Alexander McNeil
Quantitative Methods for Risk Management
Time and Place
SS 2004. Monday 11-12 and Thursday 10-12 in HG D7.2.
Target Audience
This course is an optional specialisation for students of the Master of Advanced Studies in
Finance program and is designed with these students in mind; it is
however open to all students who have the necessary background
knowledge
(see below).
Required Knowledge
This course will build on the course Empirical
Methods for Finance in WS2003-04 and it will be assumed that
students
are familiar with the material in that course. This includes: standard
methods of statistical inference; rudiments of multivariate analysis;
univariate time series analysis (ARMA and GARCH).
The course will be taught in English.
Course Description
We look at mathematical and statistical methods for the problem of
modelling market risk. Typically we consider a large portfolio,
representing the position in risky assets of a financial institution or
one of its sub-units, whose value is impacted by numerous underlying
risk factors such as stock prices, exchange rates, interest rates and
spreads. The problem is to estimate the loss distributiuon for a future
time interval, such as a day or fortnight, and associated risk measures
that describe the tail of the loss distribution. The statistical
problems that arise require techniques in multivariate analysis, time
series and extreme value theory. Of particular interest is the
phenomenon of simultaneous extreme moves in many risk factors and the
models that may be used to capture this behaviour.
In style the course will offer a combination of theory, statistical
methods and practical examples with real data using S-Plus and
S+FinMetrics.
Contents
Slides are downloadable by ETH students.
- Basics of Risk Management
- Standard
Statistical Methods for Market Risk
- Multivariate Models
for Risk Factors: Basics
- Multivariate Models:
Normal Mixtures and Elliptical Distributions
- Financial Time Series
- Conditional Risk
Measurement
- Multivariate Time
Series
- Copulas and Dependence
- Modelling Maxima and Worst Cases
- POT
Method
- Advanced EVT
Practical Examples with S-Plus
A feature of the course will be the integration of practical examples
using S-Plus and S+FinMetrics. This software is installed in the
student
computer rooms in E floor and is also available as part of the Neptun project for installation
on students' own notebook computers.
It is intended that students should be able to reproduce the analyses
demonstrated in the lecture room. For some of the scripts you will need
my extra library of scripts known as QRMLib. For Windows version download the zipped file here, unpack
in the library directory of Splus and attach using the commmand
"library".
- Standard Methods for
Market Risks
- Normal Mixtures
Models (output only; this script cannot be run by students)
- Conditional Risk
Measurement
- VAR (vector autoregressive) Models
- Multivariate GARCH
- Copulas
- Copula Fitting
- Modelling Maxima
- POT Method
Recommended Literature
- McNeil,A.J. & Frey,R. and Embrechts,P. (2004). Quantitative
Risk Management: Concepts, Techniques and Tools. Book in preparation - go to www.math.ethz.ch/~mcneil/book.html
.
- Zivot,E. & Wang,J. (2002). Modeling Financial Time Series
with S-Plus. Insightful + Springer Verlag, New York. Extremely useful book - highly recommended
for this lecture course. Contains many relevant S-Plus examples.