Risk Management

Lecture 1: Course Introduction

Chen Tong

SOE & WISE, Xiamen University

September 3, 2024

Chen Tong (SOE&WISE) Risk Management September 3, 2024 1 / 64

Syllabus Review: Instructor Info

▶ Instructor: Chen Tong; Email: tongchen@[Link]

▶ Office: B407

▶ Office hours: Tuesday 11:30-13:30 (by appointment)

▶ TA: Yijing Dang; Email: dyj18436066839@[Link]

Chen Tong (SOE&WISE) Risk Management September 3, 2024 2 / 64

Communications: QQ Group

Chen Tong (SOE&WISE) Risk Management September 3, 2024 3 / 64

Course website

▶ [Link]
▶ password: 556677

▶ The course website would be used for downloading slides,

reading materials and uploading finished assignments.
▶ Course related announcements is given in QQ group.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 4 / 64

Grading policy:

▶ The course grade will be determined based on five

components. The tentative plan is summarized as follows,
which may change according to our actual study process:

Component Percentage Notes

Participation 10% Attendance and class performance
Assignments 20% 2-3 assignments
Mid-term exam 30% In-class
Group Presentation 10% Group (5 persons) presentation
Final exam 30% In-class

▶ According to the requirement, we have at least 12 attendance

records.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 5 / 64

Course Description:

▶ The purpose of this course is to enable students to master the

basic concepts and methods of risk management, and to
develop students’ ability in quantitative analysis.

▶ Some modern quantitative methodologies for risk

measurement and management will be introduced, including
some time series forecasting methods and volatility models.

▶ An introduction to financial derivatives is also provided.

▶ Learn to use statistical software (e.g. Matlab) to analyze

financial data, construct and evaluate the risk model.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 6 / 64

Course outline (Tentative):

Section Content Week

1 Introduction to risk management 1-2

2 Probability & Statistics Review 3-4

3 Mean-Variance analysis and CAPM model 5-6

4 Multi-factor Asset Pricing Models 7-8

5 Programming and Data Analysis 9

6 Time Series Model 10-11

6 Valuation and Risk (volatility) models 12-14

7 Financial Derivatives and Risk Hedging 15-16

Chen Tong (SOE&WISE) Risk Management September 3, 2024 7 / 64

About Presentation
▶ Group (5 persons) presentation with PPTs for about 20 mins,
e.g.
▶ Lessons learned from the financial crisis 2007-08
▶ 2012 JPMorgan Chase trading loss in CDS market by trader
"London Whale"
▶ Risks to Chinese banking system when Chinese currency RMB
(CNY) could be freely exchanged
▶ Risk of internet banking (e.g. Alibaba)
▶ Recent news/reports on risk-related issues (e.g. China
Everbright Securities’s glitch August 2013)

▶ Please find a new risk-related topic and talk to TA.

▶ The presentation will be arranged in the first week after

Mid-term exam.
Chen Tong (SOE&WISE) Risk Management September 3, 2024 8 / 64
Software:

▶ Matlab, download from [Link]

Chen Tong (SOE&WISE) Risk Management September 3, 2024 9 / 64

Reading Materials:

Since a single textbook cannot satisfy all of our needs, the lecture
slides will contain all relevant contents. Nonetheless, the following
reading materials would be highly helpful

▶ Elements of Financial Risk Management (2nd Edition), by

Peter Christoffersen.
▶ Risk Management and Financial Institutions (Third Edition),
by John Hull, 2012.
▶ Quantitative Equity Portfolio Management: An Active
Approach to Portfolio Construction and Management, by
Ludwig B. Chincarini and Daehwan Kim, McGral-Hill, 2006.
▶ Time Series Analysis, by James D. Hamilton, 1994.
▶ Options, Futures and Other Derivatives, by John Hull, 2014.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 10 / 64

Course materials:

Lecture slides and additional handouts

▶ All contents covered in this course will be included in the
lecture slides.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 11 / 64

My personal website ([Link]

Chen Tong (SOE&WISE) Risk Management September 3, 2024 12 / 64

My research: Financial Econometrics, Financial Engineering

▶ My research focuses on the modeling of financial volatility,

covariance matrix, high-frequency data analysis, and their
applications in financial engineering.

▶ Some of my main contributions are associated with developing

new high-frequency-based pricing models for VIX derivatives
(e.g., VIX futures and VIX options), and proposing coherent
frameworks for derivatives pricing with time-varying risk
aversion.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 13 / 64

My websites

▶ Official website
▶ [Link]

▶ Personal website
▶ [Link]

Chen Tong (SOE&WISE) Risk Management September 3, 2024 14 / 64

Risk

▶ The American Heritage dictionary, Fourth Edition, defines risk

as "the possibility of suffering harm or loss; danger". In
finance, harm or loss has a specific meaning: decreases in the
value of a portfolio.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 15 / 64

Types of Risk

Chen Tong (SOE&WISE) Risk Management September 3, 2024 16 / 64

Types of Risk

▶ There are always NO clearly defined boundaries of these

different risk categories.

▶ They are often mixed and interacting with each others (e.g.
credit risk & liquidity risk in financial crisis).

▶ "Risk decomposition" & "risk aggregation" are both used in

risk management.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 17 / 64

What is Risk Management?

▶ Risk management is a subject for living with the possibility

that future events may cause adverse effects.

▶ Its main responsibility is to understand the portfolio of risks

that the company is currently taking, and the risks it plans to
take in the future.

▶ It must decide whether the risks are acceptable and; if they are
not acceptable, what action should be taken.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 18 / 64

What is Risk Management?

▶ More recently, derivative dealers have promoted "risk

management" as the use of derivatives to hedge or customize
market-risk exposures.

▶ Risk management is
▶ an on-going process of making risks transparent
▶ to search hidden risks, measure and manage them
▶ cycle of learning and decision making

▶ Our course is mainly concerned with the ways risks are

managed by fund manager/individual investors/ financial
institutions (rather than non-financial corporations).

Chen Tong (SOE&WISE) Risk Management September 3, 2024 19 / 64

Components of Risk Management

▶ Identify and understand the risk:

▶ What can go wrong?
▶ What are the consequences of things going wrong?

▶ Measure the risk:

▶ What is the probability of loss?
▶ How much is loss?
▶ What return is expected for taking this risk?

Chen Tong (SOE&WISE) Risk Management September 3, 2024 20 / 64

Components of Risk Management

▶ Manage the risk:

▶ boils down to making choices: change risk profile or do
nothing?
▶ How to change risk profile?
▶ What-if-analysis (sensitivity analysis)

▶ Risk management is not just reducing risk:

▶ a risk management decision might be to take on more risk
▶ e.g. a bank’s attitude to risk is not passive and defensive; a
bank is actively willing to take risk, because it searches for a
return (higher than risk-free rate) and this does not come
without risk.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 21 / 64

Risk vs Return

▶ Aim of risk management is not just reducing risk but to take

reasonable risk with proper return:

▶ There is a trade-off between risk and expected return (rather

than actual return)

▶ In an efficient financial market, statistically, the higher the

risks taken, the higher return that could be realized

Chen Tong (SOE&WISE) Risk Management September 3, 2024 22 / 64

Risk vs Return

Chen Tong (SOE&WISE) Risk Management September 3, 2024 23 / 64

The Rise and Fall of LTCM

▶ Long-Term Capital Management (LTCM) was a large hedge

fund led by Nobel Prize-winning economists and renowned
Wall Street traders.
▶ LTCM was profitable in its heyday in the 1990s, drawing over
$1 billion of investor capital by promising that its arbitrage
strategy would yield huge returns for investors.
▶ LTCM’s highly leveraged trading strategies failed to pan out
and, with losses mounting due to Russia’s debt default, the
U.S. government had to step in and arrange a bailout to stave
off global financial contagion.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 24 / 64

What can we learn from the fall of LTCM?

▶ See the video.

▶ "There is a gap between the knowledge of high minded

academics and the conditions of the real world!"
▶ But, learning and mastering the concepts and methods of risk
management is always helpful!
▶ Keep modest and respect the market!

Chen Tong (SOE&WISE) Risk Management September 3, 2024 25 / 64

Risk measure

▶ Consider the investment of $1 in N risky assets, with random

returns, r1 , r2 , ..., rN , in next period.

▶ The portfolio return:

rp = ω1 r1 + ω2 r2 + ... + ωN rN

where ωi is the money invested in asset i , with budget

constraint ω1 + ω2 + ... + ωN = 1.

▶ How to measure the risk of this asset?

Chen Tong (SOE&WISE) Risk Management September 3, 2024 26 / 64

Risk measure

▶ Suppose the p.d.f of the return rp is given by f (rp ).

▶ It seems like that a more dispersed distribution is always

associated with higher risk.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 27 / 64

Risk measure 1: volatility

▶ In most cases, we use volatility to measure the risk of an asset

2 2
σp = var(rp ) = E[(rp − µ) ]

which can be expressed as

+∞
2 2
σp = ∫ (rp − µ) f (rp ) drp
−∞

▶ You may care about the portfolio risk σp2 for a given level of
expected return µp = E(rp ).

Chen Tong (SOE&WISE) Risk Management September 3, 2024 28 / 64

Mean-Variance Analysis

▶ Mathematically, we need to solve following problem:

2
min σp
ωi

s.t. E(rp ) = µp , ω1 + ω2 + ⋯ + ωN = 1

▶ Without constraints, the optimal portfolio weights are:

ω = g + bµp

where g and b are functions of µ and Σ, defined by?

Chen Tong (SOE&WISE) Risk Management September 3, 2024 29 / 64

µ and Σ

▶ Denote the expected returns on the N risky assets by:

N×1
µ = E[r ] =∈ R

▶ The riskiness of the N risky assets is summarized by the

covariance matrix given by:

⎡
⎢ var (r1 ) cov (r1 , r2 ) . . . cov (r1 , rN ) ⎤
⎥
⎢
⎢ ⎥
⎥
⎢ cov (r2 , r1 ) var (r2 ) . . . cov (r2 , rN ) ⎥
Σ = var(r ) = ⎢
⎢
⎢
⎢
⎥
⎥
⎥
⎥
⎢
⎢ ⋮ ⋮ ... ⋮ ⎥
⎥
⎢
⎢ ⎥
⎥
⎣ cov (r ,
N 1r ) cov (r N , r2 ) ... var (rN ) ⎦

Chen Tong (SOE&WISE) Risk Management September 3, 2024 30 / 64

Shortcoming of volatility as a risk measure

▶ How good is volatility in measuring the risk of an asset class?

Actually, a pretty good one when the underlying data is
normally distributed (why?).

Chen Tong (SOE&WISE) Risk Management September 3, 2024 31 / 64

Symmetrical & Skewed distribution

3 3
Skewness = E[(rp − µ) /σp ] > 0 (or < 0)

Chen Tong (SOE&WISE) Risk Management September 3, 2024 32 / 64

Normal & heavy-tail distribution

4 4
Kurtosis = E[(rp − µ) /σp ] > 3

Chen Tong (SOE&WISE) Risk Management September 3, 2024 33 / 64

Risk measure 2: Value-at-Risk (VaR)

▶ The α percentage Value-at-Risk ( VaR) of a portfolio is defined

as the largest return such that the probability that the return
on the portfolio over some period of time is less than VaR is α

Prob (rt < VaR) = α

▶ The VaR is commonly used in financial institutions.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 34 / 64

Graphical representation of Value-at-Risk

Chen Tong (SOE&WISE) Risk Management September 3, 2024 35 / 64

Risk measure 3: Expected Shortfall

▶ (Definition) Expected Shortfall (ES) is defined as the expected

value of the portfolio loss given a Value-at-Risk exceedance
has occurred. The conditional Expected Shortfall is defined

ES = E [rp ∣ rp < VaR]

▶ Expected shortfall - also known as tail VaR - combines aspects

of the VaR methodology with more information about the
distribution of returns in the tail.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 36 / 64

A graphical representation of ES

Where VaR asks the question "how bad can things get?", expected
shortfall asks "if things do get bad, what is our expected loss?".

Chen Tong (SOE&WISE) Risk Management September 3, 2024 37 / 64

Section 3: CAPM (Capital Asset Pricing Model)

▶ Adding a risk-free asset, and some assumptions, we have

E (rj ) = rf + βj [E (rm − rf )] .

Here rf is risk-free rate, rm is return of market portfolio.

cov (rj , rm )
βj =
σ 2 (rm )

▶ This course will introduce how to derive, estimate and test

CAPM.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 38 / 64

Section 4: Multi-factor Asset Pricing Models
▶ The CAPM said that there is only one factor

E (rj ) = rf + βj [E (rm − rf )] .

▶ However, variables that have no special standing in

asset-pricing theory show reliable power to explain the
cross-section of average returns.
▶ size: ME, stock price times number of shares
▶ earnings/price (E/P)
▶ book-to-market equity (the ratio of the book value of a firms
common stock to its market value)

▶ Fama and French (1993) three factor model

E (Ri ) − Rf = bi (E (Rm ) − Rf ) + si E(SMB) + hi E(HML)

Chen Tong (SOE&WISE) Risk Management September 3, 2024 39 / 64

Section 4: Multi-factor Asset Pricing Models

▶ Fama and French(2015) proposes a five-factor model for U.S.

returns...

▶ This course will introduce how to construct factor (in

cross-sectional) and test its pricing performance.
▶ Fama-MacBeth Regressions

Chen Tong (SOE&WISE) Risk Management September 3, 2024 40 / 64

Section 6: Time-Series Forecasting (for return)

e.g. Forecasting (single) stock return Et (rt+1 ):

Deliver statistically and economically significant gains by

accommodating model uncertainty and parameter instability.
▶ economically motivated model restrictions;
▶ forecast combination;
▶ diffusion indices;
▶ regime shifts.
▶ time-series models: ARMA(p,q)

Extension to multivariate forecasting model.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 41 / 64

Section 6: Time-Series Forecasting (for volatility)

Forecasting (single) stock return volatility Vart (rt+1 ):

Some univariate volatility models

▶ GARCH model;
▶ Reduced-form volatility model;

Applications?
▶ Compute value-at-risk (VaR);
▶ Compute expected shortfall (ES);
▶ Pricing financial derivatives*;

Chen Tong (SOE&WISE) Risk Management September 3, 2024 42 / 64

Section 6: Multivariate volatility models

▶ Remember that the optimal portfolio weights are given by:

ω = g + bµp

where g and b are functions of µ and Σ. Here Σ is the

covariance matrix of assets.

▶ How to model the covariance matrix Σ?

▶ The key step is to ensure the positive definiteness of Σ.

▶ You will learn some popular multivariate volatility models.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 43 / 64

Recent trends in volatility modeling

▶ Realized Volatility from high-frequency intraday data is the

most important progress in volatility modeling during the past
decade.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 44 / 64

What is Realized Volatility (RV ) ?
▶ Suppose the daily log-return of SPX (S&P 500) is
St 2
Rt = log ∼ F (µ, σt )
St−1
How to estimate the latent volatility ?

▶ If we sample the prices more frequently (e.g. every 5-min),

then we have m observations of log-returns (ri,t ) within one
day,
m
2
RVt = ∑ ri,t
i=1
(m = 78 for 5-min sampling from 9:30 AM to 4:00 PM)

▶ Under some regularity condition, when m → +∞, we have

P 2
RVt −−→ σt

Chen Tong (SOE&WISE) Risk Management September 3, 2024 45 / 64

Realized volatility of SPX

Chen Tong (SOE&WISE) Risk Management September 3, 2024 46 / 64

Advances in multivariate volatility modeling

▶ From Engle(2002), one could use following decomposition for

modeling covariance matrix Σ,

Σ = ΛC Λ

where C is the correlation matrix, and Λ is a diagonal matrix

given by
Λ = diag(σ1 , ⋯, σN )
where σi is the volatility for i-th asset.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 47 / 64

Advances in multivariate volatility modeling
▶ In the bivariate case, we have

1 ρ
C=[ ]
ρ 1

so we could use the Fisher transformation

1 1+ρ
ρ ↦ F (ρ) ≡ log
2 1−ρ

which is a one-to-one mapping from (−1, 1) into R.

▶ Then we could model ρt in an unrestricted way: e.g.

2
F (ρt+1 ) = ω + βF (ρt ) + ut ut ∼ N(0, σ )

Chen Tong (SOE&WISE) Risk Management September 3, 2024 48 / 64

Extend the Fisher transformation

▶ How to extend the Fisher transformation into multivariate

case?

▶ Archakov and Hansen (2021, Econometrica) proposed a new

parametrization of correlation matrices, where positive
definiteness is an innate property.

▶ This parametrization can be viewed as a generalization of

Fisher transformation to higher dimensions.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 49 / 64

Archakov and Hansen (2021, Econometrica)

▶ A unique spectral decomposition for correlation matrix C:

′
⎛ λ1 0 ⋯ 0 ⎞⎛ ξ1 ⎞
⎜ ⎟ ⎜
⎜
′ ⎟
⎟
⋯ ξp ) ⎜ ⎟
0 λ2 0
⎟⎜ ⎟
⎜
⋮
⎟ ξ2
C = ( ξ1 ξ2 ⎜ ⎜ ⎟
⎜
⎜ ⋮ 0 ⋱ 0 ⎟⎜
⎟ ⎜
⎜ ⋮ ⎟
⎟
⎟
⎝ 0 ⋯ 0 λp ⎠⎝ ξp
′ ⎠

that is
N
′
C = ∑ λ k ξk ξk
k=1

and all λi are positive!

Chen Tong (SOE&WISE) Risk Management September 3, 2024 50 / 64

Archakov and Hansen (2021, Econometrica)
▶ Then the logarithm of correlation matrix is defined by:

N
log C = ∑ ξk ξk log(λk )
′

k=1

▶ And the new parametrization of correlation matrix is given by

N(N−1)/2
γ(C) = vecl (log C) ∈ R

where the vecl (⋅) extracts and vectorizes the elements below
the diagonal.

▶ Archakov and Hansen(2021) showed that γ(C) and C is a

one-to-one mapping (so the diagonal elements of log C is
determined by its off-diagonal elements).

Chen Tong (SOE&WISE) Risk Management September 3, 2024 51 / 64

 ▶ To illustrate this parametrization, consider following example

⎛⎡
⎢ 1.0 •
⎢
• ⎤
⎥
⎥ ⎞ ⎛ ⎡ ⎢
⎢
1.0 • • ⎤
⎥
⎥⎞
γ⎜⎢
⎢ ⎥
⎥ ⎟ ⎜ ⎢
⎢ ⎥
⎥⎟
⎜⎢
⎢ 0.8 1.0 • ⎥
⎥ ⎟ = vecl ⎜ ⎢
log ⎢ 0.8 1.0 • ⎥
⎥⎟
⎝⎢
⎢
⎣ 0.0 0.2 1.0
⎥
⎥
⎦⎠ ⎝ ⎢ ⎢
⎣ 0.0 0.2 1.0
⎥
⎥
⎦⎠

⎛⎡⎢
⎢
−0.53 • • ⎤ ⎥
⎥ ⎞ ⎡ ⎢
⎢
1.14 ⎤
⎥
⎥
⎜
= vecl ⎜⎢⎢
⎢ ⎥
⎥ ⎟ ⎢
⎢ ⎥
⎥
⎢ 1.14 −0.57 • ⎥
⎥ ⎟ = ⎢
⎢ −0.13 ⎥
⎥
⎝⎢⎢ ⎥
⎥ ⎠ ⎢
⎢ ⎥
⎥
⎣ −0.13 0.28 −0.03 ⎦ ⎣ 0.28 ⎦

Chen Tong (SOE&WISE) Risk Management September 3, 2024 52 / 64

 ▶ This parametrization is an extension of Fisher transformation:
2 1+ρ
1 ρ
1
log (1 − ρ ) 1
log 1−ρ
log ( )=( 2
1+ρ
2
2 )
ρ 1 1
2
log 1−ρ 1
2
log (1 − ρ )

and we have
1 ρ 1 1+ρ
γ ([ ]) = log
ρ 1 2 1−ρ

Chen Tong (SOE&WISE) Risk Management September 3, 2024 53 / 64

Section 7: What is a Derivative?

▶ A derivative is an instrument whose value depends on, or is

derived from, the value of another asset.

Pt = ft (St )

where Pt is the price of certain derivative, and the St is the

price of underlying asset. The function ft (⋅) are determined by
the type of derivative, the process of underlying asset, and
some related state variables.

▶ Derivatives vs Underlying Assets?

Chen Tong (SOE&WISE) Risk Management September 3, 2024 54 / 64

"Derivatives are financial weapons of mass destruction!"

Chen Tong (SOE&WISE) Risk Management September 3, 2024 55 / 64

Underlying Assets

▶ Stocks

▶ Fixed income

▶ Foreign exchanges

▶ Loans

▶ Corporate bonds

▶ Mortgages

▶ Commodities

▶ Real Estate

▶ ...

Chen Tong (SOE&WISE) Risk Management September 3, 2024 56 / 64

Types of Derivatives

▶ Forwards

▶ Futures

▶ Options

▶ ...

Chen Tong (SOE&WISE) Risk Management September 3, 2024 57 / 64

Why Derivatives are Important?

▶ Derivatives play a key role in transferring/managing risks in

the economy.

▶ Price discovery.

▶ Many financial transactions have embedded derivatives.

▶ The underlying assets could include stocks, currencies, interest

rates, commodities, debt instruments, electricity, insurance
payouts, the weather, etc. (Anything!)

Chen Tong (SOE&WISE) Risk Management September 3, 2024 58 / 64

Introduction to futures contract

▶ A futures contract with futures price Ft and maturity date T .

▶ On maturity date T , the holder of this futures must buy the

stock in price Ft

▶ The payoff on maturity date T

Payoff = ST − Ft

where ST is the terminal price of stock.

▶ How to protect your portfolio when you expect the stock price
decrease in the future?

Chen Tong (SOE&WISE) Risk Management September 3, 2024 59 / 64

Introduction to European call option

▶ A European call option is a contract that the holder has a

right to buy a stock S in strike price K on future date T .

▶ The payoff only occurs on maturity date T

Payoff = max (ST − K , 0)

where ST is the terminal price of stock.

▶ Pricing option? e.g. Black-Shole model...

−r (T −t) Q
Ct = e Et [max (ST − K , 0)]

That is why we call the derivatives traders as Q-quants.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 60 / 64

Recent trend in derivatives trading
▶ VIX was introduced by CBOE to measure the expected stock
market volatility for next month, derived from SPX options.
▶ The correlation between changes of VIX and SPX is up to
−71%, suggesting a diversification benefit by including VIX in
a portfolio.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 61 / 64

Correlation of Changes for VIX and SPX (ρ = −71%)

Chen Tong (SOE&WISE) Risk Management September 3, 2024 62 / 64

Volatility as a tradable asset: VIX Futures and Options

▶ CBOE introduced the VIX futures in 2004 and VIX options in

2006, which enable investors to trade volatility directly.

▶ VIX derivatives have become popular financial tools with

investors. In 2020, combined trading volume in VIX options
and futures is up to 800,000 contracts per day (source:
CBOE).

▶ Pricing VIX derivatives critically relies on modeling of

undeyling SPX volatility. It’s important to find an accurate
volatility model for SPX.

Chen Tong (SOE&WISE) Risk Management September 3, 2024 63 / 64

Daily Trading Volume of VIX Options in 2020

Chen Tong (SOE&WISE) Risk Management September 3, 2024 64 / 64