Chapter 20: Further Reading - Modeling College Sports

Academic Papers and Research

  1. Massey, Kenneth. "Statistical Models Applied to the Rating of Sports Teams." Honors Thesis, Bluefield College (1997). The foundational work on least-squares power ratings for college sports, introducing the methodology that underpins most modern college rating systems. Massey's ratings remain one of the components of the BCS and CFP selection committee's evaluation framework.

  2. Stern, Hal S., and Barbara Mock. "College Football on Thanksgiving: An Example of a Poisson Bootstrap." The American Statistician (1998). Applies statistical methods to college football prediction, demonstrating the use of bootstrap methods for uncertainty quantification in sports models with small sample sizes.

  3. Harville, David A. "The Use of Linear-Model Methodology to Rate High School or College Football Teams." Journal of the American Statistical Association (1977). A pioneering paper that established the mathematical framework for rating teams using linear models. Directly applicable to the power rating systems discussed in this chapter.

  4. Glickman, Mark E. "Parameter Estimation in Large Dynamic Paired Comparison Experiments." Journal of the Royal Statistical Society, Series C (1999). Extends the Elo and Bradley-Terry frameworks to handle time-varying team strengths and large pools of competitors. Provides the theoretical foundation for dynamic college sports rating systems.

  5. West, Brady T., and Moshe Lamm. "Predicting the Outcome of College Football Games Using the Least-Squares Method." American Journal of Undergraduate Research (2004). An accessible study applying margin-based ratings to college football with evaluation against betting markets. Demonstrates that even simple models can identify systematic edges.

  6. Boulier, Bryan L., Herman O. Stekler, and Sarah Amundson. "Testing the Efficiency of the College Football Point Spread Market." Applied Economics Letters (2006). Empirically tests the efficiency of college football betting markets and finds evidence of persistent inefficiencies, particularly early in the season and for games involving lesser-known programs.

  7. Stern, Hal S. "Who's Number 1? Massey, Colley, and the BCS." CHANCE Magazine (2004). Reviews and compares the mathematical rating systems that were used in the BCS formula. Provides accessible explanations of the Massey, Colley, Anderson-Hester, and other rating methodologies.

Books

  1. Carroll, Bob, Pete Palmer, and John Thorn. "The Hidden Game of Football." University of Chicago Press (1998). While focused primarily on professional football, the analytical frameworks introduced in this book directly inform college football modeling, particularly the concepts of efficiency metrics and opponent adjustment.

  2. Winston, Wayne. "Mathletics: How Gamblers, Managers, and Sports Enthusiasts Use Mathematics in Baseball, Basketball, and Football." Princeton University Press (2009). Includes chapters on college football rating systems with practical examples and clear mathematical exposition. A good starting point for understanding the quantitative foundations of sports rating.

  3. Peabody, Rufus. "Various Interviews and Published Work on College Football Betting." While not a single published volume, Rufus Peabody's publicly shared insights on college football betting methodology represent some of the most practical professional-level analysis available. His work on recruiting as a predictive feature and market inefficiencies is directly relevant to this chapter.

Data Sources

  1. 247Sports Composite Rankings. The primary source for recruiting data, providing individual player ratings and team-level recruiting class rankings. Available at https://247sports.com/. Essential for building the recruiting-based predictive features discussed in this chapter.

  2. Sports Reference (College Football). Comprehensive historical statistics for college football, including game logs, team and player statistics, and draft data. Available at https://www.sports-reference.com/cfb/. The most convenient free source for historical game results.

  3. Massey Ratings. Kenneth Massey's rating system, one of the longest-running college sports prediction systems, with ratings for multiple sports. Available at https://masseyratings.com/. Provides both individual ratings and a comparison of numerous rating systems.

  4. Bill Connelly's SP+ (formerly S&P+). An advanced college football rating system incorporating efficiency, explosiveness, field position, and finishing drives. Available through ESPN. SP+ provides context for the types of features that improve upon simple margin-based ratings.

  5. On3. Recruiting evaluation service that also tracks transfer portal activity, NIL valuations, and roster composition data. Available at https://www.on3.com/. Increasingly important for tracking the financial and talent dynamics of modern college football.

Online Resources and Communities

  1. Football Outsiders' FEI (Fremeau Efficiency Index). An opponent-adjusted efficiency rating for college football that separates scoring efficiency from field position advantage. Available at https://www.footballoutsiders.com/stats/ncaa/fei/overall. Provides a useful benchmark for evaluating custom models.

  2. The Power Rank. Ed Feng's sports analytics site that applies statistical methods to college football prediction. Covers topics from basic margin-based ratings to advanced techniques, with clear explanations of methodology.

  3. Pinnacle Sports Resources. Pinnacle's editorial content on college sports betting covers market efficiency, closing line value, and identifying edges in college markets. Their low-margin college football lines are widely regarded as among the sharpest available.

Methodological References

  1. Park, Juyong, and M. E. J. Newman. "A Network-Based Ranking System for US College Football." Journal of Statistical Mechanics (2005). Applies network theory to the college football ranking problem, demonstrating how the network structure of the schedule affects the accuracy of different rating systems. Particularly relevant for understanding the cross-conference comparison problem.

  2. Kvam, Paul, and Justin Sokol. "A Logistic Regression/Markov Chain Model for NCAA Basketball." Naval Research Logistics (2006). While focused on basketball, this paper introduces a Markov chain approach to sports rating that has been successfully adapted for college football. The approach naturally handles the large team pool and sparse connectivity that characterize college sports.