Sources and Variables

The entirety of the numerical data was extracted from Pro Football Focus, a database consisting of current and historical statistics for every possible football metric. Both basic and advanced stats were gathered from this source. The source had data ranging from statistics for a team’s entire season to the single game performance of individual players, both of which were extremely helpful to the analysis. The data was extrapolated from individual tables for offense and defense. Some examples of offensive statistics include total yards, receptions, yards per carry, etc. The rankings were constructed with data from the 2023 season, since it was the last completed season. In terms of structure of the data set, this source primarily had labeled data. This makes sense because the ranking model requires not only team statistics for each position group but also the team itself for comparison purposes. 

The huge advantage of Pro Football Focus compared to the source of the previous years data, Pro Football Reference, is it’s advanced metrics. Basically, PFF comes up with their own grading system for players based off of qualitative analysis. They watch every play from every player, and give a score to each player based on the outcome of the play. Then, the scores from all of the plays are summed into a final grade for each player. They monitor this throughout the whole season for an end of season grade. This is particularly useful for positions that are difficult to quantify, such as offensive linemen. While advanced metrics were not necessary for every position, they were found as a more reliable way to evaluate players in tough to quantify positions.