There is no argument that the most important position in NFL football is the QB and the ability to affect the QB’s play is paramount in the game. The average EPA per play for a QB drops by 0.3 points per play each time he is pressured and 1.9 points per play for each sack (run stops only account for <0.2 points per play of negative impact). So, a defense’s #1 priority should always be how to pressure the QB (and ideally sack him).
So, it is no secret that the ability of a defense to rush the passer is extremely important to how often an NFL team can win. That is why edge rushers (DEs in even fronts or OLBs in odd fronts) are a valuable commodity in the NFL. The ability of these players to apply pressure to the opposing QB can be a significant factor in winning games. We have seen this in the most recent Super Bowl when the Eagles pass rush was able to pressure Mahomes into multiple sacks that ended drives.
One way to measure an edge rusher’s ability to pressure the QB is by using the Pass Rush Win Rate metric (PRWR %). PRWR is a metric that measures how often a pass rusher beats a block within 2.5 seconds. Most offensive plays are designed to have open receivers within 2.5 seconds, so if a pass rusher can disrupt the QB (pressure) within that time then it can be a huge advantage for the defense.
Since 2015 NFL draft, there have been 198 edge rushers who have played at least 200 plays. The median PRWR of those players is 9.3%, while the top 10% of pass rushers have a PRWR >14.2% and the bottom 25% of pass rushers are at <7.2% PRWR. We will use these values to determine the “elite” pass rushers (>14.2% PRWR) and the “inferior” pass rushers (<7.2% PRWR). (NOTE: all stats are from PFF Premium Stats – highly recommend signing up for PFF Premium Stats).
For the NFL draft, the question always is: does college production translate to NFL production? And, if so, what college metric best translates to NFL success? In reviewing edge rusher performance prediction in the NFL, it became clear that, yes, college production does translate to NFL production, but with one caveat.
First, below, you can see how NFL PRWR correlates to college PRWR.
NFL PRWR % vs College PRWR % (all NFL Edge Rushers drafted since 2015 with >200 plays)

Now, you can see that there is an obvious correlation from college PRWR to NFL PRWR (R-squared is 0.22). Edge rushers that excel getting pressure on the QB in college tend to excel in the NFL. Of the 20 elite pass rushers in the NFL (PRWR >14.2%), 15 of them had college PRWR >16%. Of the 79 college edge rushers with PRWR >16%, 15 of them were elite (19%) and only 7 were inferior pass rushers (9%).
So, PRWR by itself is a decent predictor of NFL success (a 2-to-1 odds of finding an elite player vs a bust or inferior player). However, as an NFL owner, are you happy with those odds? Can we improve them further?
I think yes….now, any analytical model will never have 100% perfect predictive odds. The game of football has too much of a human element with too much complexity.
Yet, I believe my College-to-NFL Edge Rusher model does just that. My model takes into account several statistics – PRWR, PRP (pass rush productivity), hits/hurries/sacks, and similar statistics against “true pass set” (see PFF for all stat definitions). There is one additional variable that is very important for evaluating NFL edge rushers that my final model takes into account – RAS (relative athletic score), which is a measure of how athletic a player is compared to similar players of same size (see link for RAS explanation).
The below chart shows the College-to-NFL Edge Rusher model without taking into account RAS. This model shows a better correlation to NFL edge rusher PRWR than just college PRWR alone. Of the 60 players who had a predicted PRWR >10.5%, 15 of them ended up elite (25%) while 5 were inferior (8%). Our odds of finding an elite pass rusher, while avoiding a bust, has now increased to 3-to-1.
NFL PRWR % vs College-to-NFL Edge Rusher Model Predicted PRWR % (w/o using RAS) (all NFL Edge Rushers drafted since 2015 with >200 plays)

But, something that has been studied significantly is the impact of athleticism for NFL edge rushers. Typically, the most athletic edge rushers have translated well to the NFL. Of the 20 elite pass rushers in the NFL, 16 of them had enough athletic testing to calculate RAS – of those 16, 11 of them had RAS >9.5 (see chart below).
NFL PRWR % vs RAS (all NFL Edge Rushers drafted since 2015 with >200 plays who have enough athletic testing to calculate RAS)

Of the 39 edge rushers who had RAS >9.5, 11 of them were elite in the NFL (28%) and 6 were inferior (15%), so less than 2-to-1 odds of finding an elite player vs a bust (worse than just college PRWR by itself).
So, athletic testing (RAS) is important but a player must first have the college production. However, if we do add in RAS to our College-to-NFL Edge Rusher model, we can see an even better performance of the model (see below).
NFL PRWR % vs College-to-NFL Edge Rusher Model Predicted PRWR % (w/ RAS) (all NFL Edge Rushers drafted since 2015 with >200 plays)

Incorporating RAS into our model improves the odds of finding an elite edge rusher. Of the 40 players who had a predicted PRWR >11% (w/ RAS), 14 of them were elite (35%) and none were inferior (<7.2% PRWR) with 34 of 40 players (85%) with NFL PRWR >9.3% (above average).
We went from a 2-to-1 or 3-to-1 elite-to-bust odds for edge rushers to totally eliminating any odds of a bust and having 2.3-to-1 odds of finding an elite rusher vs above average edge rusher.
Now, you might ask why did I use 10.5% limit in the 1st model (w/o RAS) versus 11% limit in the 2nd model (w/ RAS). RAS has typically added 0.45% to the predicted PRWR for an edge rusher. A RAS of 7.6 adds no additional value to the PRWR, while the average RAS of this sample was 8.8. So, I had to adjust the limits based on how much RAS moved the overall sample. But, this is also a good metric of how much RAS can impact PRWR predictability – take the RAS, subtract 7.6, and multiple by 0.38 then that will tell you how much RAS will move the PRWR that was predicted without using RAS.
So, this model is very, very predictive at finding elite edge rushers, while avoiding any possible inferior players (or “busts”). If you were an NFL owner, wouldn’t you sleep better at night knowing these odds before spending millions of dollars and draft capital on an edge rusher coming out of college?
Now that we have a predictive model – how does the 2025 NFL draft prospects score on this model? How many above-average to elite edge rushers are there (scoring >11% w/ RAS or 10.5% w/o RAS)? That’s the next article.
Leave a Reply