Modeling NFL Prospective Success
Quick Summary
Analyzed 169 NFL quarterback prospects spanning the past 25 years
Combined college performance, efficiency metrics, rushing production, NFL Combine data, and draft position
Used regression models, clustering, and PCA to identify consistent patterns
Found that passing efficiency, volume, and rushing value matter more than raw athletic testing
Successful quarterbacks tend to have balanced profiles rather than extreme traits
What Actually Predicts NFL Quarterback Success?
Every NFL Draft, teams convince themselves they’ve found their quarterback of the future. And every year, most of them turn out to be wrong.
Even with decades of data, advanced metrics, and countless scouting hours, quarterback success in the NFL remains incredibly hard to predict. That challenge is what motivated this project: can measurable data from college and the NFL Combine tell us anything meaningful about which quarterbacks will actually succeed?
Motivation
A lot of existing research looks at quarterback evaluation from only one angle — either athletic testing at the NFL Combine or college performance metrics. In practice, teams don’t evaluate players in isolation like that. They consider efficiency, production, mobility, awards, draft position, and athletic traits together.
The goal of this project was to combine all of those factors into a single modeling framework and see which variables actually matter when predicting NFL quarterback success.
The Data
I built a dataset of 169 quarterbacks drafted over the past 25 years who had complete data across college statistics, NFL Combine testing, rushing production, awards, and draft position. To keep the outcome definition clear, quarterback “success” was treated as a binary variable based on meaningful NFL achievements such as Pro Bowl or All-Pro selections, major awards, or starting a playoff game.
Only about 22% of quarterbacks in the dataset met this definition of success, which immediately highlights how difficult this problem is.
Approach
I used three main analytical techniques throughout the project:
Regression models to identify which variables were most strongly associated with NFL success
K-means clustering to group quarterbacks into profile archetypes based on their college and combine data
Principal Component Analysis (PCA) to reduce the high-dimensional data into two interpretable dimensions for visualization
Rather than relying on a single model, the goal was to see whether consistent patterns emerged across different methods.
Visualizing Quarterback Profiles
The interactive plot below shows quarterbacks projected onto two principal components derived from their college and combine metrics. PC1 is more closely related to athleticism and rushing ability, while PC2 reflects passing efficiency and volume.
Hover over each point to see individual player details.
Key Patterns
One of the most interesting findings from this visualization is where successful quarterbacks tend to cluster. Instead of sitting at the extremes, successful quarterbacks generally fall near the center of the plot, indicating more balanced profiles.
Pure pocket passers without strong efficiency showed the lowest success rates, while quarterbacks with some level of mobility and efficient passing were more likely to succeed. Extremely athletic or efficiency-only players can succeed, but those outcomes appear more as exceptions than the rule.
What the Models Showed
Across the regression models, a few patterns were consistent:
Combine metrics alone were weak predictors of success
Passing efficiency and volume mattered more than raw totals
Rushing production was a meaningful indicator in the modern NFL
Draft position improved predictive power but did not dominate the models
The best-performing models balanced athleticism, efficiency, and context rather than relying on any single variable.
Limitations
This project isn’t meant to perfectly predict NFL quarterback outcomes. Success at the professional level is heavily influenced by coaching, injuries, scheme fit, and opportunity — factors that are difficult to quantify. The intent of this analysis was not certainty, but better context when evaluating quarterback prospects.
Final Takeaway
There is no single metric that predicts NFL quarterback success. However, combining passing efficiency, rushing value, athletic traits, and draft context provides a much clearer picture than looking at any one factor in isolation.
This project reinforced why quarterback evaluation remains one of the hardest problems in sports — and why thoughtful data analysis can still meaningfully improve the process.
Works Cited
“2024 NFL Scouting Combine Best 40 Times and Results.” Sharp Football Analysis, 5 Mar. 2024,
www.sharpfootballanalysis.com/analysis/2024-nfl-scouting-combine-best-40-time-vertical-jump-broad-jump-bench-press/.
“2025 Ohio State Pro Day Results.” The 33rd Team, 2025,
the33rdteam.com/ohio-state-pro-day-results-2025/.
DraftHistory.com. “Failure Rates for Even the Top Quarterbacks Are Surprisingly High.” DraftHistory,
www.drafthistory.com/index.php/articles/view/failure-rates-for-even-the-top-quarterbacks-are-surprisingly-high.
ESPN. “College Football Awards.” ESPN,
www.espn.com/college-football/awards/_/id/12.
“Michael Pratt NFL Draft Profile and Pro Day Results.” Hero Sports, 15 Mar. 2024,
www.herosports.com/fbs-nfl-draft-michael-pratt-tulane-pro-quarterback-cpcp/.
Meil, Andrew James. Predicting Success Using the NFL Scouting Combine. California State University, Fullerton, 2018. Master’s thesis, ScholarWorks,
https://scholarworks.calstate.edu/downloads/1z40kt64t.
“College Football National Championship History.” NCAA.com,
www.ncaa.com/news/football/article/college-football-national-championship-history.
Pro-Football-Reference, www.pro-football-reference.com/.
“Tyler Shough Draft Profile.” Sports Illustrated, 2025,
www.si.com/college/louisville/football/tyler-shough-draft-profile.
Sports-Reference.com, www.sports-reference.com/cfb/.
Table Studio, table.studio/convert/txt/to/csv.
“Joe Milton III 2024 NFL Draft Combine Results and Scouting Report.” The 33rd Team, 2024,
www.the33rdteam.com/joe-milton-2024-nfl-draft-combine-results-scouting-report-for-new-england-patriots-qb/.
Winchester, Niven, and J. Dean Craig. Predicting the National Football League Potential of College Quarterbacks. AUT Economics Working Paper Series, no. 2020/11, Faculty of Business, Economics and Law, Auckland University of Technology, 2020. EconStor,
https://www.econstor.eu/bitstream/10419/242583/1/aut-ewp202011.pdf.
NFL Draft Buzz, www.nfldraftbuzz.com/.
