Introduction

This book aims to introduce machine learning (ML) tools to asset managers to help them discover economic and financial theories rather than just back-test potential trading rules. The author emphasizes that successful investment strategies are the specific realization of general theories, and strategies that lack theoretical basis are likely to be wrong.

This book details the advantages of machine learning in finance, including its focus on out-of-sample predictive power, the use of computational methods to avoid unrealistic assumptions, learning complex specifications (including nonlinear, hierarchical, and discontinuous interaction effects in high-dimensional spaces), and decoupling between variable search and specification search to be robust to multicollinearity and other substitution effects. The main mathematical techniques and applications covered in the book include core concepts and practical methods such as Denoising and Detoning, Distance Metrics, Optimal Clustering, Financial Labels, Feature Importance Analysis, Portfolio Construction, and Testing Set Overfitting. This book also provides Python code snippets and empirical analysis of Monte Carlo experiments, aiming to help readers understand the theory and apply it to asset management practice.

Main features

• Solid theoretical foundation
• Rich practical cases
• Clear operating instructions
• Suitable for practical applications

Suitable for the crowd

• Quantitative Analyst
• Algorithmic Trader
• Financial Engineer
• Data Scientist