it can also be expensive.
2 Accordingly: We consider approaches to improve efficiency.
3 Recall:
√ The width of the confidence interval is inversely proportional to
M.
4 This means that in order to get an extra digit √
of accuracy, the confidence
interval has to be shrunk by a factor of 10 = ( M), which requires
100 = (M) as many samples.
5 However,
p we also saw that the confidence interval also scales with
Var (Xi ).
6 This motivates the idea of replacing Xi , with another set of i.i.d. random
variables, with the same mean as Xi , but with smaller variance.
7 This is the fundamental idea behind the approach of variance reduction.
Antithetic Variates: Example of Uniform Distribution
1 We begin with an illustration of the antithetic variates.
2 Suppose that we apply Monte Carlo in order to approximate the expected
value: √
I = E e U , U ∼ U (0, 1).
3 It may be noted that I = 2 (actual integral).

4 A Monte Carlo estimate of I is given by:
M
1 X
IM = Yi ,
M i=1
√
Ui
where Yi = e with i.i.d. Ui ∼ U (0, 1).
e2 − 7
5 Note: Var (Yi ) = .
2
6 We now consider the antithetic variate Monte Carlo estimator:
M
√ √
1 Xb e Ui + e 1−Ui
IM =
b Yi , where Yi =
b , Ui ∼ U (0, 1).
M i=1 2
Antithetic Variates: Example of Uniform Distribution (Contd ...)
1 This antithetic version “re-uses” Ui in the form 1 − Ui .
√
2 Note that 1 − Ui ∼ U (0, 1) so E Ybi = E e U .
3 In terms of random number generation, both have the same costs, but
there is some overhead associated with IbM .
4 Twice as many evaluations of the exponential and square root functions
are required.
5 We have:
√ √
√ √ √
!
Ui 1−Ui
e +e 1
Var = Var e Ui + Cov e Ui , e 1−Ui
2 2
6 Now:
√ √ Z1 √ √
Ui 1−Ui
(A) E e e = e x+ 1−x dx.
0
√ √ Z1 √ √
(B) Cov e Ui , e 1−Ui = e x+ 1−x dx − 22 .
0
Antithetic Variates: Example of Uniform Distribution (Contd ...)
1 Thus:
Z1
e2 15 1 √ √
x+ 1−x
Var (Ybi ) = − + + e dx.
4 4 2
0
2 Using numerical quadrature for numerical integration, we get:
Var (Ybi ) ≈ 0.001073.
3 Hence:
Var (Yi )
≈ 181.2485.
Var (Ybi )
Analysis of the Uniform Case
1 To understand the antithetic variate technique works, consider the more
generalized case of approximating:
I = E (f (U)), where U ∼ U (0, 1),
for some function f .

2 The standard Monte Carlo estimate is:
M
1 X
IM = f (Ui ), with i.i.d. Ui ∼ U (0, 1).
M i=1
3 The antithetic alternative is:

M
1 X f (Ui ) + f (1 − Ui )
IbM = , with i.i.d. Ui ∼ U (0, 1).
M i=1 2
4 Thus:

f (Ui ) + f (1 − Ui ) 1
Var = Var (f (Ui )) + Cov (f (Ui ), f (1 − Ui )) .
2 2
Analysis of the Uniform Case (Contd ...)
1 The success of the new scheme is contingent on whether
1
Var (f (Ui ) + f (1 − Ui )) is smaller than Var (f (Ui )).
2
2 This, then boils down to making Cov (f (Ui , f (1 − Ui ))) as negative as
possible.
3 We want: f (Ui ) to be big (relative to its mean) when f (1 − Ui ) is small
(relative to the mean).
4 Intuitively: This approach will work when f is monotonic.
5 Loosely speaking: Antithetic variate attempts to compensate for samples
that are above the mean, by adding samples which are below the mean,
and vice versa.
6 We can now convert this intuition into a mathematical result.
7 (A) f is monotonically increasing: x1 ≤ x2 ⇒ f (x1 ) ≤ f (x2 ).
(B) f is monotonically decreasing: x1 ≤ x2 ⇒ f (x1 ) ≥ f (x2 ).
(C) If f and g are both monotonically increasing or both monotonically
decreasing functions, then (f (x) − f (y ))(g (x) − g (y )) ≥ 0.
Lemma
If f and g are both monotonic increasing functions or both monotonic
decreasing function, then for any random variable X :
Cov (f (X ), g (X )) ≥ 0.
Proof of the Lemma

1 Let Y be a random variable that is independent of X with the same
distribution.
2 Then (f (X ) − f (Y ))(g (X ) − g (Y )) ≥ 0.
3 So the random variable (f (X ) − f (Y ))(g (X ) − g (Y )) must have a
non-negative expected value.
4 Hence:
0 ≤ E [(f (X ) − f (Y ))(g (X ) − g (Y ))]

= E [f (X )g (X )] − E [f (X )g (Y )] − E [f (Y )g (X )] + E [f (Y )g (Y )]
= 2E [f (X )g (X )] − 2E [f (X )]E [g (X )]
= 2Cov (f (X ), g (X )).
5 Therefore the results follows.

Analysis of the Uniform Case (Contd ...)
1 Note that:
(A) If f is a monotonic increasing function, then so is −f (1 − x).
(B) If f is a monotonic decreasing function, then so is −f (1 − x).
2 In either case, applying our Lemma gives Cov (f (X ), −f (1 − X )) ≥ 0.
3 Equivalently Cov (f (X ), f (1 − X )) ≤ 0.
4 Hence:
f (Ui ) + f (1 − Ui ) 1
Var ≤ Var (f (Ui )),
2 2
if f is monotonic.
5 For monotonic f , the variance in the antithetic sample is always less than
or equal to half of that in the standard sample.
Analysis of the Normal Case
1 In case of normal random variates, for I = E (f (Z )), where Z ∼ N (0, 1),
the standard Monte-Carlo estimate is:
M
1 X
IM = f (Zi ), with i.i.d. Zi ∼ N (0, 1).
M i=1
2 Consequently, the antithetic variate is:

M
1 X f (Zi ) + f (−Zi )
IbM = , with i.i.d. Zi ∼ N (0, 1).
M i=1 2
3 Since the N (0, 1) distribution is symmetric about the origin, rather than
1
about , the antithetic estimate used is “−Zi ” rather than “1 − Zi ”.
2
4 Further “−Zi ” is also a N (0, 1) random variable.
5 The analysis done earlier can now be repeated to give:

f (Zi ) + f (−Zi ) 1
Var ≤ Var (f (Zi )),
2 2
when f is monotonic.
Lecture 39

Antithetic Variates in Option Valuation
1 The Application: Monte-Carlo estimation of path dependent options.
2 Discretization: The time interval [0, T ] is discretized and the risk-neutral
prices are computed at the time points {ti }N i=1 , with ti = i∆t, N∆t = T .
3 On each increment, the price update uses a N (0, 1) random variable
coming from the i.i.d. sequence {Z0 , Z1 , . . . , ZN−1 }.
4 We wish to compute the expected value of some payoff function.
5 Therefore: We are looking for the expected value of a function of the N
i.i.d. N (0, 1) random variables {Z0 , Z1 , . . . , ZN−1 }.
6 The antithetic variates technique is to take the average payoff from one
path with samples {Z0 , Z1 , . . . , ZN−1 } and another path with samples
{−Z0 , −Z1 , . . . , −ZN−1 }
An Example: Up-and-In European Call Option using Basic Monte-Carlo
for i = 1 : M
for j = 0 : N − 1
Compute an N (0, 1)
sample ξj
√

1 2
Set Sj+1 = Sj exp r − σ ∆t + σ ∆tξj
2
end
Set Simax = max Sj
0≤j≤N
If Simax > B
Set
Vi = exp(−rT ) max(SN − K , 0)
Else
Vi = 0
end
M
1 X
Set aM := Vi
M i=1
M
2 1 X
Set bM := (Vi − aM )2 .
M − 1 i=1
An Example: Up-and-In European Call Option using Antithetic Variate Monte-Carlo
for i = 1 : M
for j = 0 : N − 1
Compute an N (0, 1) sample ξj
√

1 2
Set Sj+1 = Sj exp r− σ ∆t + σ ∆tξj
2
√

1 2
Set S̄j+1 = S̄j exp r− σ ∆t − σ ∆tξj
2
end
max
Set Si = max Sj
0≤j≤N
max
Set S̄i = max S̄j
0≤j≤N
If Simax > B
Set Vi = exp(−rT ) max(SN − K , 0)
Else
Vi = 0
end
If S̄imax > B
Set V̄i = exp(−rT ) max(S̄N − K , 0)
Else
V̄i = 0
end
bi = 1 Vi + V̄i

Set V
2
end
M M
1 Xb 2 1 X
bi − aM )2 .
Set aM := Vi and bM := (V
M i=1 M − 1 i=1
Variance Reduction by Control Variates
1 The approach of control variate relies upon finding samples that have
some general known correlation.
2 The control variate approach is less generic than antithetic variates, as it
requires some knowledge about the underlying random random variables
in the simulations.
3 However, when it works it can be very powerful.
4 Goal: We wish to estimate E (X ).
5 Suppose: We can somehow find another random variable Y , that is close
to X , with known mean of E (Y ).
6 Then the random variable:
Z = X + E (Y ) − Y
satisfies:
E (Z ) = E (X ) + E (Y ) − E (Y ) = E (X ).
Variance Reduction by Control Variates (Contd ...)
1 Hence we can apply Monte Carlo to Z instead of X .
2 In this context Y is called the “control variate”.
3 Further:
Var (Z ) = Var (X − Y ).
4 Hence to get some benefit from this approach, we would like X − Y to
have a smaller variance than X , which is what is meant by “close”.
5 Note: It may be more expensive to sample Z than to sample X .
6 If Var (Z ) = R1 Var (X ) for some R1 < 1 and the cost of sampling Z is R2
times that of sampling X , then we get an overall gain in efficiency, if
R1 R2 < 1.
1 Generalized case:
Zθ = X + θ(E (Y ) − Y ),
for any θ ∈ R.
2 Since we still have E (Zθ ) = E (X ), therefore we may apply Monte-Carlo to
Zθ .
3 Accordingly:
Var (Zθ ) = Var (X − θY ) = Var (X ) − 2θCov (X , Y ) + θ2 Var (Y ).
4 As θ varies, the value of θ that minimizes this quadratic is given by:
Cov (X , Y )
θmin = .
Var (Y )
5 It can be shown that Var (Zθ ) < Var (X ) if and only if θ lies between 0
and 2θmin .
1 On a general problem we typically do not know Cov (X , Y ) and hence
cannot find θmin .
2 However, it is possible to estimate Cov (X , Y ) and hence θmin during a
Monte-Carlo simulation.
3 The name “control variate” comes from the fact that the term
“E (Y ) − Y ” controls the Monte-Carlo process.
1 Suppose: The covariance is positive, that is
E [(X − E (X ))(Y − E (Y ))] > 0 and θ > 0.
(A) In this case, when X > E (X ), then we would “expect” Y > E (Y ).
(B) Generally, adding the negative amount θ(E (Y ) − Y ) helps to correct
the overestimate of E (X ) from the sample of X .
2 Suppose: The covariance is positive, that is
E [(X − E (X ))(Y − E (Y ))] > 0 and θ > 0.
(A) In this case, when X < E (X ), then we would “expect” Y < E (Y ).
(B) Generally, adding the positive amount θ(E (Y ) − Y ) helps to correct
the underestimate of E (X ) from the sample of X .
3 A similar argument holds when Cov (X , Y ) < 0 and θ < 0.
Control Variates: Arithmetic Average Price Asian Call Option
1 Arithmetic Average Price Asian Call Option: The payoff is:
" n
#
1X
max S(ti ) − K , 0 .
n i=1
2 Geometric Average Price Asian Call Option: The payoff is:


n
!1 
Y n
max  S(ti ) − K , 0 .
i=1
3 Goal: Determine the price of “Arithmetic Average Price Asian Call

Option”.
4 Control Variate: Closed form price of “Geometric Average Price Asian
Call Option”.
Closed Form Price: Geometric Average Price Asian Call Option
1 Define:
σ 2 (n + 1)(2n + 1)
(A) b2 :=
σ .
6n2
1 1 n+1
(B) b := σ 2 + r − σ 2
µ .
2 2 2n
S0 1 2

log K + µ b + 2σb T
(C) db1 := √ .
√σ b T
(D) db2 := db1 − σb T.
2 Time zero price of geometric average price Asian call option:

CGA = e −rT S0 e µbT N (db1 ) − K N (db2 ) .
Primary Classes of Traders
We differentiate between three primary classes of traders:
1 Fundamental (or noise or liquidity) traders: Those who are driven by
economic fundamentals outside the exchange.
2 Informed traders: Traders who profit from leveraging information not
reflected in market prices, by trading assets, in anticipation of their
appreciation or depreciation (May include arbitrageurs, which sometimes is
considered as a separate class).
3 Market makers: Professional traders who profit from facilitating exchange
in a particular asset and exploit their skills in executing trades.
4 (A) Fundamental: Active/Aggressive trading.
(B) Informed: Active/Aggressive trading.
(C) Market makers: Passive/Reactive trading.
Trading in Electronic Markets

A way for people to signal their willingness to trade, and a matching engine to
match those wanting to buy and those wanting to sell.
Orders and the Exchange
An electronic market has two types of orders:
1 Market Order (MO):
(A) These are usually considered aggresive orders that seek to execute a
trade immediately.
(B) Sending an MO, means a trader wants to buy/sell a certain quantity
of shares at the best available price and will (usually) result in an
immediate trade (execution).
2 Limit Order (LO):
(A) These are considered as passive order.
(B) A trader sending in an LO indicates the desire to buy/sell at a given
price up to a certain maximum quantity of shares.
(C) Since the price offered in an LO is usually worse than the current
market price, it will not result in an immediate trade.
(D) Consequently, it will have to wait until either it is either matched
with a new order (and executed) or it is withdrawn (cancelled).
Managing of Orders
A Matching Engine: A Limit Order Book (LBO):

Uses a well defined algorithm that estab- Keeps track of incoming and
lishes when a possible trade can occur, outgoing orders.
and if so, which criterion is going to be
used to select the orders that will be ex-
ecuted.
More on Managing of Orders
(1) Most markets prioritize MO’s over LO’s and then use a price-time priority
as follows.
(2) If a MO to buy comes in, the buy order will be matched with the standing
LO’s to sell in the following way:
(A) First, the incoming order will be matched with the LO’s that offer
the best price (in case of buy order, the sell LO’s with the “lowest
price”).
(B) Then, if the quantity demanded is less than what is on offer at the
best price, the matching algorithm selects the oldest LO’s, the ones
that were posted earliest, and executes them in order, until the
quantity of the MO is executed completely.
(C) If the MO demands more quantity than that offered at the best
price, after executing all standing LO’s at the best price, the
matching algorithm will proceed by executing against the LO’s at the
second best price, the third best price and so on, until the whole
order is executed.
(D) LO’s that have increasingly worse prices are referred to as LO’s that
are deeper in the LOB, and the process by which an entering MO
executes against standing LO’s in the LOB is called
“Walking-the-Book”.
Alternate Exchange Structures
(1) An alternative matching algorithm, such as “pro-rata” rules are used in
some money markets. MO’s are matched against posted LO’s available at
the best prices in proportion to quantities posted i.e., there is no
“time-priority” rule.
(2) Some markets (for example, in case of futures) mix the two, pro-rata and
time-priority.
(3) There are a number of other variations in the way exchanges organize
offers and trades.
(A) Some markets introduce an additional priority to orders coming from
a certain type of trader.
(B) Many exchanges use auctions at particular points in time (such as
starting of trading day/closing of market/after trading halt).
Lecture 40

Aspects of Transparency
1 An important aspect of characterizing exchange is the degree (and cost)
of transparency.
2 In the US there is a clear legal distinction between regulated exchanges
(such as NASDAQ and NYSE) which have specific obligations to publish
information regarding the status of LOB’s AND other electronic markets
(Electronic Crossing Networks (ECNs), dark pools, broker-dealer
internalization).
3 Beyond the legal definitions, we generally distinguish lit (open order book)
from dark markets, based on whether LOB information is publicly
available or not.
Colocation
1 Exchanges also control the amount and degree of granularity of the
information you receive:
(A) You can use consolidated/public feed at low cost.
(B) Pay a relatively much larger cost for direct/proprietary feeds from
the exchanges.
(C) They also monetize the need for speed by renting out
computer/server space next to their matching engines, a process
called “colocation”
2 Through colocation, exchanges can provide uniform service to trading
clients at competitive rates.
3 Further, having traders trading engines at a common location owned by
the exchange simplifies the exchange’s ability to provide uniform service,
as it can control the hardware connecting each client to the trading
engine: Cable (all have the same cable of same length) and Network.
4 This ensures that all traders in colocation have the same fast access and
are not disadvantaged (at least as far as exchange-provided hardware is
concerned).
Figure 1.1
Figure: Figure 1.1

Figure 1.2
Figure: Figure 1.2

Figure 1.3
Figure: Figure 1.3

Artificial LOB: Addition of LO to LOB
(1) Recall: Electronic exchange are fundamentally described by an LOB and a
matching algorithm.
(2) Price-time priority: An incoming LO joins the LOB at the order’s price
and is placed last in the execution queue at that price.
(3) LO’s are displayed as blocks of length equal to their quantities.
(4) LO’s are ordered in terms of time-priority from right to left.
(5) Example: When a buy LO comes in at $23.09 (purple block) it will be
added to the line of blocks, already resting at that price.
(6) The new LO joins the queue at the point closest to the Y-axis, thereby
becoming the third LO waiting to be executed at $23.09.
MO Walks the LOB or is Re-routed
(1) Assume that the venue’s best bid is the best buy quote that the market,
across all venues, currently displays.
(2) A new MO (to sell) 250 shares enters the market (sum of green blocks).
(3) The matching engine goes through the LOB, matching the existing
(posted) LO’s (to buy on the bid side) with the entering MO following the
rules in the matching algorithm.
(4) In the LOB, there are two LO’s at the best bid of $23.09 (represented by
two red blocks), both for 100 units, thereby, totaling 200 units.
(5) The 200 units are executed at the best bid.
(6) Question: The final 50 units depends on the order type and the market it
is operating in.
MO Walks the LOB or is Re-routed (Contd ...)
(1) Standard market: Remaining 50 units will be executed against the LO’s
standing at $23.08, ordered in terms of time-priority (MO will walk the
book).
(2) Summary: MO coming in, is split into 3 blocks, the first 2 are matched
with LO’s at $23.09 and the last with LO’s at $23.08.
(3) Recall that: In the US, there are order protection rules to ensure that
MO’s get the best possible execution and which (depending on the order
type) may require the exchange to re-route the remaining 50 units to
another exchange, that is also displaying the best bid price of $23.09.
(4) In that case, a part of the remaining 50 units (the light blue block) is
re-routed to another venue(s) with liquidity posted at $23.09.
(5) Only after all liquidity at $23.09 in all exchanges are exhausted, can the
remaining shares of the MO return and be executed in this venue, against
any LO resting at (the worse price of) $23.08.
(6) In the example, 25 units were re-routed to alternate exchanges, and 25
units returned to the venue and walked the book.
MO Walks the LOB or is Re-routed (Contd ...)
(1) The MO could, in-principle, be an Immediate-or-Cancel (IOC) order,
which specifies that the remaining 50 shares, that cannot be executed at
the best bid, should be canceled entirely.
(2) Because of order-protection rules in the US (no such rule in European
market) it is seldom observed that in the US, a MO walks the book
straight away.
(3) Rather, it may be observed that a MO is chopped up and executed
sequentially in several markets in a very short span of time.
A Snippet
(1) As depth disappears (for instance, as observed during the Flash Crash of
6th May, 2013), an MO at the end of the sequence, may be executed
against very poor prices, or in the worst case, matched with “stub prices”.
(2) “stub prices” are LO’s at such absurdly low prices that clearly suggests
that they are not expected to be executed (such trades were observed
during the Flash Crash).
(3) Thus LOB serves to keep track of LO’s and supply the algorithm that
matches incoming orders to existing LO’s.
More on LOB (Characteristics)
(1) LOB is defined on a fixed discrete grid of prices (price levels).
(2) Size of the step (difference between one price level and the next) is called
the “tick”.
(3) In the US, the minimum tick size is 1 cent, for all stocks with a price of
above $1.
(4) Other markets may have other ticks (Paris Bourse OR Blosa de Madrid:
¿0.001 to ¿0.05, contingent on the price at which the stock is trading
at).

Combinepdf

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Combinepdf

Uploaded by

Copyright:

Available Formats

MA 592: Mathematical Finance

Prof. Siddhartha Pratim Chakrabarty

1 2002 − 2005 Low interest rates led to a bubble, driven by mortgage

Prof. Siddhartha Pratim Chakrabarty

3 Capital set equal to be at least 8% of the risk-weighted assets.

12.5 × max (VaRt−1 , mc × VaRavg ) + SRC .

Here VaRt−1 is the previous day’s VaR, mc ≥ 3 is the multiplicative

0.08 × [Credit Risk RWA + Market Risk RWA] .

0.08 × [Credit Risk RWA + Market Risk RWA] .

2 Supervisory review and consequent intervention.

where EAD: Exposure at Default, LGD: Loss Given Default, WCDR:

max (VaRt−1 , mc × VaRavg ) + max (sVaRt−1 , ms × sVaRavg ) .

2 The default probabilities can be estimated using several approaches, such

Prof. Siddhartha Pratim Chakrabarty

VaR α (X ) = − inf [x ∈ R|α < FX (x)] .

2 A commonly used approach to estimate VaR is the historical simulation

2 Consequently, its sensitivity to a change in the interest rate is:

3 We see that we cannot extend in a straightforward manner, the analysis

4 It is clear that the sensitivity of a bond to interest rate changes (measured

Prof. Siddhartha Pratim Chakrabarty

3 We will invest a proportion α of the initial capital in the bond with

and therefore, we buy:

Again, the increase in the value of the portfolio approximately matches

D = 0.66 × 1 + 0.34 × 4 = 2.02.

Therefore the bond is more sensitive to an increase in interest rates.

6 Using a Taylor’s Series expansion up to the second-order term, we obtain:

potential impact of extreme and highly unlikely events occurring

FX (x) = P(X ≤ x),

which is right-continuous and non-decreasing.

q α (X ) = inf{x|α < FX (x)},

is called the upper α-quantile of X .

is called the lower α-quantile of X .

q α (X ) = inf{x|α < FX (x)} = inf{FX−1 (α) < x} = FX−1 (α).

VaR α (X ) = −q α (X ) = − inf{x|α < FX (x)}.

4 Let us observe that, since X denotes the gain from an investment,

VaR α (X ) = −q α (X ) = q1−α (−X ),

6 This means that the probability of the loss exceeding VaR α (X ) is no

Prof. Siddhartha Pratim Chakrabarty

q α (X ) = inf {x|α < P(X ≤ x)} ,

4 Hence: VaR α (X ) = −q α (X ) = −xkα .

Index Portfolio value ($000s)

Day Date DJIA FTSE 100 CAC 40 Nikkei 225

Table: Data on stock indices for historical simulation calculation

3 The portfolio therefore has a gain of $21,502 under Scenario 1.

Prof. Siddhartha Pratim Chakrabarty

4 Central Limit Theorem: For large n, Sn behaves like N (nµ, nσ 2 ).

Prof. Siddhartha Pratim Chakrabarty

E∗ [[f (S(T ))|S(0) = S0 + h] − [f (S(T ))|S(0) = S0 ]]

Table: Results of Monte Carlo simulation for exotic option pricing

Prof. Siddhartha Pratim Chakrabarty

2 It can be shown that v 2 = 2σ 4 .

3 It may be noted that I = 2 (actual integral).

2 Using numerical quadrature for numerical integration, we get:

Var (Ybi ) ≈ 0.001073.

I = E (f (U)), where U ∼ U (0, 1),

for some function f .

3 The antithetic alternative is:

Proof of the Lemma

0 ≤ E [(f (X ) − f (Y ))(g (X ) − g (Y ))]

5 Therefore the results follows.

2 Consequently, the antithetic variate is:

Prof. Siddhartha Pratim Chakrabarty

Var (Zθ ) = Var (X − θY ) = Var (X ) − 2θCov (X , Y ) + θ2 Var (Y ).