Let’s cut off the Patriots fans before they can begin typing in Boston accents: the fact that Manning’s 2013 numbers dwarf Brady’s 2013 numbers does not mean Manning’s career >>> Brady’s career. And it doesn’t even mean (although it strongly implies) that Manning was a better quarterback in 2013 than Brady was. There’s no doubt that Denver’s supporting cast, at least on offense, is much better than New England’s. Manning has Brady’s favorite target from last year, Wes Welker, along with Demaryius Thomas, Eric Decker, and Julius Thomas. Brady has dealt with a very inexperienced set of receivers following massive turnover. The Patriots have had to replace Welker, Rob Gronkowski, Aaron Hernandez, and Danny Woodhead with Julian Edelman, 12 games worth of Danny Amendola, 8 games of Shane Vereen (although he’ll be around on Sunday), 7 games of Gronkowski (he won’t be around on Sunday), and Aaron Dobson and Kenbrell Thompkins. Each quarterback is down a star tackle (Ryan Clady, Sebastian Vollmer) but has an All-Pro caliber guard (Louis Vasquez, Logan Mankins).
But whatever the reason for the discrepancy, one conclusion is inescapable: this is not a meeting of equal passing attacks. On one hand, you have one of the greatest passing offenses ever. On the other, you have an above-average passing offense. And that’s the real story. The Broncos averaged 10 more points per game than New England, while Manning (as representative of the Denver passing attack) averaged 2.75 more adjusted net yards per attempt than Brady (as representative of the Patriots passing attack).
Let’s put these passing attacks into proper historical perspective. In 2013, Manning averaged 8.87 Adjusted Net Yards per Attempt, and the league average was 5.87 ANY/A. That means Manning has a Relative ANY/A of +3.00. Brady averaged 6.13 ANY/A, giving him a RANY/A of +0.25. That gap of 2.75 ANY/A is pretty large by conference championship game standards. Of the 871 conference championship games since 1970, only 11 games featured starting quarterbacks with a larger gap in their regular season RANY/A averages. And the quarterback with the much higher average went 10-1 in those games.
Curiously, the largest gap of the 87 games came in the one loss. That was in 1974, when reigning MVP Ken Stabler and the Raiders hosted the Pittsburgh Steelers. Terry Bradshaw was far from a polished quarterback back then — he split time with Joe Gilliam during the season and didn’t turn the corner until 1975 — so on paper, this was a huge mismatch. But in the AFC Championship Game, it was the Steelers running attack that won the game, as Franco Harris and Rocky Bleier carried 47 times for 209 yards, while the Raiders were limited to just 29 yards on 21 carries.
The table below shows the starting quarterback from each of the conference championship games since 19702, and here’s how to read it. In 1974, the Raiders faced the Steelers (for each game, the chart is listed from the perspective of the statistically superior quarterback). Oakland’s quarterback was Stabler, who had a Relative ANY/A of +3.12. Pittsburgh’s quarterback was Bradshaw, who during the regular season had a RANY/A of -0.99. That means Oakland had a +4.12 advantage in ANY/A, and the two quarterbacks combined for an average ANY/A that was 1.06 adjusted net yards per attempt better than average. Despite the apparent passing edge, Oakland lost the game.
A few thoughts before moving on:
- If you sort by the “Avg” column you can see the best and worst matchups. Surprisingly, the very best quarterback duel was the first one: In 1970, John Brodie and Craig Morton finished 1-2 in ANY/A, NY/A, and AY/A. Brodie finished 1st in sack rate and 2nd in completion percentage, while Morton led the league in yards per completion. The second best game was the 1998 NFC title game between Minnesota and Atlanta. Randall Cunningham led the NFL in ANY/A, TD% (you may recall Randy Moss was on this team), and passer rating, while Chris Chandler was third in ANY/A but first in Y/A, thanks to an absurd 16.6 yards per completion average that is the highest of the last 30 years.
- Only two games grade out as below-average passing battles, and it’s not too surprising to see which games they are. The 1979 NFC Championship Game was won 9-0 by the Rams and Vince Ferragamo over Doug Williams and the Bucs, so it lived up to its hype of a battle of bad passing attacks. And in 2008, Ben Roethlisberger and Joe Flacco put up pedestrian numbers but were surrounded by elite defenses. The AFC title game felt more like a bare knuckles brawl than a football game, ultimately won by the Steelers.
- The numbers are the numbers, of course, but it certainly feels wrong to put Manning/Brady in the vicinity of like Manning/Mark Sanchez or Drew Brees/Rex Grossman. But the 2.75 ANY/A edge is significant, and is simply a testament to the dominance of Denver’s passing offense has been. It’s also a reflection of the struggles — which were to be expected — of the Patriots passing offense. And one could argue that ANY/A actually isn’t the right tool for the job and underrates the gap between the Denver passing attack and New England passing offense. The Patriots might look worse, I think, if you used NY/A or passer rating.
But just because one team appears to have a big passing edge doesn’t mean that is how the championship game will unfold. The next table looks at how these championship games actually played out. If you type “Brady Manning” into the search box, it will pull up the two times these two met in the AFC Championship Game. Here’s how to read the line from 2006. That year, Indianapolis played New England and Manning had a Relative ANY/A of +2.55 during the regular season compared to just +0.70 for Brady. Then, during the AFC Championship Game, the Colts had a team ANY/A of 6.1, as a result of going 27/47 for 349 yards with 1 touchdown and 1 interception, and had 3 sacks for -19 yards. New England as a team averaged 5.74 ANY/A in the AFCCG, completing 21 for 34 passes for 232 yards, with 1 TD and 1 INT, and one sack for 6 yards. That means Indianapolis had a slight edge of +0.36 ANY/A in the AFC title game, which of course Indianapolis won. The 2003 AFC title game was another story, as Manning’s Colts were completely dominated in the passing game.
But here’s something interesting. With a sample size of one game, the regular season values are close to meaningless. The correlation coefficient between the ANY/A difference during the regular season and ANY/A difference during the conference championship game is close to zero (0.06, to be precise). Take a look a the same data above, but plotted on a chart. The X-Axis shows the difference between the ANY/A averages of the two teams during the regular season, from the perspective of the statistically superior quarterback/team passing attack. The Y-Axis shows the ANY/A difference during just the conference championship game, from the perspective of the better passer during the regular season.
If there was a strong correlation — i.e., if the better quarterback during the regular season usually played better in the championship game — then the dots would come close to forming a diagonal line from the bottom left to the top right. Instead, the data points are all over the graph:
This isn’t necessarily surprising, as a one-game sample is not very meaningful and many of the teams that had bad passing attacks also had great pass defenses, which would make it harder for the statistically superior quarterback to excel. One good sign for Broncos fans is that if there is any trend to be seen, it’s that the farther out you go on the X-axis, the results do appear to be more consistent with expectation (with the exception of well, quite a few games, with the 1974 AFC Championship Game and two recent AFC title games involving Manning and Brady being notable ones). That cluster of games on the right represents the 10-1 record mentioned earlier. Other games go wildly off script, like the 1989 AFC title game or 2000 NFC Championship. The 1990 AFC Championship Game appeared to be an even match, until Jim Kelly threw for 300 yards on 23 passes and Jay Schroeder threw six interceptions.
Here’s another observation. Just two years ago, Brady was Manning and Flacco was Brady. Entering the 2011 AFC Championship Game, it looked like an uneven fight: Brady averaged 2.54 more ANY/A than Flacco. But in that game, very nearly won by Baltimore, Flacco wound up with much better numbers (to be fair, the Patriots defense was much worse than the Ravens defense).
Manning, as a proxy for the Denver passing offense, appears to have an enormous edge. But in one game, anything can happen. Oh, and about all that Brady/Manning hype? Let’s end with some notes about this rivalry and pretend that quarterback wins are a real thing:
- As you probably know by now, Brady has a 10-4 record against Manning.
- Brady has been the host for nine of the fourteen games. At home, Brady is 7-2. On the road, he is 3-2.
- Since 2005, Brady and Manning are 4-4 against each other. Brady is 3-2 at home and 1-2 on the road.
- Since the game on Sunday is in Denver, it seems relevant to mention that the 10-4 record is about as relevant at this stat: Manning has lost a home game to Brady once in the last ten years.
- Sunday will mark the first road playoff game for New England in 7 years!
- In 10 of the 14 matchups, the team that won the passing battle (by ANY/A) also won the game. But 3 of the 4 games in which the better passing team didn’t win broke in Brady’s favor — which is pretty much the Brady-Manning debate in a nutshell.3
Prediction: Denver 31, New England 26
- I excluded the 1972 NFC Championship Game between Washington and Dallas, because Roger Staubach just threw 20 passes during the regular season that year. I included three other games where one quarterback had fewer than 100 pass plays, but I marked those games with a footnote. [↩]
- Including both games this year, but excluding the 1972 NFC title game. [↩]
- This is Neil’s contribution. [↩]