One of the fundamental aspects of quarterback style is how aggressive he is throwing the ball down the field. There is a clear difference between newly minted teammates Jameis Winston and Drew Brees. The former pushes the ball down field with reckless abandon while the latter is more methodical, often takes the safe option underneath the defense, and pieces together longer drives. Is one style inherently better than the other? My initial assumption was there is not a better option, but I needed to support it with evidence. To address my hypothesis, I explored the relationship between aggressiveness and effectiveness (in the form of EPA per throw) and visualized the results.
How can we define aggressiveness? There is the tried and true method of using average depth of target (aDOT, the amount of air yards per passing attempt). Another way to conceptualize it is by looking at target depth with respect to the location of the first down marker. An aggressive quarterback theoretically does not just want to throw to the marker, but he wants to generate huge chunk plays well past the marker. On every throw, you can look at the yards required to move the chains and the amount of air yards per throw. The difference in air yards per throw and yards to the first down marker per throw I referred to as yards past the sticks per throw. Using aDOT is a bit more intuitive, at least to me, but yards past the sticks seems to better capture what we are looking for because it accounts for the context supplied by the distance required for a first down. I would prefer to use the more intuitive measure for ease of understanding, so to check if there was a massive difference between the two I looked at their correlation for passers in 2019 with at least 200 attempts.
They align extremely well, to the tune of a 0.952 R-squared (95.2 percent of the variance in yards past the sticks per throw can be explained by variations in aDOT). Since they are so closely aligned, from here on out I will use aDOT.
To further break down how quarterbacks distributed their throws, I looked at the percentage of their total pass attempts that went to receivers at short depths, intermediate depths, and those that went deep. Short passes are those that travel no more than five air yards. Intermediate throws were more than five and at most 20 air yards and deep throws were anything more than 20 yards down the field.
The notorious dink and dunkers Derek Carr, Drew Brees, and Teddy Bridgewater top this list of players who concentrated most of their targets short. Kirk Cousins is a bit of a surprise here given the Vikings propensity for play-action, which are longer developing plays with deeper first and second reads. On the other side of the coin, it should not be surprising to NFL fans that Russell Wilson, Jameis Winston, and Ryan Fitzpatrick loathe to dump the ball off. The two interesting players here are Ryan Tannehill and Dwayne Haskins. Tannehill's placement on this leader-board can be attributed to Arthur Smith's heavy reliance on play-action and their personnel at receiver with A.J. Brown and Corey Davis. Washington last season ran the ball a ton, even when trailing so it appeared on the surface that they were trying to protect Haskins from being exposed. But he seemed to be having none of it by rarely choosing to check down and target receivers just past the line of scrimmage. His reputation coming out of Ohio State would make believe he would be at the top here after he was tasked with throwing a lot of screens and shallow crossers to OSU's yards-after-catch monsters. Haskins will be someone I am keeping a close eye on early in the season. Now, for the other areas of the field:
Under former Seahawks offensive coordinator Darrell Bevell, Matthew Stafford was the most aggressive player pushing the ball down the field (the dichotomy between aggressively throwing the ball and being one of the most run-happy teams on early downs was certainly interesting. Very similar to what went/goes on in Seattle). After being one of the lowest aDOT guys who threw among the most passes in the league year in and year out with play-caller Jim Bob Cooter, Stafford totally reinvented himself in 2019. Look at the gainers and losers in aDOT from 2018 to 2019 (for passers who threw at least 200 passes each season):
The extent to which Stafford lapped the field is comical. Also worth noting Dak Prescott's emphasis on throwing the ball deeper in 2019. He led the league in passes at the intermediate level under new play-caller Kellen Moore. These two examples from last season give credence to the idea that aDOT is both a function of the quarterback and scheme. Quarterbacks aDOTs are relatively consistent year over year, but we see a coaching change can really shake things up. Baker Mayfield and Fitzpatrick are guys who played the entire 2019 season with a new coaching staff (in Baker's case Freddie Kitchens came onboard in 2018 but changed the way he called plays with more 11 personnel in 2019, in Fitzpatricks case he changed teams entirely). Flacco was another team changer and Darnold was a scheme-changer. Among players with large deltas and ostensibly no scheme change, I would point out Goff, Mahomes, Allen (who has been absolutely dreadful throwing the ball downfield), Watson, Rivers, and Lamar Jackson (who to be fair had a small sample of throws in 2018). In general, aDOT is relatively stable year over year and most guys on the extreme ends tend to regress slightly (except if you are Jameis Winston or Derek Carr). As you can see, the big outliers are those aforementioned scheme-changers.
Does aggressiveness have anything to do with overall effectiveness? We can look at the relationship between aDOT and EPA per throw to find out.
The correlation is tiny, only 6.82 percent of the variation in EPA per throw can be explained by aDOT. You might be able to say that, all else being equal, the best quarterbacks are not afraid to uncork deep throws at above average rates. But given the correlation, I would say this can be chalked up to noise. Style does not have anything to do with effectiveness. You can be great without being aggressive, look at Drew Brees the past few years. What really matters is accuracy. Using the completion percentage over expected model from the nflfastR package (expected completion percentage is based on the depth of target, whether or not the throw is to the middle of the field, if the game is in a dome, and a few other factors), I compared the accuracy of a quarterback (in terms of completion rate over expected per throw) and his effectiveness.
Despite David Blough and Haskins' attempts to break the axis scales, we can see a clear relationship between accuracy and effectiveness. 61.5 percent of the variance in EPA per throw is captured by accuracy. For every 1 percent in completion rate over expected a quarterback adds, he should expect to add about 0.05 points per throw, which, when accounting for the league average EPA per throw around 0.07, is substantial.
We see Jackson, Mahomes, Prescott, and Stafford buck the trend after sterling seasons, which we should give them a lot of credit for but also realize they will have to be more accurate this year to sustain effectiveness in 2020. Brees and Cousins, despite dinking and dunking, seem to have "deserved" their success last year as did Tannehill, who did so on a completely different diet of throws. Tannehill had by far his best season in the NFL and was aided by A.J. Brown adding almost nine yards after the catch per reception, by far the highest rate in the league for a wide receiver and something that is not sticky year over year. Rivers had another solid year for the Chargers even with that terrible offensive line. His age is a red-flag, but playing behind a great offensive line in Indianapolis should help stave off some age-related regression. Carr was excellent last year and the Raiders quietly had a fringe top-ten offense by EPA per play. They have a ton invested in their starters at offensive line (the most in terms of cap dollars per over the cap) and added two receivers in the draft in Henry Ruggs and Bryan Edwards. Carr has no guaranteed money left on his deal, so him returning to Vegas in 2021 will be contingent on the Raiders maintaining their offensive gains. Darnold and Matt Ryan should expect some positive regression after not performing up to snuff given their accuracy, though with Darnold you have to be weary of the coaching and scheme.
Finally, for some more insight into potential performance movers, I looked at the largest swings in EPA per throw for passers from 2018 to 2019:
Tannehill and Jackson were on another planet last year. In my post about offenses that should regress next year I focused heavily on the Titans and Ravens. Teams that make that large a jump in scoring rate and red zone efficiency basically never sustain those gains the following year. I think Tannehill is a fine player and Lamar is a great player, but I do not expect either of them to turn into super-charged versions of prime Peyton Manning in terms of effectiveness (definitely not style). The fact that Mahomes was so good last year after losing about 0.06 points per throw is preposterous. His touchdown rate took a step back going from a league-leading 8.6 percent to 5.3 percent and the Chiefs offense went from scoring touchdowns on about 48 percent of their drives to 36 percent. Given the league average touchdown rate for passers is around 4.3 percent and Mahomes is one of the best passers this century, I think he will settle in around his 2019 level of play. Watson is another passer who I expect to reproduce his excellent 2019 results despite losing Hopkins to trade. They still have a solid group of receivers in Houston and at this point he is used to playing behind mediocre-to-terrible offensive lines. As I said before, Matt Ryan should see a bump in performance after a down 2019 season. The Falcons, on the whole, regressed as an offense in 2019. With improved line play (they made substantial draft-day investments last year) and the continued excellence of Julio Jones and Calvin Ridley, Ryan should again be among the best quarterbacks in the league. Goff and the Rams are yet another bounce-back team. I mentioned in the team offense post, Goff had throws where the receiver got tackled at the one yard line and he failed to throw a touchdown after. Goff is often maligned as a game-manager and a pawn for coach Sean McVay, but I think in general public analysis gives too much credit to coaching at the expense of the players on the field. Goff was excellent in 2017 and 2018, I would bet he is closer to that level than his results in 2019 (Rams are at +575 to win that division for what it's worth). Tom Brady, probably even more than Lamar Jackson for his 2019 encore, will be under the public microscope after leaving New England. He was not the same player last year in terms of effectiveness and aggressiveness pushing the ball down field. Many attributed that to his less-than-stellar supporting cast, which I think is fair. Now in Tampa Bay, he will have one of the best collections of receiving talent at his disposable. If he continues to get worse this season, we might have to put the days of him being among the league's best behind us.
No comments:
Post a Comment