As shown in the previous post, I have developed a prediction model based on college QB performance utilizing 68 different statistics to predict NFL QB performance (based on EPA per play) with an 86% accuracy. This model accurately predicted controversial draft picks such as Josh Allen, Patrick Mahomes, Lamar Jackson, while also predicting busts, such as Zach Wilson, Dwayne Haskins, and Will Levis.
So, with this new model at hand, can we use this to predict which QBs in the 2025 NFL Draft will be elite, average, or a bust? Many draft experts have said the 2025 NFL Draft has no 1st round talent at QB. Teams picking in the top 10 should just stay away from all of these QBs and instead focus on elite talent at the edge rusher and WR positions.
But, I am here to tell you that they are wrong – there is one QB that is an elite talent in this draft and one QB that also scored high in the College-to-NFL QB prediction model (but that is it, unfortunately).
A quick refresher on the College-to-NFL QB Prediction Model – this model utilizes 68 different statistics (out of over 900) from the PFF database and focuses on statistics that are broken down by play style (play action vs non-play action, blitz vs non-blitz, etc). This allows better QB comparison based on actual plays versus looking at overall statistics, which can be misleading due to offensive systems or how defenses played each team.
Below is how well the model has predicted NFL QB performance since 2015:
NFL EPA Per Play vs Predicted NFL EPA Per Play based on College-to-NFL QB Prediction Model (QBs with at least 150 snaps since 2015)
As you can see from the correlation plot, this model accurately predicts NFL EPA per play for QBs with an R2 of 0.86 – the single best prediction model you will see for NFL QBs (no model will ever be 100% for a complex human system such as football).
So, how did the 2025 NFL QB Draftees do in this model? Here is the results and some explanation on why they scored so well:
Predicted NFL EPA Per Play by Draft Year based on College-to-NFL QB Prediction Model
As you can see from the chart, Cam Ward is the highest scoring QB from the 2025 NFL Draft based on this prediction model. He actually scores so high that he is in the same realm as Patrick Mahomes and Jayden Daniels. We will dig into Cam Ward more later, but the 2nd best QB might surprise some of you.
The 2nd best QB in the 2025 NFL Draft is not the consensus best QB (or 2nd best QB for some), Shedeur Sanders. The 2nd best QB in the 2025 NFL Draft is Jaxson Dart. Dart scores in the same region as Brock Purdy, Jordan Love, and Joe Burrow.
Shedeur Sanders scores close to zero for predicted NFL EPA per play (an average QB). He scores similarly to Aidan O’Connell and between Trevor Lawrence and Justin Fields (those 2 QBs are probably his ceiling and floor). The other interesting thing about the 2025 NFL Draft prediction is that the other QBs (Kyle McCord, Jalen Milroe, Will Howard, Quinn Ewers, and Dillon Gabriel) all score close to Shedeur Sanders.
So, if a team is trying to decide on choosing Sanders in the 1st round, I would recommend passing, select an elite talent at another position, and select one of these other QBs (McCord, Milroe, or Howard) in the later rounds.
Let’s first dive into why Dart scores so much higher than Sanders (we will focus on Ward later).
Dart vs Sanders – why Dart is better
As mentioned in the previous posts, there are some college QB statistics that show better correlation or have higher weighting in the College-to-NFL QB prediction model that allow for the high accuracy. So, how did Dart perform in these versus Sanders?
BTT Rate on Plays >2.5sec vs Pressure-to-Sack Rate against No Blitz (2025 NFL Draftees highlighted)
Two of the top weighted statistics in the prediction model are BTT rate on plays >2.5sec and Pressure-to-Sack rate on plays without a blitz. In this chart, you want to be in the top left of the chart for best performance in the predicton model (assuming you score well in the other 66 statistics).
Reviewing these 2 statistics, you can see Dart outperforms Sanders significantly in BTT rate (almost 2x) and also has lower Pressure-to-Sack rate. These statistics shows Dart’s ability to make plays down the field when the play breaks down while also avoiding those costly sacks when defenses are just rushing four (no blitz).
Accuracy % on Throws between 10-19 yards vs Pressure-to-Sack Rate (2025 NFL Draftees highlighted)
As shown in previous posts, accuracy on intermediate throws is the most critical accuracy that translates to NFL QB performance. Now, Sanders has been hyped as the most accurate QB coming out in this draft, yet, he is actually slightly less accurate on these intermediate throws than Jaxson Dart (and also Cam Ward).
Now, overall Pressure-to-Sack rate was the highest correlating overall statistic in college (not broken down by play style) for NFL QB performance. Out of 14 elite NFL QBs (EPA per play >0.1), 13 of them had pressure-to-sack rates below 20% (only Jayden Daniels broke that at 20.2%).
Well, Shedeur Sanders is also at 20.2%, but without Daniels’ BTT rate (over 13% on plays lasting longer than 2.5sec).
So, Dart has a lower Pressure-to-Sack rate, slightly better accuracy on intermediate throws, and a much higher BTT rate. That is why Dart scores better on this model than Sanders.
Now, the final analysis of the 2025 NFL Draft – the one elite QB: Cam Ward.
Why is Cam Ward elite?
In this 2025 NFL Draft, there is only one QB who scored in the truly elite realm: Cam Ward.
But, why did Cam Ward score so high? Let’s look at some statistics that explains just how good Cam Ward is:
BTT Rate on Plays >2.5sec vs Pressure-to-Sack Rate against Blitz (2025 NFL Draftees highlighted)
This chart is the same chart above that was used to compare Dart vs Sanders. However, let’s look at Cam Ward. He also scores very well in both of these statistics – he has a very high BTT rate on plays >2.5sec as well as a low pressure-to-sack rate on non-blitz plays (and low overall pressure-to-sack rate).
So, these explain why both Dart and Ward score so high (Dart does have a predicted NFL EPA per play slightly above 0.1), but why does Ward score even higher? Let’s look at a couple of stats that I think explains what makes Ward better than Dart.
TWP Rate against Blitz vs Pressure-to-Sack Rate against Blitz (2025 NFL Draftees highlighted)
In this chart, we are looking at TWP Rate versus Pressure-to-Sack rate on blitz plays (with 2025 NFL Draftees highlighted). Defensive coordinators top decision in calling plays is deciding if they should blitz (send more than 4 rushers) or not. If you blitz, you leave fewer defenders in coverage. If you don’t blitz, you give the QB more time to find open receivers.
So, how QBs perform against the blitz is critical for NFL QB play. In the chart above, you want to be in the bottom left (low TWP rate and low Pressure-to-Sack rate). Cam Ward outperforms both Jaxson Dart and Shedeur Sanders. His pressure-to-sack rate of 17.2% is lower than both QBs while his TWP rate is still near the bottom of the overall distribution (albeit slightly higher than Dart and Sanders). So, defenses forced a turnover-worthy play or sack when blitzing at a lower rate with Ward at QB vs Sanders or Dart.
BTT Rate against No Blitz vs Pressure-to-Sack Rate against No Blitz (2025 NFL Draftees highlighted)
In this chart, we can see that on plays without a blitz, Ward had a very low Pressure-to-Sack rate while also posting the highest BTT rate on plays without a blitz. So, if defenses didn’t blitz, Ward was able to make them pay with big-time throws. If defenses did blitz, then he avoided sacks and turnover-worthy plays.
So, this is why Cam Ward sticks out – he is the type of QB that is nearly impossible to scheme against on defense (think Mahomes, Allen, Jackson).
Now, Dart, the 2nd highest rated QB, has good Pressure-to-Sack rate and average BTT rates on non-blitz plays also. However, his pressure-to-sack rate was on the high side when blitzed at 23.5%. So, teams are able to sack him with blitzes at a high rate, and sacks have a very high negative EPA (-1.87 for each sack). So, this is why Dart scores lower than Ward, although still scoring very well (and definitely worth a high draft pick).
Sanders, however, has a high Pressure-to-Sack rate against the blitz at 26% and also a high Pressure-to-Sack rate against no blitz. He also has a low BTT rate against no blitz at 4.2%. So, if teams blitz him, then he gives up sacks at a high rate. If teams don’t blitz, then he can’t make enough big-time throws. Sure, his TWP rate is low (one of the lowest), but his inability to avoid sacks and not able to make enough big-time throws really limits him in the NFL.
This shows the power of breaking out these statistics by different play styles – it shows how these QBs performed under various plays so that you can compare how QBs did against the blitz/no blitz, pressure/no pressure, play action/no play action, etc. It eliminates offensive play styles and play calling and the type of defenses faced.
This is why the College-to-NFL QB Prediction model is so powerful – it looks at actual QB performance metrics for more direct comparisons and focuses on the metrics that best translate to NFL performance.
This is a powerful model that can allow teams to avoid costly mistakes (like drafting Shedeur Sanders in the 1st round) while not passing on elite QBs (like Cam Ward).
NFL EPA Per Play vs Predicted NFL EPA Per Play based on College-to-NFL QB Prediction Model (QBs with at least 150 snaps since 2015)
Predicted NFL EPA Per Play by Draft Year based on College-to-NFL QB Prediction Model
I hope you enjoyed these posts – the next set of posts will focus on edge rushers in the NFL Draft and we will see just how deep this draft is for those edge players.