Peyton Manning’s 2004 Season Remains the Gold Standard

Considering Peyton Manning’s brilliant start to the 2013 campaign, there has been talk that when the year is over, he may have posted the greatest quarterback season ever.

While there’s no denying Manning’s stellar play through 10 weeks in 2013, the season may not even be the best of his career, let alone NFL history.

Numbers strongly suggest 2004 was Manning’s best season, and unlike his current season, Manning’s 2004 may very well be the best quarterback season of all time.

None of this is to disparage the Broncos quarterback’s current season, as he has been measurably better than his peers. Among quarterbacks with at least 200 attempts this season, Manning leads the league in net yards per attempt, adjusted net yards per attempt, touchdown pass rate, traditional quarterback rating, ESPN’s QBR, defense-adjusted yards above replacement and value over average (via Football Outsiders), Win Probability Added, Expected Points Added, and Success Rate (via Advanced NFL Stats).

Manning is the MVP in the 2013 and anyone who says otherwise isn’t being fair; now time to look at some history.

Before going forward, this discussion must be contextualized in terms of both volume and era. Two quarterbacks may both throw for 4,500 yards in a season; all other things being equal, reaching this total with 400 attempts is more impressive than hitting it on 600 attempts.

Also, presume two quarterbacks throw for 7.0 yards per attempt; the one who reached this average in 1972 deserves much more credit than the one who did this in 2012.

Volume vs. Efficiency

On volume first, it is unfortunate discussions like this one are so often framed by basic, counting stats: who threw for more yards and touchdowns.

Consider Brett Favre, who threw for more yardage, touchdowns and interceptions than anyone else. Is Favre the best quarterback ever? Probably not, and he is certainly not the worst ever.

When we see Favre has also attempted more passes than anyone else, his touchdown, yardage, and interception totals make more sense.

A look at Vinny Testaverde’s career illustrates this point as well. Testaverde is ninth all-time in career touchdown passes, ahead of Joe Montana and behind Johnny Unitas. Testaverde is also eighth in career passing yardage.

No one can seriously suggest Testaverde is of the same caliber as Montana or Unitas, or that he is one of the 10 best quarterbacks ever. Counting stats like touchdowns and passing yards suggest he is.

The real explanation, of course, is that Testaverde played for 21 seasons and is seventh in career passing attempts.

With these examples in mind, it makes much more sense to use rate/efficiency stats, that put production in the context of total attempts. This is true when looking at both quarterbacks’ careers and single seasons.

In 2011, Drew Brees set the record for most passing yards in a single season with 5,476. That same season, Matt Stafford threw for the sixth-most yards in a season with 5,038.

Stafford and Brees were also first and second, respectively that season in pass attempts, and when we look at their yards per attempt totals, we see their passing yardage owes more to volume of attempts than efficiency.

Brees finished sixth in Y/A* in 2011 (100th in NFL history) and Stafford was 13th (outside history’s top 232 Y/A seasons).

(*In analysis I prefer to use net yards per attempt than simple yards per attempt, as NY/A factors in yardage lost to sacks; NY/A is more predictive and quarterbacks do have at least some control over their sack rate. However, for this discussion, I felt using standard Y/A would allow for an apples-to-apples comparison, as Brees and Stafford’s yardage totals do not include sacks. Using an efficiency stat that does include sacks would shift the scales).

So thus whenever talking about a quarterback’s season, it makes much more sense to talk in terms of efficiency stats than volume. This way, the comparison is not skewed by one player’s team playing from behind frequently–leading to more attempts–or one player being led by pass-happy coach (as was the case with Brees and Stafford in 2011).

Historical Context

Johnny Unitas finished his career with an interception rate of 4.9%. Mark Sanchez has posted an INT rate of 3.7% during his career. Fair for Jets fans to gripe about why Sanchez is ridiculed as an erratic passer while Unitas is hailed as an all-time great? These labels should clearly be reversed, right?

No, of course not. But why? “You believe in numbers right, Sal? So why kill Sanchez for interceptions and not kill Unitas?” (And no, it’s not just because Unitas is practically the patron saint of the Cacciatore family, Colts fans since the 1950s).

The real answer is really simple: The game was different during the 1950s and 60s than it is today. Unitas played before the 1978 rules changes, which liberalized the passing game, while Sanchez enjoyed these rules which make things harder on the defense.

There is a way to compare the two though. Rather than throw our collective arms up in the air and say “different eras, can’t compare Unitas and Sanchez,” we can still pit the two against each other. Rather than comparing their stats to each other right away, we need to compare their stats to the league averages when they played, and then we can compare the results.

Obviously, I’m using an extreme example to make a point–Mark Sanchez isn’t in the same area code, country, or galaxy as Unitas in any facet of the game.

Still, if we wanted to make such a comparison, we’d use Pro-Football-Reference’s index stats as a tool. These index stats compare a player’s stats to the league averages for his career (or a single season) and put them on a scale where 100 is average. Numbers higher than 100 are naturally better, and anything below 100 is below average.

Unitas posted a interception rate index (INT%+) of 108, while Sanchez’s career interception rate index is 87. Unitas was therefore 21 units better than Sanchez in terms of interceptions, despite posting a higher raw interception rate.

This makes sense when we see that the league average-interception rate was 7.3% in 1956 (Unitas’ rookie season), 5.7% in 1967 (one of his three MVP seasons), and 5.3% in 1973, Unitas’ final season.

In Sanchez’s first season (2008), the NFL-average INT% was 2.8, and has since fluctuated slightly, with a high of 3.1% in 2009 and a low of 2.6% in 2012.

Glory to Unitas; Sanchez, you’re still horrible.

Historical context matters, and this is true of individual seasons as well: in 1937, future Hall of Famer Sammy Baugh led the league in his rookie season by averaging 6.6 yards per pass; today, a 6.6 average would not come close to leading the league, as it would be below the league average of 7.2 Y/A (Baugh’s Y/A index in 1937 was 141 and the league-average Y/A was 5.6).

Back to the Original Topic…

All of this is a long-winded way of saying context matters. If we want to compare Manning’s 2013 season to his 2004 season or Tom Brady’s 2007 season or Norm Van Brocklin’s 1954 season, we need to consider volume and era.

So here’s my case for why Manning’s 2004 remains the best ever:

First, Manning’s adjusted net efficiency index (ANY/A+) in 2004 was 153, the best ever (minimum 200 attempts).

We should put a lot of stock into this, as ANY/A might be the most useful “traditional” passing stat available. ANY/A modifies basic NY/A (pass yards – sack yardage/attempts + sacks) by deducting 45 yards per interception and adding 20 yards per touchdown (for the skeptic, here’s more on how we arrive at these numbers; for even more on these adjustments, read the Hidden Game of Football).

As Chase Stuart of Football Perspective explains, ANY/A works well because it is both predictive and explanatory. While traditional quarterback rating also has use as an explanatory stat, it weighs interceptions and completions too heavily; this reduces quarterback rating’s predictive power, making ANY/A more useful.

Still, if you like traditional quarterback rating, Manning’s 2004 tops that metric too. Manning’s 121.1 quarterback rating that season translate to a 151 passer rating index, the best ever.

Manning’s touchdown percentage index in 2004 was 168, and as you may have guessed, that is the best of all time as well. Aaron Rodgers in 2011 and Brady in 2007 are tied for second (153) while Manning’s 2013 season is fourth (150) (Dan Marino in 1984 and Ken Stabler in 1976 are tied for fifth).

Brady of course holds the record for most touchdowns in a season, with 50 (one more than Manning’s 49 in 2004). Manning comes out on top because his un-adjusted TD% in 2004 (9.9%) was higher than Brady’s 8.7% in 2007 (Brady threw 578 times in 2007, while Manning had 497 attempts in 2004).

Even more advanced metrics show how good Manning’s 2004 season was. In terms of defense-adjusted value over average, Manning’s ’04 was the best season among quarterbacks with more than 200 attempts since at least 1989 (which is as far back as Football Outsiders’ DVOA stats go).

Manning’s DVOA in 2004 was 58.9%. Brady’s DVOA in 2007 was 54.1% and Manning’s DVOA in 2013 is currently is 50.6%.

According to Advanced NFL Stats, Manning’s Win Probability Added in 2004 was 7.05, the best since 1999, which as far back as the ANS stats go. This means that Manning alone was worth a staggering seven wins to the 12-4 Colts

To recap, the areas where Manning’s 2004 was the best ever include ANY/A+, QB Rating+, TD%+, DVOA, and WPA.

In the metrics where this season is not the best, it still comes out looking very good.

Looking just at NY/A (without TD or INT adjustments), Manning’s 2004 is third behind Kurt Warner’s 2000 and Dan Marino’s 1984.

Still, Manning’s best season comes out ahead of Marino’s in terms of ANY/A+, QB rating+, INT%+, TD%+, and Y/A+ (Marino edges Manning in terms of sack rate index, where his 136 sack%+ in 1984 was the best ever).

Warner’s 2000 season is weighed down by his 18 interceptions (5.2% of his attempts), which equated to an interception rate index of 70; in other words 30 units below average.

In terms of FO’s advanced metrics, Brady in 2007 edges out Manning in 2004 in terms of defense-adjusted yards above replacement. But, DYAR is a total value stat, so the fact that Brady threw almost 100 more times than Manning weighs heavily here.

Aaron Rodgers posted 0.4 expected points added per play in 2011, compared to Brady’s 0.39 play in 2007, and Manning’s 0.38 in 2004 (via ANS).

On the whole, there is an overwhelming statistical case that Manning’s 2004 season was better than his still-stellar 2013; in addition, there is substantial evidence this 2004 campaign was the best single-season campaign a quarterback has ever had.

This entry was posted in Football and tagged , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s