Appendix and Bonus Insights
Here, you will find all of the plots, visualizations, and data used to build the rankings. This section will also contain all player rankings at each position, and will even include extra insights that were not included in the report for fun!
Ranking Construction Plots and Data
Here are all the cluster plots, random forest confusion matrices, GINI Coefficient Tables, and Data Tables used to create the rankings for all positions.
Player Rankings
Quarterback
Running Back
Offensive Tackle
Defensive Tackle
Linebackers
Here are snippets of Player Rankings at every position. The button will take you to the full rankings for all position groups, and free agent rankings. Note that the player rankings are based off of the end of season rankings, and they do not reflect the changes in free agency.
Cornerbacks
Wide Receiver/Tight End
Interior Offensive Linemen
Edge Rushers (DE and OLB)
Safeties
Team Position Rankings
These visuals break down the rankings and grades for each individual position. If you want to see the position group rankings, the button below will take you to the results page.
End of Season Rankings
Post Free Agency Rankings
Change in Rankings
Green: Upgraded at Position
Red: Downgraded at Position
White: Neutral
End of Season Grades
Post Free Agency Grades
Change in Grades
Green: Upgraded at Position
Red: Downgraded at Position
White: Neutral
From these rankings, a draft prioritization was also created with the top 5 individual positions. Below is the draft prioritization for this scenario.
The 5 Player Model vs ESPN
For fun, the NFL draft results were compared to the 5 positions model and ESPN’s recommendation. Below are the results.
Offense, Defense, and Overall Rankings
These rankings here sort the teams by their entire offense, defense, and team. These rankings assume that every position group is treated equally. The overall ranking is simply the average of the offense and defense grades.
Offense
Defense
Overall
Model vs Real Life
Model End of Season Rankings
This section shows the model’s projections in division and conference standings, and compares it to what actually happened. In the conference standings, positions 1-4 are reserved for the division winners, while the remainder of the table is sorted from best to worst. Positions 5-7 made the playoffs.
Below the model vs actual comparison is the model’s projection for the 2024 NFL season, not including the draft results.
The model actually performed quite well compared to the actual standings. In the division, it correctly ranked three divisions perfectly and was one change away from three more. It was also able to predict 9 out of the 14 playoff teams. 21 of the model’s predictions also landed within two places of their actual ranking. The model can also identify teams that overachieved and underperformed in the 2023 season. For instance, the New Orleans Saints were projected as a division winner and the Atlanta Falcons were predicted to be the first team out of the playoffs. However, the Tampa Bay Buccaneers won the division despite ranking 11th in the NFC and 3rd in the NFC South. The Saints ended up being the second team out of the playoffs in the NFC and the Falcons were 12th. From this analysis, the Buccaneers overachieved while the Saints and Falcons underperformed. This could be due to coaching performance, as the Falcons coach actually ended up getting fired after the season ended. On the AFC side, the Pittsburgh Steelers and Cleveland Browns overachieved and made the playoffs, and this can be explained by coaching. The Steelers coach has never had a losing season as a head coach, and the Browns coach ended up winning coach of the year.
Now, here is the model’s projected playoff field after free agency and before the draft. Remember that the top four go to the division winners.
Actual NFL Standings
Some interesting statistics over the years with the NFL Standings, and the model’s prediction:
On average, there are 6 new playoff teams from the previous year to the next year.
The model’s projection has 7 teams that will do this.
Those teams are the Atlanta Falcons, New Orleans Saints, New York Jets, Cincinnati Bengals, Las Vegas Raiders, Jacksonville Jaguars, and the Indianapolis Colts.
On average, 4 of the 8 division winners change from the previous year to the next year.
The model’s projection has 4 teams that will do this.
Those teams are the Atlanta Falcons, New York Jets, Las Vegas Raiders, and the Cincinnati Bengals.
In 19 of the past 21 seasons, at least one team that finished last in their division finishes first the next year.
There is one team that does this according to the model. It’s the Cincinnati Bengals.
Considering these parameters and weird nuggets, the model projection produces a hyper realistic scenario. However, there are too many unknown factors to say that this is what will happen. Some of these factors include how the NFL Draft will play out and the hiring of new coaches and rookies. Therefore, don’t take these projections too seriously. The Kansas City Chiefs were considered to be an average team in the end of season rankings, but they went on to win the Super Bowl. The model does not like them as much in 2024 due to lost talent and unknown player contributions, but they have proven success with less. Another example is the Houston Texans’ hiring of DeMeco Ryans as well as the CJ Stroud. They were the worst-to-first team last year, and nobody could’ve projected that a new head coach and rookie quarterback would produced immediate success. This projection will be updated one more time to account for the draft results, so stick around!