Chapter 18: Further Reading - Modeling the NHL

Academic Papers and Research

  1. Macdonald, Brian. "A Regression-Based Adjusted Plus-Minus Statistic for NHL Players." Journal of Quantitative Analysis in Sports (2011). Pioneering work on isolating individual player contributions in hockey using regression techniques. Provides the foundation for understanding how player-level analytics translate to team-level predictions.

  2. Thomas, Andrew C., Samuel L. Ventura, Shane T. Jensen, and Stephen Ma. "Competing Process Hazard Function Models for Player Ratings in Ice Hockey." Annals of Applied Statistics (2013). Develops a sophisticated framework for modeling scoring as a competing hazards process, with applications to player evaluation and game outcome prediction.

  3. Schuckers, Michael, and James Curro. "Total Hockey Rating (THoR): A Comprehensive Statistical Rating of National Hockey League Forwards and Defensemen Based Upon All On-Ice Events." MIT Sloan Sports Analytics Conference (2013). Presents a comprehensive player rating system that integrates all on-ice events, relevant for understanding how individual player deployment affects team xG.

  4. Pettigrew, Stephen. "Assessing the Offensive Productivity of NHL Players Using In-Game Win Probabilities." MIT Sloan Sports Analytics Conference (2015). Uses win probability framework to evaluate player contributions, analogous to EPA in football. Demonstrates how hockey's continuous play creates unique analytical challenges.

  5. Lopez, Michael J. "How Often Does the Best Team Win? A Unified Approach to Understanding Randomness in North American Sport." Annals of Applied Statistics (2018). Quantifies the role of luck versus skill across the NHL, NBA, MLB, and NFL. Demonstrates that the NHL has the highest ratio of luck to skill among major sports, explaining why process-based metrics (xG) are so much more predictive than outcomes.

Books

  1. Vollman, Rob. "Stat Shot: The Ultimate Guide to Hockey Analytics." ECW Press (2016). Comprehensive introduction to hockey analytics covering Corsi, Fenwick, PDO, xG, and player evaluation. Accessible to non-technical readers while providing sufficient depth for modeling applications.

  2. Awad, Thomas. "Hockey Abstract." Thomas Awad (annually updated). Applied analytics compendium covering team evaluation, player projection, and historical analysis. Useful for benchmarking model outputs against independent evaluations.

  3. Desjardins, Gabriel. "Behind the Net: A History of Hockey Analytics." Various publications (2010--2020). Chronicles the development of hockey analytics from the early Corsi days through the modern xG era, providing context for how the field has evolved and where remaining inefficiencies lie.

Data Sources

  1. NHL API / Hockey-Reference. The NHL provides public APIs for game data, play-by-play events, and shot coordinates. Hockey-Reference (https://www.hockey-reference.com/) offers comprehensive historical statistics. Together, these sources provide the raw data needed for xG model construction.

  2. Natural Stat Trick. Free resource providing team-level and player-level shot metrics, expected goals data, and game state breakdowns. Available at https://www.naturalstattrick.com/. The most widely used public source for xG and shot metric data.

  3. MoneyPuck. Free expected goals model and data resource maintained by Peter Tanner. Available at https://moneypuck.com/. Provides pre-computed xG values, team ratings, and game-level predictions. A valuable benchmark for evaluating custom models.

  4. Evolving Hockey. Advanced NHL analytics resource providing RAPM (Regularized Adjusted Plus-Minus) player ratings, xG data, and goaltender evaluation tools. Available at https://evolving-hockey.com/. Subscription required for full access but the most comprehensive public resource for player-level analytics.

  5. Corsica / HockeyViz. Historical hockey analytics sites that pioneered public xG models and shot location visualizations. While some have been discontinued, their methodology papers and archived data remain valuable references for model construction.

Online Resources and Communities

  1. The Athletic's Hockey Analytics Coverage. Regular applied analytics articles by Dom Luszczyszyn, Shayna Goldman, and others. Covers team evaluation, player projection, and model-based analysis. Subscription required. Luszczyszyn's "Game Score" model is particularly well-documented and influential.

  2. JFresh Hockey (@JFreshHockey). Prolific data visualization and analytics account providing team and player-level charts, xG analysis, and contract evaluation. An excellent source for understanding how public-facing analytics are communicated.

  3. InStatHockey / Sportlogiq. Commercial data providers offering proprietary event-level tracking data with richer features than the public NHL feed (e.g., pre-shot puck movement, zone entries, controlled entries vs. dump-ins). Subscription required but provides features that significantly improve xG models.

Betting Market Resources

  1. Pinnacle Sports Betting Resources. Pinnacle's editorial content covers NHL market efficiency, closing line value, and how sharp bettors approach hockey. As a reduced-juice book, Pinnacle's closing lines serve as the benchmark for market sharpness.

  2. Unabated. Line-shopping platform tracking opening lines, line movement, and market consensus across sportsbooks. Essential for identifying the best available number on NHL games, where line differences between books can be significant.

  3. DailyFaceoff. The most reliable source for goaltender confirmations and projected lineups. Available at https://www.dailyfaceoff.com/. Monitoring this site (or its API feed) is essential for goaltender-based betting strategies that require timely information.

  4. Left Wing Lock / DobberHockey. Resources for tracking NHL line combinations, deployment patterns, and recent usage. Understanding which forwards and defensemen are playing together helps contextualize team-level xG data and identify potential lineup-driven edges.