Post by Funkytown on Apr 7, 2018 10:38:12 GMT -6
This is a piece that I thought our Mr. Reignman would like:
[OC] Adjusted Kicking Score - a New, Better Kicking statistic by Tripudelops
...
Plenty more at the link: https://www.reddit.com/r/nfl/comments/8a9q92/oc_adjusted_kicking_score_a_new_better_kicking/
[OC] Adjusted Kicking Score - a New, Better Kicking statistic by Tripudelops
Introduction
Most positions in football can rack up statistics in multiple categories. Offensive skill positions can record rushing and receiving yards, touches, averages, scores, and more. Defensive skill positions can record tackles, turnovers, sacks, PDs, etc. Conversely, kicking really only has one measurable statistic--made kicks--but it contains all sorts of information that completely goes to waste when simplified down to simple Field-Goal Percentage. Honestly, there's an argument to be made that fantasy points are a more useful barometer for a good kicking season than FG%, since deeper kicks are worth more.
In 2017, NFL kickers converted every attempt from between 18-21 yards out, and made only 50% of their 61-yard attempts. As far as Field-Goal Percentage is concerned, however, both kicks are worth the same. I've been increasingly frustrated with a lack of a weighted option for kicking statistics, so I decided to make my own - Adjusted Kicking Score (AKS).
I used Pro-Football-Reference to gather data on every kick since 2000. PFR is an amazing resource. With this data, I learned some cool stuff about kicking tendencies over time.
Most positions in football can rack up statistics in multiple categories. Offensive skill positions can record rushing and receiving yards, touches, averages, scores, and more. Defensive skill positions can record tackles, turnovers, sacks, PDs, etc. Conversely, kicking really only has one measurable statistic--made kicks--but it contains all sorts of information that completely goes to waste when simplified down to simple Field-Goal Percentage. Honestly, there's an argument to be made that fantasy points are a more useful barometer for a good kicking season than FG%, since deeper kicks are worth more.
In 2017, NFL kickers converted every attempt from between 18-21 yards out, and made only 50% of their 61-yard attempts. As far as Field-Goal Percentage is concerned, however, both kicks are worth the same. I've been increasingly frustrated with a lack of a weighted option for kicking statistics, so I decided to make my own - Adjusted Kicking Score (AKS).
I used Pro-Football-Reference to gather data on every kick since 2000. PFR is an amazing resource. With this data, I learned some cool stuff about kicking tendencies over time.
So now that you have a good idea of how kicking is changing over time, let's jump in to the statistic!
Methodology
I compiled data--all field goal attempts for the past five seasons, including playoffs (because why the hell would I ignore playoff attempts? f*cking hell, FG% is so dumb) from PFR. I calculated the average make percentage at every distance.
I sorted the data by make percentage. Generally, the closer you are to the uprights, the more likely you are to successfully kick a field goal, but there are several exceptions (24-yard attempts were only the 13th easiest attempts despite being the 7th closest to the uprights, for example). I then created a quadratic regression that modeled the rate of decline.
Using the rate of decline, I calculated an "Adjusted Value" to each distance, which grows as the difficulty of the kick increases. The Adjusted Value increases at a rate that echoes the regression, which assigns a "fair" value to kicks as they increase in difficulty (again, not distance, but difficulty!). I flipped this data and set aside a "Negative Adjusted Value" for misses, which penalizes more for missing easier kicks (again, all proportional to the rate of decline), leaving me with two sets of adjusted value, one for makes, and one for misses.
I ran each individual player's numbers, and sorted them by distance, just like the data pool, then calculated the difference in make percentage at each distance between the player and the 5-year average. I divided this number by the proportion of attempts for that player that came from that range to normalize the data from player-to-player. This number was multiplied by the appropriate "Adjusted Value" created for that respective distance (based on whether they outperformed the spread from that distance) to get an individual-distance Kicking Score. Any kickers with fewer than 20 attempts in 2017 were not considered. Low sample size is the achilles heel of any stat, and when there are low attempt numbers, each kick contributes more significantly to AKS.
Distance Kicking Scores were compiled and multiplied by a (technically unnecessary) constant, which made the numbers a little nicer to look at. This left us with an Adjusted Kicking Score.
Methodology
I compiled data--all field goal attempts for the past five seasons, including playoffs (because why the hell would I ignore playoff attempts? f*cking hell, FG% is so dumb) from PFR. I calculated the average make percentage at every distance.
I sorted the data by make percentage. Generally, the closer you are to the uprights, the more likely you are to successfully kick a field goal, but there are several exceptions (24-yard attempts were only the 13th easiest attempts despite being the 7th closest to the uprights, for example). I then created a quadratic regression that modeled the rate of decline.
Using the rate of decline, I calculated an "Adjusted Value" to each distance, which grows as the difficulty of the kick increases. The Adjusted Value increases at a rate that echoes the regression, which assigns a "fair" value to kicks as they increase in difficulty (again, not distance, but difficulty!). I flipped this data and set aside a "Negative Adjusted Value" for misses, which penalizes more for missing easier kicks (again, all proportional to the rate of decline), leaving me with two sets of adjusted value, one for makes, and one for misses.
I ran each individual player's numbers, and sorted them by distance, just like the data pool, then calculated the difference in make percentage at each distance between the player and the 5-year average. I divided this number by the proportion of attempts for that player that came from that range to normalize the data from player-to-player. This number was multiplied by the appropriate "Adjusted Value" created for that respective distance (based on whether they outperformed the spread from that distance) to get an individual-distance Kicking Score. Any kickers with fewer than 20 attempts in 2017 were not considered. Low sample size is the achilles heel of any stat, and when there are low attempt numbers, each kick contributes more significantly to AKS.
Distance Kicking Scores were compiled and multiplied by a (technically unnecessary) constant, which made the numbers a little nicer to look at. This left us with an Adjusted Kicking Score.
... and neither is this Chris guy.