Kicker and Punter Index Methodology
What’s Behind our Numbers?
Dr. Marc A. Rhorer
There are many metrics for rating college football players, with methods that range from merely summaries of opinion to scientific processes. Our ratings are examples of the latter approach, using systematic measurements collected on every kicker and punter’s applicable actions in all FBS college football games. Utilizing the scientific method, which is also a guiding principle business analytics (after all, we are a business school!), we evaluate player performance. These scientific principles create a level playing field for evaluating performance relative to all other players in similar game situations. These are the foundations our model uses to rate players.
How are our ratings different than most? The answer to that question is complex, but by providing an overview and some examples of how we gauge performance of player activities, a clearer understanding of the advantages of our models should emerge. Our algorithms that gauge performance are based on complete kicking and punting data from prior NCAA FBS seasons. In other words, details of every field goal attempt (field position, distance, and success) and punt (field position, yardage, and net yardage [punt yards minus return or touchback distance]) are factored into the process. These data points may have similarities to those of other rating systems (the few that exist), but information about the methodologies of other models is rare. Some rating schemes in the public realm tend to have unclear methodology and rely heavily on opinion of perceived top performers and highly-ranked teams, or readily available game statistics, without further scrutiny. Gathering the data is merely the starting point for our model, and we compile extensive data. For example, in the first two weeks of the 2018 season we amassed over 45,000 individual data points from play-by-play game logs solely related to kicks and punts. Some of these data points required visual verification (by watching game video of plays when official statistics were questionable), so the labor burden of collecting and accurately recording the game data is massive. The following provide sections more details on our kicker and punter rating models.
Now that you are familiar with our data collection process, as well as the verification methods to ensure the numbers used in our model are accurate, let’s reveal more about the models themselves. Without revealing the exact formulas of our algorithms, understanding the factors considered in the ratings will provide support and consensus behind the methodology. The kicking model has fewer components than the punter metric, because there are fewer variables for consideration. For kickers, we factor in the obvious components: field position of the kick attempt, and if the kick is made. A likely difference with our model, compared to others, is the precision in how we award points for field goal attempts. Based on historical data from prior FBS seasons, we know what percent of attempts should be made relative to zones on the field (e.g. 1-10 yard line, 11-20, 21-30, and 31-40, etc.). So, scoring for field goal attempts made or missed is relative to the success of other kickers’ attempts from the same general field position. For example, a 25-yard field goal is expected, so success gains some points, but failure at an action most FBS kickers are expected to make carries a heavy penalty (loss of points in the model). On the other hand, missing a 53-yard field goal, which is extremely challenging for even the best kickers, is penalized much less (with points deducted) in our model, but success at such a long attempt is highly rewarded. Again, this model provides a relative value score of the kicker (and likely success at field goal attempts) to the overall success of the team. When watching televised games this value of a proven, accurate kicker at longer distances is most evident by the location of the green “field goal range” line. Teams with television’s green “field goal range” line at the 30-yard line, or higher, likely will have kickers who potentially rank near the top of our ratings. The ability to hit long field goals is important in our model, but so is accuracy at shorter ranges.
In addition to field goal attempt distance and success, we also factor other variables into our rating model. Two prominent variables are stadium altitude and field goal timing (and impact on game momentum and outcomes). First, let us consider altitude. Excessive altitudes directly impact field goals, as the air is less dense, which provides an atmosphere with less air pressure to push the ball downward with less drag. Though not as relevant at lower altitudes, for the 11 FBS stadiums at least 3,500 feet above sea-level, kickers are likely to benefit from a bonus in terms of distance (from a range of 4-7 extra yards) and trajectory for field goal attempts, when compared to kicks at stadiums nearer, or below, the mean US elevation of 2,500 feet. We control for this elevation bonus in our algorithm.
Another factor of our kicker metric calculation is the timing of made field goals, especially ones that impact game momentum or the final game outcome. For example, a 30-yard field goal made just prior to half-time in a tight game is weighted slightly higher (10%) than the same field goal made in the first quarter. Our rationale for the variance is that scores (or missed opportunities to score) immediately prior to the half-time break have more influence on overall game momentum. Also, they are more difficult to execute due to increased pressure of the situation. Similarly, walk-off, potentially game-winning field goals in the final seconds are weighted substantially more (50% premium if made, and 50% deduction if missed) than the same effort at a less crucial point of the game. In these instances, the kicker holds the outcome of the game in his feet. Also, successful field goals in the final minutes games that narrow the score differential in close game situations (e.g. making it a two-score game or less), also carry more value in our model. All these actions have pre-determined values based on their impacts on games from prior seasons, so the emotions of particular games, teams, and situations are absent from the scoring mix. The counting is uniform, across all players, conferences, and teams, as the application of the scientific method requires.
The scoring range for kicking is not based on averaging (e.g. .500 is not the mean) but is merely a system based on points earned. The season score is a total of the points earned to date. There is not a limit on kicking points. Points are a direct product of the success, or failure, on the total number of kicks and their distances, with other noted factors factored into the metrics.
The punting model has more data components than the kicking metric, but the model and evaluation process is easier to explain. The Boom Rating (our name for the punting score) ranges from 0-1 and conveys a punter’s performance relative to others who kicked in similar situations. A Boom Rating of .500 is the average. A weekly Boom Rating of .750 or higher will likely place a punter among the top five. An end-of-season cumulative Boom Rating of over .610 will probably place a punter among the top 10. Punts are scored primarily based on the net yardage of each punt relative to all other FBS punters with attempts from the same general location on the field. Many other attempts to gauge punting rely on the weak measure of average distance, without regard to relative field positioning of each attempt. For example, consider a punt from the opponent’s 45-yard line, the maximum possible distance is 44 yards (if downed at the 1-yard line), compared to a punt from a player’s own 20-yard line having a potential maximum of 79 yards. Our Boom Rating is relative, in that players in the two scenarios above are awarded points relative to all other punters who have kicked from similar field positions. In addition to field position relevance, we also factor results of the punt play. We only use punt net yards in our metrics, relative to other punters in prior seasons who kicked from the same area on the field. As an example of our scoring method from 2018, week three’s Best Performance winner, Tyler Sumpter of Troy University. Tyler’s week three Boom Rating was .856 on five punts. This rating means that, on average, his five punts, when each was measured against all other punters’ performances from similar field positions in prior seasons, were better than 85.6% of those attempts.
We also factor in a slight variation in points for punts from the precarious locations of deep within the end zone, when the line of scrimmage is at the 10-yard line or less, as historical data tells us that failure or extremely poor performance are more likely outcomes. Punt distance is not directly relevant to scoring, as the maximum score for each punt (a 1.000) is earned by the best net punt yards possible from any field position. For example, a punt from the opponent’s 40-yard line that is downed at the one yard line (a 39-yard punt) earns the maximum (1.000); similarly, a 70-yard punt (from the 10-yard line, downed at the 20) is also worth nearly the maximum (.997). As you can see, the play of the entire special teams unit is a factor in the punter’s overall performance scoring, just as the receivers, running backs, and offensive line greatly impact quarterbacks.
As you can see from these highlights of our metrics, the scientific-based methodology is complex, but most importantly, it is objective. We are a business school, therefore precision in measurement and analysis of quantitative data is a critical component of our ratings process, just as it is in financial modeling, investment analysis, operations management, and most components of business processes. We apply principles of business data analytics to address deficiencies common in many ratings models. Our goal is to use the methods of business analytics to provide reliable and valid rating systems for kickers and punters, positions that are often overlooked but are critical to teams’ success. In this process, we apply business analytic methods creating a rating system that benefits millions of college football fans.
Our Methodology for Best Punter Award
You may think that our award recipient would simply be determined by the player with the highest Boom Rating each week; however, the formula is more complex. While the Boom Rating is a significant factor in our measure of weekly performance, another factor is considered for the award.
After the 2018 season we evaluated our methodology and discovered opportunities to become even more precise with our calculations. To improve on our model we refined our system to produce a more accurate weekly evaluation of punters.
We found quantity matters. While players with a minimum of three punts are eligible for award consideration, high performance is significantly more challenging with increased punt attempts. Examining the 2018 season, we discovered that those with fewer punts were given a slight artifical advantage. To correct this in the 2019 model, we used historical data to develop a methodology that controls for bias regardless of the quantity of kicks.