 | Financial Econometrics and Forecasting
October 25-27, 2010 (Module 1 - Modern Techniques)
October 28-29, 2010 (Module 2 - Advanced Techniques)*
FRANCIS X. DIEBOLD
Over the past twenty-five years, a revolution in financial modeling and forecasting has swept both academic research and the financial services industry. This course surveys both traditional and new methods of forecasting financial markets, their successes and failures and their future potential. Hands-on application using modern forecasting software is an integral part of the course, as are daily detailed afternoon tutorials of cutting-edge research papers.
* the two modules can either be taken separately or in combination
Objectives: The course develops an appreciation and understanding of methods of modeling and forecasting the fundamentals that underlie financial asset returns, the financial asset returns themselves and their volatility and correlation, as well as the pitfalls and opportunities that arise as technologies move forward. The level of the discussion is designed to strike a balance between intuition and mathematical rigor.
Target audience: Professionals in the financial services industry, central banks and international organizations from a variety of backgrounds, including banking, asset management, risk management, insurance and consulting, as well as financial engineers and analysts, economists, managers and statisticians who want to understand and use financial forecasting models.
Fees: The fee for Module 1 and 2 taken together is CHF 6.500 (incl. VAT). If taken separately:
Module 1 – Modern Techniques: CHF 4.700
Module 2 – Advanced Techniques: CHF 3.500
The fee covers tuition, extensive course material (including pre-course readings), lunches and official event when appropriate.
Accreditation: CFA 36 CE credits for Module 1 and 2 taken together. If taken separately, CFA 21 CE credits for the Modern Techniques Module, CFA 15 CE credits for the Advanced Module
COURSE CONTENTS
Module 1 - Modern Techniques - Monday-Wednesday
Key topics: trend; seasonality; cycles;vector autoregressions; volatility; forecast evaluation and combination; model selection; structural change.
Intensive course on the elements of financial forecasting in a variety of contexts, covering most of Professor Diebold’s book, Elements of Forecasting, as well as articles on special topics. All participants will receive Professor Diebold’s book, as well as copies of his lecture slides and articles discussed in the course.
Pre-course required reading: Elements of Forecasting, Thomson, South-Western, Chapters 1-9
- The six considerations relevant in all forecasting situations: The decision environment and loss function, the forecast object, the forecast statement, the forecasting horizon, the information set, the parsimony principle.
- Components models: Modeling and forecasting trend, seasonal and cyclical components.
- Model selection and structural change: Optimizing out-of-sample forecast performance; parsimony principle; AIC, SIC and degrees-of-freedom penalties; diagnosing structural change; recursive estimation and structural change diagnostics; cusum and related procedures.
- Modeling and forecasting volatility and correlation: GARCH and related models; leverage effects; long and short-run variance components; exogenous variables affecting volatility; time-varying market risk premia; fat-tailed conditional densities.
- Multivariate models: Dynamic regression models for fundamentals and returns; explanatory versus forecasting models; vector autoregressions; predictive causality; impulse-response functions
- Backtesting: Measuring and evaluating forecast accuracy; comparing forecast accuracy; testing for differences in forecasting accuracy; forecast encompassing; forecast combination
- The cutting edge: Each day, tutorials of new and cutting-edge research will be given.
Module 2 - Advanced Techniques - Thursday-Friday**
Key topics: State space modeling; Kalman filtering; nonlinear filtering; simulation and Monte Carlo; Markov chain methods; Bayesian methods.
New and intensive course on advanced financial forecasting, introduced from a state space perspective. Covers dozens of cutting-edge models in a unified and powerful state-space framework for measurement, modeling, simulation and forecasting. Covers both classical and Bayes approaches using cutting-edge simulation methods, including Gibbs, Metropolis, and particle filtering. Familiarity with basic matrix algebra, as well as general mathematical maturity, is assumed.
**Note: Professor Diebold encourages all those who took his 5-day course in earlier years to register for this new course. This is all new material, no redundancy with previous years!
- A classical, linear, Gaussian perspective: Conditionally linear and Gaussian state space representations; optimal filtering, smoothing, and prediction with the Kalman filter; likelihood-based analysis; simulation; Monte Carlo and variance reduction methods; numerous applications including dynamic factor models, unobserved components models, integration, cointegration and error correction, structural change and time-varying parameters, and many more.
- A Bayesian, nonlinear, non-Gaussian perspective: A Bayesian interpretation of state space and optimal filtering; nonlinear and non-Gaussian state space representations; optimal filtering, smoothing, and prediction via simulated Bayesian posteriors; Markov chain methods; Gibbs and Metropolis-Hastings; particle filtering; numerous applications including regime-switching and threshold models, stochastic volatility models, dynamic stochastic general equilibrium (DSGE) models, and more.
- The cutting edge: Throughout, several presentations of new and cutting-edge research will be given.
| |
