Preface

"The goal is to win the Super Bowl. Analytics is just one tool to help us get there." — Anonymous NFL General Manager

Why This Book?

The National Football League has undergone a quiet revolution over the past two decades. What began as a handful of teams experimenting with statistical analysis has transformed into an industry-wide embrace of data-driven decision-making. Today, every NFL franchise employs analytics professionals, and the language of football has expanded to include terms like Expected Points Added, Win Probability, and Completion Percentage Over Expected.

Yet despite this transformation, there has been no comprehensive textbook designed to train the next generation of football analysts. Students interested in the field have had to piece together knowledge from blog posts, academic papers, and online tutorials. Practitioners entering NFL front offices often arrive with strong technical skills but limited football-specific training.

This textbook fills that gap.

Our Approach

This book takes a hands-on, code-first approach to football analytics. Every concept is accompanied by working Python code that readers can run, modify, and extend. We believe that understanding comes through doing, and we've structured the material to get readers working with real data as quickly as possible.

At the same time, we don't sacrifice rigor for accessibility. Each analytical method is grounded in proper statistical foundations, with careful attention to assumptions, limitations, and appropriate use cases. Football generates noisy data with small samples, and we address these challenges head-on rather than glossing over them.

The book is organized in a logical progression:

Part I: Foundations establishes the groundwork—the data ecosystem, programming tools, and statistical concepts that underpin everything that follows.

Part II: Player Analytics develops metrics for evaluating individual performance across all positions, from quarterbacks to special teams.

Part III: Team Analytics zooms out to examine offensive and defensive schemes, game theory, and team-level efficiency metrics.

Part IV: Predictive Modeling introduces forecasting methods for games, players, and prospects.

Part V: Advanced Topics explores cutting-edge techniques including tracking data, computer vision, and deep learning.

Part VI: Applications brings everything together with system design and business applications.

Who Should Read This Book

This textbook is designed for undergraduate students who have completed introductory courses in statistics and Python programming. A basic understanding of football rules and gameplay is assumed, though we provide context where necessary.

The book is also suitable for:

  • Graduate students seeking a comprehensive introduction to sports analytics
  • Working professionals transitioning into football analytics
  • Data scientists looking to apply their skills to sports
  • Serious fantasy football players who want to go beyond surface-level analysis

How to Use This Book

Each chapter follows a consistent structure:

  1. Opening Hook: A compelling question or scenario that motivates the topic
  2. Conceptual Foundation: Intuitive explanations before formal definitions
  3. Mathematical Framework: Formulas and derivations at an appropriate level
  4. Python Implementation: Working code with detailed walkthroughs
  5. Case Studies: Applications to real NFL scenarios
  6. Exercises: Problems ranging from basic to research-level
  7. Self-Assessment Quiz: Immediate feedback on comprehension

We recommend working through chapters sequentially, as later material builds on earlier concepts. However, readers with specific interests can jump to relevant sections—each chapter clearly lists its prerequisites.

The exercises are not optional. Football analytics is a skill, and skills require practice. We've included far more exercises than any student could complete, allowing readers to choose problems matching their interests and skill level.

A Note on Data

All analyses in this book use publicly available data, primarily from the nfl_data_py package and related sources. While NFL teams have access to proprietary data that we cannot replicate, the public data is sufficiently rich to demonstrate every concept and technique.

We've made a deliberate choice to use real NFL data rather than simulated datasets. This means readers will encounter the messiness, inconsistencies, and limitations of real-world data—exactly what they'll face in professional settings.

Acknowledgments

This book would not exist without the football analytics community that has openly shared methods, code, and insights over the past two decades. We are particularly indebted to the creators of nflfastR and nfl_data_py, who have democratized access to play-by-play data.

We also thank the countless researchers, analysts, and writers who have advanced our collective understanding of football through careful analysis and public discourse.

Looking Forward

Football analytics is a young and rapidly evolving field. The methods in this book represent our best current understanding, but they will inevitably be refined and extended. We encourage readers to approach the material critically, to question assumptions, and to develop their own improvements.

The goal is not to memorize formulas or replicate our code. The goal is to develop the analytical mindset that allows you to ask good questions, gather relevant evidence, and communicate insights effectively. These skills will serve you well regardless of how the specific methods evolve.

Welcome to football analytics.

The Authors January 2024