CFBNumbers cuts through the noise of preseason hype by admitting a hard truth: ranking college quarterbacks is a statistical nightmare because the gap between a "death machine" defense in Georgia and a FCS opponent is too wide to ignore. Rather than offering a gut-feeling list, the author builds a custom linear mixed-effect model to isolate individual quarterback value from team performance, a method that surprisingly demotes highly touted freshmen in favor of seasoned veterans.
The Problem with Traditional Metrics
The piece begins by dismantling the idea that raw stats tell the whole story. CFBNumbers writes, "There is much debate on how to judge Quarterback performances with numbers, as it is extremely hard to separate their individual abilities from the team around them." This is the foundational premise that elevates the analysis above standard box-score recaps. By acknowledging that a quarterback's efficiency is often a product of the offensive line and the quality of the opponent, the author sets the stage for a more nuanced evaluation.
The author introduces two key metrics: Expected Points Added (EPA) and Completion Percentage Over Expectation (CPOE). The goal, as CFBNumbers puts it, is to find the players "able to complete passes adjusted primarily for distance... and is overall efficient on their passes and rushes." This approach highlights Duke transfer Darian Mensah as a standout, noting that he "takes the crown for EPA/Play + CPOE." This is a bold claim for a transfer, suggesting that his impact will be immediate and measurable regardless of the system he enters.
However, a counterargument worth considering is that these advanced metrics, while sophisticated, still rely on historical data that may not account for sudden scheme changes or coaching adjustments in the upcoming season. A player's past efficiency doesn't guarantee future success if the supporting cast collapses.
Building a Custom Model for 2025
To solve the limitations of existing stats, CFBNumbers constructs a proprietary model designed to answer a single, critical question: "How many points does this QB add to their team?" The author explains the methodology, stating, "I built myself a linear mixed effect model that attempts to predict how many points a QB adds to his team." By treating the quarterback and the opponent as random effects, the model attempts to strip away the noise of varying defensive strengths and isolate the quarterback's true contribution.
The results of this model challenge conventional wisdom regarding the hierarchy of talent. While names like Arch Manning and Dylan Raiola dominate social media hype, the data shows they rank lower due to a lack of sample size. CFBNumbers notes, "Arch Manning didn't even make the graph above because he has less than 100 passes and rushes in his young career." This is a crucial distinction; the model is descriptive of what has happened, not predictive of what might happen. It favors experience, placing established players like Carson Beck at the top.
"We can also look at a QBs point rating and what their 247 composite high school recruit rating was when they were coming into college."
The author uses this data to highlight the value of under-recruited players who have outperformed their initial projections, such as Jalon Daniels and Parker Navarro. This reframes the narrative from "who was the best recruit" to "who has actually delivered value." Critics might note that this heavy weighting on past performance could penalize young talents who are on the verge of a breakout, potentially causing fans to overlook the next star until it is too late.
Conference Strengths and Draft Implications
The analysis shifts to a conference-by-conference breakdown, revealing that the Big 12 leads with ten of the top thirty quarterbacks. CFBNumbers observes, "While the Big 12 and ACC have some proven QB talents, I would argue the Big Ten and SEC have more talents we hope will step up this season." This suggests a potential shift in the power dynamic where unproven talent in major conferences could outperform the established stars in others.
The piece also touches on the correlation between this custom model and Pro Football Focus (PFF) metrics. The author finds that their model "appears to favor QBs with more rushing abilities," whereas PFF captures more of the volume of snaps. This distinction is vital for understanding why certain dual-threat quarterbacks might rank higher in this analysis than in traditional scouting reports.
In the final section, the author offers personal NFL draft rankings, admitting that their gut feeling sometimes diverges from the data. CFBNumbers writes, "I am also higher on Aidan Chiles than the point rating system is, as I believe he has a chance to put everything together with one final year." This admission of human intuition alongside algorithmic rigor adds a layer of credibility, showing that the author understands the limits of data.
Bottom Line
The strongest part of this argument is the transparent methodology that prioritizes experience and isolates individual impact, effectively neutralizing the bias toward high-profile freshmen. Its biggest vulnerability is the inherent risk of relying on past performance to predict a volatile season where coaching changes and injuries can upend even the most robust statistical models. Readers should watch to see if the "proven" veterans hold their ground or if the data's skepticism toward the new generation proves wrong in real-time.