Professional Documents
Culture Documents
Connor Walsh
October 17, 2021
SPEX 660: Data-Driven Decision-Making
Summary of Recommendations
2
Recommendations
The goal of this project was to assist teams in the National Football League (NFL) with
their evaluation of college quarterback prospects. To accomplish this, all thirty-two current
starting quarterbacks in the NFL grouped together based on the four archetypes assigned to them
by the Madden 21 video game (“Field General”, “Improviser”, “Scrambler”, and “Strong Arm”)
with the primary point of comparison being their passer ratings from the 2020 season. Operating
under the theory that quarterbacks classified under the “Field General” archetype will generate a
higher passer rating in a given season because of their high accuracy ratings on Madden, this
analysis compared the averages of individual attribute ratings in the video game (accuracy, throw
power, throw on the run, play action passing) as well as actual season statistics related to passer
rating.
Upon completion of said analysis however, that theory was proven to be incorrect.
Despite the fact that half of the thirty-two starting quarterbacks fall under the “Field General”
archetype, in 2020 “Improviser” quarterbacks actually generated the highest average passer
rating of 99.68 compared to a 92.62 for those classified as a “Field General”. That is why the
official recommendation from this analysis is for NFL teams to draft quarterbacks who fall under
the “Improviser” archetype. In fact, the last three NFL MVPs (Patrick Mahomes, Lamar Jackson,
interception percentage, completion percentage, and adjusted yards per attempt when evaluating
college prospects. That is because according to the correlation test completed for this analysis,
these were the metrics that shared the closest relationship with overall passer rating.
3
“Improviser” quarterbacks also lead the way in each of these metrics with the exception of
completion percentage.
Strengths/Weaknesses
With these recommendations in mind, the analysis that led to them did have its share of
strengths and weaknesses. The main strength to draw from this process was that the comparison
of each quarterback archetype was supplemented by 2020 season statistics to evaluate their
performances as a group. Analyzing individual attribute ratings from the Madden dataset was
helpful with applying qualitative values to the skillsets of each archetype, but passer rating
serves as a direct indicator of quarterback performance. The use of a correlational test then
further strengthened this analysis because it identified other related statistics that typically lead to
a higher passer rating, providing a blueprint for talent evaluators in terms of what metrics to
On the other hand, the main weakness of this analysis would have to be the sample size
used from the original dataset, although part of that is due to the way NFL rosters are structured.
In the Madden dataset there were 111 quarterbacks listed, but the issue is that there are only
thirty-two starters in the NFL barring any injuries. As a result, less than a third of active
quarterbacks put forth any game statistics, meaning that the rest do not help with this type of
evaluation. What could have been done to mitigate this however is the use of a progression
model that utilized quarterback statistics over the course of multiple seasons.
When completing this kind of analysis, it is important to consider that there are external factors
quarterback out of the draft there are other things that need to be in place in order for that player
to thrive, with the first being organizational stability. Constant turnover in the coaching and
general manager positions leads to a lack of continuity, and as a result some quarterbacks have to
That is why fitting a young quarterback to a coaching staff and system that develop and
accentuate his abilities is so important for them to succeed, though it does not always happen this
way (Leonard, 2021). For example, out of the thirteen quarterbacks selected in the first round of
the NFL Draft between 2011 and 2017, only four had the same head coach for their first four
years. As a result, players drafted for their physical abilities are forced to play in offensive
schemes that are not tailored towards their strengths (Barnwell, 2018). To compound that issue,
many young quarterbacks are then stuck on rosters with a lack of surrounding talent, including
poor blocking by offensive lineman, and underwhelming wide receivers and running backs to
give the ball to. All of these variables are capable of hindering a quarterback’s development,
causing them to come up short of their expected output in terms of passer rating and other related
metrics.
5
Bibliography
Barnwell, B. (2018, April 10). History tells us the NFL is terrible at evaluating Quarterbacks.
Here's what it means for 2018. ESPN. Retrieved September 13, 2021, from
https://www.espn.com/nfl/story/_/id/23039883/history-tells-us-nfl-terrible-evaluating-
quarterbacks-means-2018-draft-prospects#Part.
Leonard, P. (2021, April 27). Hits and misses: Why drafting a STAR QB is so difficult in the
https://www.nydailynews.com/sports/football/ny-nfl-draft-quarterback-assessment-
20210427-6unfpqsd4fbe7gg4kt5qigul5i-story.html.