Postseason QB Drive Stats and DVOA

by Scott Kacsmar

Each year since 2014, Football Outsiders has written about playoff drive stats for quarterbacks, continuing work I did before arriving here. With Super Bowl LI complete, we've got another year of results to add to our database. This leaves us with the same 35 quarterbacks that we had in our sample a year ago, but we finally have some new data on Eli Manning and Matt Ryan. Ryan did a lot to change his playoff persona this year, at least among rational fans, even if things did fall apart at the end of Super Bowl LI. Meanwhile, aside from the two improbable Super Bowl runs, Manning has led the Giants offense to just 0.93 points per drive in four one-and-done appearances. For context, the 2016 Rams were dead last this season at 1.17 points per drive.

Drive stats can provide a lot of context into an offense's performance. Great quarterbacks typically lead offenses with great drive stats. The following is an annual update to the quarterback playoff drive stats I first introduced six years ago at Pro Football Reference. You can read last year's edition here. Each year this study grows and it now includes 5,011 drives.

Since these are stats for the whole offensive unit, we always need a warning here.

Disclaimer: While passing stats are not entirely individual stats, the drive stats are even more teammate-focused. There are, of course, drives where the quarterback never drops back and just hands the ball off every play. The entry "Matt Ryan" is another way of saying "2008, 2010-12, and 2016 Atlanta Falcons," and also an abbreviation for Julio Jones, Roddy White, Tony Gonzalez, Harry Douglas, Michael Turner, Alex Mack, Devonta Freeman, offensive coordinator Kyle Shanahan, etc. Drive stats are not adjusted for opponent.

Most of these stats will be familiar to users of Football Outsiders' drive stats -- my inspiration for collecting this data in the first place. Yards per drive still measures net yards, including penalties. As always, kneeldown drives are excluded, and any kneeldown is removed from the drive stats. Generally, only games started by the quarterback are included, but there are some significant backup performances included for Steve Young, Troy Aikman, John Elway, Mark Brunell, Dave Krieg, Bernie Kosar, and Randall Cunningham. Every quarterback with at least six playoff starts is included.

The other exciting part about this year's update is the addition of passing DYAR and DVOA for the 1986, 1987, and 1988 postseasons. We will get into those numbers in the second half of this article.

Drive Stats

Our first table includes general drive stats for things like yards (Yds/Dr), points (Pts/Dr), and starting field position (LOS/Dr) per drive. The average scoring margin at the start of each drive is also included in the final column. The 15 active players are in bold, and the average of the sample is included at the bottom.

Obviously, it would be great if Colin Kaepernick started some more playoff games in his career, because all of his success is tied to the 2012 and 2013 seasons with San Francisco. The prospects of that look at least a little better than Mark Sanchez, another six-playoff game gunslinger, ever returning to relevance. Then again, we just had a postseason where Connor Cook, Brock Osweiler, and Matt Moore started on the same weekend, so really, anything is possible.

The playoffs bring a plethora of possibilities and a drastic reversal of perception if things work out well. Just a year ago, Matt Ryan ranked 20th in yards per drive (30.71) and 30th in points per drive (1.74) in this same 35-quarterback sample. After his best season and best postseason, Ryan is up to fourth in yards per drive (35.24) and seventh in points per drive (2.33). Sure, Atlanta fans will eternally wish that he gained about 10 more yards to set up Matt Bryant for three more points to put away the Super Bowl, but it was a very successful postseason for the MVP. More on that later.

Ryan still has the third-worst average starting field position in this sample, though that is heavily slanted towards the quarterbacks who have started a lot of games since kickoffs returned to the 35 six years ago. Yes, it still feels weird that Neil O'Donnell had the best average starting field position in our sample group, but he also averaged the most plays per drive and played with the biggest leads on average thanks to a couple of blowouts. Sanchez and Alex Smith are the only active quarterbacks to rank in the top 16 for best starting field positon. The touchback change of 2016 will also have an impact on this data going forward, assuming the NFL doesn't reverse the change this offseason. There were 39 drives in this study that started at exactly the 25 this past season. There were 131 such drives in the other seasons combined, so 22.9 percent of the drives starting at the 25 were from the seven veteran quarterbacks studied this postseason.

The next table includes more drive stats. PEN +/- is a per-drive measure of the net difference in penalty yards for the quarterback's offense. So if an offense drew an illegal contact penalty and had a 10-yard holding penalty on the same drive, the net would be -5 for that drive. 3Outs/Dr is the percentage of drives ending in a three-and-out. Please note that drives where an offense ran three plays, gained zero first downs, and then kicked a field goal because they started in ideal field position still count as three-and-out drives. QBTO/Dr refers to the turnovers by only the quarterback, including interceptions and lost fumbles. Not only are the rates of touchdowns per drive shown, but the average distance of each quarterback's touchdown drive is also sorted from longest to shortest.

It makes sense that Smith has the fewest giveaways, especially matched with a high three-and-out rate. However, did you expect Mr. Buttfumble to be No. 2 in that category? Feels like we almost need a Mark Sanchez disclaimer in this column each year, and we haven't even gotten to his absurd DVOA ranking yet. Sanchez has only appeared in two postseasons, and in those six games he had two good full games (vs. 2009 Bengals and 2010 Patriots), two good halves in his AFC Championship Game losses (vs. 2009 Colts and 2010 Steelers), and two poor games where he at least managed game-winning drives (vs. 2009 Chargers and 2010 Colts). The Jets weren't overly productive behind Sanchez in the playoffs, but he kept the turnovers down and took advantage of some favorable field position. Some great quarterbacks on this list do not have the type of numbers you would expect, so it only makes sense that the reverse would be true for some of the lesser players, especially in just six games.

Passing DVOA for 1986-2016

One of the best things about introducing DVOA for another old season like 1988 is that we also have the postseason data. Although we haven't fully written up 1986 and 1987 yet, we were able to include the playoff data from all three of these seasons in this new analysis. That means every John Elway Super Bowl season is now covered. So are all but one of the playoff games with Joe Montana and Jerry Rice together, and every NFL playoff game for Jim Kelly and Warren Moon. This also means we now have DVOA and DYAR for every postseason since I was born. below, We've put together the playoff passing DVOA for all 51 quarterbacks with at least 150 passes since 1986.

(Ed. Note: Full 1987 commentary is taking a little extra time because we want to write up interesting notes on the scab games. It won't be ready for publication until I return from a week of vacation. The current plan is to run 1987 in early March and then 1986 in late March or early April. -- Aaron Schatz)

If you heard his recent interview, then you know that Joe Montana is not ready to crown Tom Brady as the greatest quarterback in NFL history. If one irrationally chose to use playoff performance as the sole tool of evaluation, then Montana would have a great case. Montana's own place in the history of playoff quarterbacks is likely secure at No. 1, because even though our 1986-1988 update adds two of his absolute worst performances, Montana is still at 50.9% DVOA over his last 14 playoff games. (He was at 62.3% DVOA before this most recent update.) In one of the biggest playoff upsets of all time, Montana posted minus-99 DYAR and a minus-70.6% DVOA against the 1987 Vikings despite playing at home. A year earlier, he suffered a minus-42.5% DVOA on 15 passes against the 1986 Giants in a 49-3 rout. Montana did not finish either game, completing a three-game playoff losing streak that started in the 1985 season. Montana will eventually drop again in playoff DVOA after that 1985 loss to the Giants is included, but Montana's first eight playoff games, including a dominant Super Bowl XIX performance against the 1984 Dolphins, will keep him as the playoff DVOA GOAT, or the "PODVOA GOAT" if you will.

As for Brady's 14th and latest postseason, he finished with a 24.0% DVOA, his eighth-best postseason. His 345 passing DYAR ranks third among his own postseasons, bested by 2011 (469) and 2014 (361). In terms of just Super Bowls, Montana blows away Brady and the competition among the 11 quarterbacks with multiple starts since 1986. Montana's Super Bowl DVOA so far is 87.7%, while Brady ranks fifth at 25.4% (a little above his typical playoff level), which ranks outside of the top 10 and sixth among his active peers. If Brady had the dominant playoff performances to back up his team's winning reputation, then Montana (and anyone else) would no longer have an argument, but that has never been the case over his career.

With 31 years of this data, we looked at the top 10 postseasons in passing DYAR and passing DVOA (min. 50 passes).

Montana's 1989 run remains the gold standard, but he was also very good the previous year when the 49ers were less dominant as a whole. Montana is the only quarterback to appear multiple times in both tables.

Then we have Matt Ryan's 2016, which managed to crack the top 10 in both DYAR and DVOA. Again, one more beneficial gain to get Ryan over 500 DYAR (and the Falcons to more points) is about all Atlanta fans would ask from their quarterback, but that just did not happen. Still, it was a very strong postseason for someone who infamously entered this tournament with harsh criticism for a 1-4 playoff record. Just three games later and Ryan is up to eighth in DVOA after he ranked 24th a year ago at 9.1%.

Ryan became the latest regular-season MVP quarterback who failed to win a Super Bowl, but their performances in the playoffs have not been that bad overall. Here's a look at each MVP season since 1986.

Only four of the 24 seasons produced Super Bowl wins, while the most common result was a Super Bowl loss, which happened 10 times. However, only Tom Brady (2010), Aaron Rodgers (2011), and Boomer Esiason (1988) produced negative DVOA, and Esiason's season was unlike any other on this list. The 1988 Bengals led the NFL with 2,710 rushing yards, but Esiason still had a fantastic regular season, ranking first with a DVOA of 37.5% with 3,572 yards on 388 attempts. But in the playoffs, Cincinnati's passing offense was anemic each week. Esiason went 18-of-39 for 202 yards with one touchdown and two interceptions in wins against Seattle and Buffalo. (Yes, those were his combined numbers for the two games, not his averages.) In the Super Bowl, he only threw for 144 yards and took five sacks. His DVOA was -43.5% in the Super Bowl, which was better than what it had been against Seattle (-57.6%) or Buffalo (-64.3%). The weather in Cincinnati that January was nothing like the 1981 Freezer Bowl, so this was really one of the greatest disappearing acts by a top passing offense in playoff history.

Conversely, Montana had at least 150 passing DYAR in each of his three playoff games that season. He finished off Esiason's Bengals in that Super Bowl with a classic drive, hitting John Taylor for the game-winning touchdown in the final minute. Montana had six straight playoff games with at least 150 passing DYAR in 1988 and 1989.

Speaking of great performances, the top playoff game from 1986 to 1988 belongs to Washington's Doug Williams in the Super Bowl against the 1987 Broncos. Williams had 223 DYAR, including 197 DYAR in the second quarter alone. Williams also had a DVOA of 108.3%. This came a year after the Broncos allowed Super Bowl MVP Phil Simms to post 209 DYAR and a 102.2% DVOA through the air. Two years later in 1989, the Broncos were back, allowing Montana to hit 283 DYAR and 136.3%.

Going back to our postseason passing DVOA table, at least half of the 26 players listed on the left (ranked 1-26) will be in the Hall of Fame one day, including nine already there. Bernie Kosar was a pleasant surprise as someone above 20.0% DVOA, but he often played well in January, only to see Elway and the Broncos rip his team's heart out. As for the lesser achievers, the players listed 27 to 51 on the right side of the table feel mostly right. None of those quarterbacks are in the Hall of Fame. Ben Roethlisberger seems like a lock at this point, and he has plenty of time left to jump up from 28th and join the top 26. Roethlisberger is nestled in very closely with Tony Romo and Dan Marino. Players such as Cam Newton and Andrew Luck can take solace in Ryan's last two postseasons for how quickly things can change here.

We also have a little shake-up at the bottom of the table. Andy Dalton is no longer the bottomless pit of postseason quarterback play. Jay Schroeder actually went 3-2 as a playoff starter, but he completed 45.6 percent of his passes, never had a 200-yard passing game, and managed just 5.01 yards per attempt. He saved some real gems for his two losses in conference championship games: a 20-of-50 stinker against the 1986 Giants in a 17-0 shutout, and a five-interception dud in a 51-3 rout at the hands of the 1990 Bills when he was with the Raiders.

Speaking of the Raiders, I wonder where Jim Plunkett will eventually land in these stats. But we must conquer the 1980s before answering that question.