How Important Are Quarterbacks?
How many clichés about how to win in football can you think of in 15 seconds or less? "Defense wins championships", "It's a team effort", "You've got to win the turnover battle", or even "You have to make halftime adjustments".
Two clichés that may not be totally untrue, though, is that the NFL today "is a pass-happy league," and "quarterback is the most important position on the field". These are commonly held, even clichéd notions, but just how factual are they?
We can certainly see how the league has swung towards favoring the passing game just by looking at the record books. Four of the top-five marks for pass attempts in a season were set in the last four years, all of the top-five seasons for passing yardage were set in the last three years, and the top-three passing touchdown seasons have all happened in the last 10 years.
What about the other cliché? Are quarterbacks the most important position on the field? They are the only position that touches the ball on every play, so it seems like this answer is fairly obvious, but to what extent?
This is where we can turn to our good old friend, Net Expected Points, which examines the value of a player or unit's contribution to their team's scoring output. If this is a measurement of success for players, and winning is a measure of success for teams, perhaps we can find a connection between quarterback success and winning games in the NFL.
The goal of this study was to see if quarterback play (or perhaps another team factor) would provide strong indicators of team success as a result of the quarterback. Make no mistake, this isn't supporting the fallacy of quarterback wins. Rather, this is an examination of the relationship between a quarterback to his team's success, or making the playoffs.
Some questions I had going into this research were: Is there a correlation between the quarterback's play (Quarterback Passing & Rushing Net Expected Points) and team wins? Has this increased over the last decade? Can we predict future playoff teams? And what should teams hoping to get to the playoffs look for in their roster construction?
You may be surprised by the results.
While we're talking about clichés, I can't go without mentioning Jim Mora when talking about the playoffs. Indianapolis' head coach was shocked that anyone would mention the word in the wake of a devastating 1998 loss, and it's similarly as shocking to see some of the teams and names that have made the playoffs since 2004, where I began this study.
Some of the lowlights: The 2005 Bears (Kyle Orton, -99.94 Passing NEP), the 2007 Giants (Eli Manning, -46.01), the 2009 Jets (Mark Sanchez, -68.36), the 2011 Broncos (Kyle Orton/ Tim Tebow, combined -47.98). What's interesting to note is that all of those teams, excepting the 2011 Broncos, had above average to elite defenses that likely carried them into the playoffs, rather than the strength at the quarterback position. But no, let's not go declaring that "defense wins championships" just yet - we're interested in big sample sizes and trends, not incidental flukes.
From a cursory glance at the data, we can see that one noticeable trend is that the NFL hasn't gotten kinder or easier when it comes to earning your way into the playoffs. There were 12 teams from 2004-2008 with single-digit wins who found themselves in the postseason, while only eight under that win total from 2009-2013 played into January. Though a small difference, the average wins for a playoff team in the first half of the decade was 11.067, and 11.117 in the latter half. In simple terms: winning has become more important if you want a chance at a ring.
In addition, only nine playoff teams have had quarterbacks (or quarterback committees) with a negative Total NEP. And only three of those have occurred since 2009. Therefore, we can draw a very rough conclusion that, as teams win more and quarterbacks perform at higher levels, those two metrics might be connected. But how closely are they tied?
"Better Than Their Record Indicates"
The beauty of using NEP to analyze the success of a team is that we can go behind the box score, beyond the standings, and really dig into what made a team effective or ineffective in a given year. We can see which teams "deserved" to be in the playoffs, and which had a few lucky bounces that may have gotten them in (I'm looking at you, 2013 Packers).
The real question here, however, is not whether a team deserves to be in the postseason, but what got them there.
I charted each playoff team over the past decade along with their Adjusted NEP (team offense adjusted for strength of schedule), Adjusted Defense NEP (team defense adjusted for strength of schedule), and their quarterback's (or quarterbacks') Passing and Rushing NEP. I then calculated the correlation between each of these metrics and the team's regular season wins, to see if there was any indication of a formula for success in the NFL.
For reference, a correlation of 1 or -1 means a perfect one-to-one ratio of cause and effect. Anything closer to zero means they are less related. A correlation above 0.7 or -0.7 means a strong, though not ironclad, correlation, and anything above 0.50 or -0.50 is a significant, but imperfect correlation. The metric with the strongest correlation to team wins? Quarterback Total NEP. But let's see if we can't debunk some of these other myths first.
Let's start with cliché number one, the most rampant of these myths about winning: "defense wins championships".
One of the clearest conclusions that can be gleaned from this research is that there has been practically no correlation between defense and winning over the past decade. Since we measure Adjusted Defense NEP in negatives (that is, a better unit has a lower value, as they limit points scored against them), the correlation score for this relationship is in the negatives, but it is pretty insignificant: -0.19 over the past decade. Over the last five years, this correlation drops even more, to -0.07. Some prime examples of these playoff teams winning despite atrocious defenses include the 2006 Colts (12 wins, 91.49 Adjusted Defense NEP), the 2011 Patriots (13 wins, 101.25 Adj. Defensive NEP) and the 2012 Colts (11 wins, 102.47 Adj. Defensive NEP).
On the other side, there were plenty of teams that squeaked into the playoffs with single-digit wins but had stellar defenses, like the 2008 Eagles (9 wins, -133.36 Adj. Defensive NEP), the 2009 Jets (9 wins, -149.64 Adj. Defensive NEP), and the 2009 Ravens (9 wins, -57.35 Adj. Defensive NEP).
Even when we look at singular seasons, there are only two years where defensive prowess had a significant correlation to teams winning (2004, -0.67; 2013, -0.62). The most telling thing is that there is also no trend to the yearly correlation between the two statistics. In just the past five years alone, the correlation scores were: 0.56 (positive (!), indicating that a bad defense was more likely to correlate to winning), -0.32, 0.37, -0.53, and -0.62. Nearly completely random.
Another common theme we've started to hear is that running quarterbacks are the future. Dual-threat signal-callers seem to be the revolution in the NFL, especially when you consider that two of them matched up against one another in this year's NFC Championship (Russell Wilson and Colin Kaepernick).
We can look at all the mobile quarterbacks in the league, and we can view their wins as rookies and say that they're the "next big thing", but do the numbers play out favorably for them? Again, a quick glance at the data shows a clear negatory. There is even less of a correlation between team wins and quarterback Rushing NEP. Over the past decade, it has been a -0.05; that's about as insignificant as you can get. If we want to consider just the most recent three years, as some people say the rushing quarterbacks and the read-option have become more successful recently with Kaepernick, Wilson, Robert Griffin III, and Cam Newton, that correlation actually drops to -0.03. So, it's not the read-option rushers either.
Our final category before we check out quarterbacks in full is Adjusted NEP. Does a team's total offense show the correlation with wins we're looking for? The answer, to an extent, is yes. Over the past decade, this metric actually outstrips quarterback Total NEP by just a bit, at a 0.50 correlation. However, it's clear that in the past few years the game has become much more quarterback-centric, that pass attempts have skyrocketed, and the emphasis on rushing and defense has diminished. So what is the correlation over the past five years? 0.56. Some teams that made it in on the coattails of historic offenses? Those 2006 Colts (177.55 Adjusted Net Expected Points) and 2011 Patriots (236.18 Adj. NEP), for two, not even considering the gunslinging teams that were the 2011 Packers (229.70 Adj. NEP) and Saints (220.90 Adj. NEP).
This shows that offense is indeed the best defense (I'm guilty of using clichés too) in today's NFL, yet the reason I'm not highlighting this metric as the best predictor is its variation in providing strong correlations (again, it's got to be greater than 0.7) over the past decade. While it may have given a few more significant correlations (greater than 0.5), only two seasons in the last 10 years have had strong correlations between Adjusted NEP and team wins, and only one of those has occurred in the last five years. This is a good indicator, but not the best that we have.
"He Just Puts the Whole Team on His Shoulders!"
We finally arrive at quarterback NEP. The "field general", the "signal-caller", the "audibilizer": quarterbacks touch the ball on every play, so we know they're important. But how important are they?
Over the past decade, quarterback Total NEP (which factors in both rushing and passing for quarterbacks) has had a 0.44 correlation with team wins. "Aha!" you'll say. "That's a negligible correlation, so quarterbacks are not the best indicator!"
However, we all know that over the past few years teams have heavily shifted away from the strong running game that dominated the NFL's early years, and even the early half of this decade. If we look at just the last five years, we'll see that quarterback Total NEP has a 0.62 correlation with team wins, the strongest such indicator we've seen over a period longer than a year. Indeed, since 2009, this correlation has trended: 0.83 (!), 0.47, 0.71, 0.74, and 0.30. Three of the last five seasons having strong correlations is a pretty clear indicator that, despite a few outlier seasons, there is a clear connection between the success of a quarterback and the success of a team in making the playoffs. In fact, if we exclude 2013, a massively down year, the correlation of quarterback Total NEP with team wins from 2009-2012 is a stunning 0.68.
In this new offensive-focused NFL, there is a clear relationship between successful quarterback play - mostly passing - and winning. You can say that it's picking and choosing, but there are only two seasons in the past decade where defense dominated (2004, 2013) and there was an insignificant correlation between team wins and offense. The only other outlier was 2008, where there was a negligible correlation between all metrics and team wins. Every other year, especially in the last few, general offense and specific quarterback play clearly correlate best with team wins.
"Knocking on the Door"
So, we've established that the best way to get to the playoffs is to have an above-average quarterback, but how above-average? Remember, no team made the playoffs in the past two years with a negative quarterback Total NEP (even the 2013 Packers, whose starters included Seneca Wallace, Scott Tolzien, and Matt Flynn - yuck).
These are very rough calculations, but if we average out the last five years' wins for a playoff team, that gives us around 11 wins as the magic number to give a team a good chance of making the postseason. Over the last five seasons, the average playoff quarterback Total NEP was 102.33; this may be a good guideline number for a team wanting to make the playoffs. To cross-check this, I averaged both the last five seasons' team wins per quarterback Total NEP to find the rate, then divided the average team wins by this rate. Sure enough, this "threshold Total NEP" (I'll call it that) comes out to 90.78 over the last five years.
This threshold for quarterbacks indicates that, to have the best chance at making the playoffs, a team will want to secure a quarterback that can net them at least 90 Total NEP in a season. Indeed, 22 out of the last 36 postseason teams have had a quarterback with around 90.00 Total NEP.
Unfortunately, the only free agent quarterback who fits this mold is impending Bears free agent Josh McCown, who is unlikely to repeat the same type of performance he did in 2013.
As for success in the playoffs, well, there's a lot of variance. Last year, the Super Bowl Champion Ravens had a mediocre quarterback in Joe Flacco (25.1 Total NEP) and a below-average defense (22.62 Adjusted Defense NEP). The year before that, the Giants won with an above-average quarterback and a below-average defense when they faced the Patriots' elite quarterback and terrible defense.
This year, it's interesting that we have two teams that went 13-3 in the regular season, both top seeds, and the record-setting offense of the Denver Broncos against the league's top defense of the Seattle Seahawks. 2013 was a year of defense matching with winning, but if I were a betting man, I'd put my money on the best offense in history. The NFL, as we've seen, is a quarterback's world; the rest of us are just playing in it.