The 199th pick in the 2000 NFL draft is a story cemented in NFL history. That pick, Tom Brady, ended up leading one of the most decorated careers on the way to a future trip to Canton. Brady’s success has seemingly altered prior beliefs on the success rate of late-round draft selections. As the NFL draft coverage moves to day three this year, prepare yourself for mentions of pick 199. Fans love the story of an “underdog” and were drawn to the last selection in last year’s NFL Draft, Brock Purdy.
In terms of production (and frankly most metrics), the Tom Brady draft selection was an anomaly. Comparing the most recent late-round quarterback to experience some measure of success to Brady fails to appreciate Brady’s career. Typically, prospects with an expected draft position in the late rounds are viewed through the lense of usable traits or raw skills that can be developed. After being drafted, a lot of these players help fill out a roster or provide value in the third phase of the game: special teams.
The chart below looks at the average performance percentiles in a player’s first three years based on an Adjusted Plus-Minus (on-field vs. off-field effect to team performance) Above Replacement metric (largely based on the work of Paul Sabin here) converted to points per game where replacement is considered a 10th percentile performer (or a typical member of the practice squad). Ultimately, the hit rate on prospects drafted in a later round decreases, but the hit rate on a quarterback drafted in later rounds decreases at a higher rate given the scarcity of players available each year. The stories of successful late round picks like Dak Prescott or Tom Brady can be categorized as survivorship bias (small instances of degrees of success overlook many instances of nonsuccess) and can alter our perception of late round draft selections.
This chart also illustrates the potential value of taking a quarterback within the first round and why we typically see more quarterbacks taken early in the draft. 69% of high value quarterbacks (90th percentile of points per game) have come in the first round. This is the highest proportion of any position, followed by offensive tackle. High value performers at other positions can be seen in other rounds, but the scarcity at quarterback is why we see teams mortgage future draft capital to move up in the draft to select their next signal caller.
One approach could be to consistently draft a quarterback early on to get more chances at hitting big. However, it is important to urge caution and recognize the difficult task of scouting quarterbacks. First, the comparison space for evaluating quarterbacks that played within their first three years is very small. Second, the quarterback class has the highest volatility both year to year and even within the same draft class (as expressed below in points per play). Similarly, offensive skill positions like wide receiver and running back have high volatility in terms of performance from year to year.
One hypothesis driving the annual variance is a lack of draftable depth or scarcity at the position from season to season. An unusual commonality (small size coincidence) between Tom Brady and Brock Purdy is that both players saw only one quarterback drafted in the first round of their respective drafts. This lack of draftable depth could create situations for a Brock Purdy or Tom Brady to flourish relative to other players at the position.
In studying some of the late-round quarterbacks who have outperformed expectations in the past few years, we observe that year-to-year variance and lack of draftable depth were hypothesized reasons for them to flourish. Dak Prescott was a fourth-round selection as part of the 2016 class, which headlined Carson Wentz, Jared Goff (both had productive rookie contracts), Paxton Lynch, and Christian Hackenburg. Brock Purdy and Bailey Zappe were a part of the 2022 class (only one quarterback drafted in the first round), and Gardner Minshew (would have been listed 21st below) was a part of the 2019 class headlined by Kyler Murray, Daniel Jones, Dwayne Haskins, and Drew Lock.
Although there is a low sample size of quarterbacks selected each year, draftable depth could be worth observing. An added layer of complexity here is that a team’s window for winning, given their surrounding value, may force their hand into drafting a quarterback (or at least gambling on a higher value quarterback) early in the first round.
Eight of the last ten Super Bowls have featured a quarterback on a rookie contract, which hints that it may be worth the gamble to move up to get a quarterback in the first round if you have high surrounding value and depth. That gamble becomes winning the lottery if a team can find a quarterback in the later rounds with a high performance above replacement with the salary of a late round quarterback like Dak Prescott.
In trying to hit on a late round quarterback selection, we can look at the advanced stats for some guidance on what traits to consider. Looking at last year’s Mr. Irrelevant, Brock Purdy, we can build a profile to compare Purdy to other players in the upcoming draft class. Time to throw variance, middle of the field throw rate, pressure to sack rate, ADoT, and experience all help inform us on who is poised to be the next underdog quarterback that overachieves in the league.
Although Brock Purdy’s success may be due to year-to-year variance at the position group and a strong team, his first year is worth studying. His final college season adjusted PFF passing grade (adjusted for competition and scheme) had him as a 25th percentile passer, which is concerning on the surface. Players like Justin Herbert and Dak Prescott had lower adjusted passing grades their final season in college than Brock Purdy did, and Zach Wilson had one of the highest adjusted final season passing grades in recent memory. This helps highlight how difficult it is to evaluate the quarterback position using solely advanced metrics.
In traditional quarterback scouting, decision making (and making those decisions quickly) is key. The skill variance gap at the college level makes evaluating how a college quarterback makes decisions more difficult. Getting the ball out quickly may be due to other contextual factors such as offensive scheme, throwing to the first read, having a solid offensive line, or a good surrounding receiving corps.
To better measure this trait, one suggestion is to look at time to throw (in seconds) variance. High variance, in terms of time to throw, may suggest a quarterback is truly reading what the defense is giving them, waiting for a receiver to get open, or evading pressure to make a play.
Brock Purdy had a high time to throw variance in his final college season. Quarterbacks such as Patrick Mahomes and Jalen Hurts had higher time to throw variances in college, and Mac Jones had one of the lowest variances from play to play in his final college season within an offensive juggernaut at Alabama (with other first-round players such as Najee Harris, DeVonta Smith, and Jaylen Waddle). A quick release is important within mechanics, but evaluating time to throw metrics at the college level needs proper context.
Another area that Brock Purdy excelled at in college was his ability to get the ball out over the middle of the field. My colleague, Tej Seth, highlighted that EPA per pass has been highest in the middle of the field in the NFL in his article, The NFL’s Mid-Range Jump Shot. In college, this can be especially important. However, when looking at the college space, this may be more difficult to identify as the rate at which players throw over the middle of the field is higher than that in the NFL.
The mean “out” (direction “left” or “right”) rate in a quarterback’s final collegiate season was 53% of throws (middle of field being 47% as a result). One historical separation of first-round quarterback talent is their higher rate of getting the ball over the middle led to higher passing grades. Brock Purdy threw the ball over the middle 51% of the time, putting him in the 69th percentile of NFL drafted quarterbacks. His nearly even split can, again, highlight his ability to read through progressions and get the ball to his playmakers in space. This ability, along with the 49ers scheme, combined for a recipe for success (again highlighted by Tej Seth here).
Another area that Brock Purdy was better than average at was his ability to evade sacks when in pressure situations. The ability to handle pressure and evade sack situations is something that directly translates at the next level when the game gets faster. Brock Purdy ranked in the 67th percentile amongst quarterbacks drafted since 2015 in his ability to avoid sacks. Patrick Mahomes, Kyler Murray, Kenny Pickett, Gardner Minshew, and Jordan Love were among quarterbacks who ranked higher than him in this category.
Next, Brock Purdy was slightly better than average (58th percentile amongst drafted quarterbacks) with regards to his average depth of target (ADoT) in his final college season. This metric explains how far the player is targeting a receiver down the field. A higher ADoT may suggest a player is looking downfield more, and a lower ADoT may suggest a player is more inclined to dump the ball off.
Lastly, we do not typically want to heavily weight a counting stat such as snaps, but it is worth mentioning in this context. For quarterbacks, snaps at the position can add to an individual’s ability to command or be creative in an offense. Brock Purdy ranked in the ~96th percentile in career passing snaps amongst quarterbacks drafted since 2015. Again, Kenny Pickett, Patrick Mahomes, and Justin Herbert were all players with an immense amount of passing snaps in college.
It is important to be cautious when looking at the raw numbers and attempt to properly contextualize passing snaps. Quarterbacks with a high expected draft position may choose to forgo their final season and declare for the draft causing less total collegiate snaps. Also, solely using passing snaps can miss the additional context of level of opponent faced. For Brock Purdy, ranking high in passing snaps is impressive given he had high quality running backs to give the ball to in David Montgomery (in 2018) and Breece Hall.
Taken together, the ability to read through progressions, evenly distribute the ball across the field, evade pressure, move the ball down field, get the ball to playmakers, and execute meaningful reps at the position are all part of the recipe for success in the NFL.
From all of that, we can build a profile of Brock Purdy to compare to this year’s draft class. Data would suggest Clayton Tune from Houston, who has a similar profile as Brock Purdy, is poised to overperform his current later round draft expectations. Tune had a final season time to throw variance of 1.31s (Purdy was at 1.37s and Bryce Young led the class). His final season middle of the field throw rate was 47%, which would put him near the mean for QBs in the chart above. Dorian Thompson-Robinson and Bryce Young were among the best in the 2023 class in getting the ball over the middle of the field.
His final season quarterback pressure to sack rate would put him at 2nd lowest in the 2023 class only behind Anthony Richardson. His near 75th percentile ADoT also rated him as above average in ability to move the ball down the field. His collegiate passing snaps label him as the one most experienced QB in this class (outside of Holton Ahlers from East Carolina). As an added bonus, Clayton Tune had one of the highest final season adjusted PFF passing grades compared to top prospects.
Much of the public will be focusing on the order of quarterbacks selected in the 2023 draft. Anthony Richardson, Bryce Young, C.J. Stroud, and Will Levis are all poised to be potential first-round draft selections. Most of the value at the quarterback position has historically come from the first round, and we have seen a steeper curve in performance in later rounds as a result. History would also suggest that variance in terms of quarterback performance may be swinging back to positive performance with a deeper draftable class.
First round quarterbacks generally possess more readily available abilities, but their expected draft position locks in their future market rate (for good reason given their ability to convert to being high performers at a higher percentage). When targeting a quarterback in later rounds, it is a better strategy to target quarterbacks with a few tangible high quality traits that you are getting at a market rate bargain (i.e., ability to pass over the middle, escape pressure, etc.) and then identify which of their deficiencies will be acceptable or minimalized within that team’s system (e.g., identify quarterbacks who will play more to their strengths in that team’s system and form gameplans to minimize weaknesses).
Regardless, we know the narrative entering the later days of the NFL Draft will be highlighting the next Tom Brady, Dak Prescott, or even Brock Purdy. Look out for Clayton Tune’s name as the next in line for a team hoping to hit on their lottery ticket.