Machine Learning Methods in Financial Econometrics

Lecturer

Dr Hasan Fallahgoul, Monash University

Synopsis

This course explores the theoretical foundations and practical applications of machine learning techniques in financial econometrics. Students will examine how traditional econometric methods can be enhanced through AI/ML approaches to address challenges in financial modelling, prediction, and analysis. The course balances theoretical frameworks with implementation examples using Python, establishing connections between classical econometric principles and modern computational methods. Students will develop a critical understanding of both the potential and limitations of applying machine learning to financial data, with an emphasis on model interpretation and validation in economic contexts.

Course Overview

Week 1: Foundations of Financial Econometrics and Machine Learning

• Introduction to financial time series properties and challenges
• Review of classical econometric models (ARIMA, GARCH)
• ML fundamentals relevant to financial data (supervised/unsupervised learning)
• Data preprocessing techniques for financial time series
• Lab: Python implementation of basic financial econometric models

Week 2: Machine Learning Models for Financial Prediction

• Linear models and regularization techniques (Ridge, Lasso, Elastic Net)
• Tree-based models (Random Forests, Gradient Boosting)
• Feature engineering and selection for financial variables
• Model evaluation in time series contexts (cross-validation strategies)
• Lab: Predicting financial returns with ensemble methods

Week 3: Deep Learning for Financial Time Series

• MLPs and LSTMs for financial data
• Attention mechanisms and Transformers for market data
• Transfer learning in financial contexts
• Interpretability and explainability of deep models
• Lab: Building and training neural networks for market prediction

Week 4: Theoretical Frameworks and Applications

• Portfolio optimization using machine learning approaches
• Theoretical frameworks for combining econometric models with ML techniques
• Challenges and limitations: model risk, overfitting, and non-stationarity
• Lab: Group projects implementing integrated ML/econometric solutions

Prerequisites