Inside the Numbers: The 2009 Cowboys Passing Game (Part I)

The 2009 Cowboys offense broke records left and right. The franchise record 6,390 yards on offense eclipsed the 1978 Cowboys (5,916), and was good enough for the number two spot in the NFL, a paltry 71 yards behind the New Orleans Saints.

In the passing game, Miles Austin set a franchise record with 250 receiving yards in a single game against Kansas City. The Cowboys receivers racked up a league leading 2,154 yards after catch. Tony Romo threw for a career high and franchise record 4,483 yards, and entered the "official" NFL leaderboards by surpassing 1,500 career passing attempts and thus qualifying for the NFL record books. Here's an (updated) excerpt from ColdHardFootballFacts.com:

So Romo enters the record books today in some very, very lofty company. His career passer rating of 95.6 is second all time. Here's the top five in career passer rating:

Steve Young: 96.8, Tony Romo: 95.6, Peyton Manning: 95.2, Kurt Warner: 93.7, Tom Brady: 93.3

Now look at the top five in career passing yards per attempt:

- Otto Graham: 8.63, Sid Luckman: 8.42, Norm Van Brocklin: 8.16, Tony Romo: 8.10, Steve Young: 7.98

Tony Romo is not only in the top five all time, he's the most prolific passer since Norm Van Brocklin retired at the end of the 1960 season. He's also one of just two players who appears on both lists, with Steve Young. This indicates that he's produced the highly efficient rating indicative of the modern game, with the very high average per attempt more common in earlier years of NFL football, before the "ball-control" style passing game became all the rage."

Was the Cowboys’ passing game really as good as as these stats suggest, and is all that glitters really gold? Follow me as we turn the incorruptible eye of Expected Points Value (EPV) on our passing game to uncover the truth behind the stats.

Dallas Cowboys passing game EPV

Tony Romo was the only QB to throw passes for the Cowboys in the 2009 season. So the numbers we'll look at as we examine the 2009 Cowboys passing game are Romo's numbers. But I prefer to talk of the 'Cowboys passing game' instead of 'Romo's passing game' because the EPV that I use is the value of each play, not the value of each player (Go back to my previous post if you would like a refresher on EPV).

As the old saying goes, "Success has many fathers, while failure is an orphan". Football is a team sport, and the credit or blame for each pass is shared by many players and coaches.

The base numbers of the passing game we'll be looking at today include touchdowns, interceptions, completions and incompletions (two spikes are excluded). Just as a reminder, interceptions weigh particularly heavy on EPV, because EPV measures the expected points value before the play and after the play, based on down, distance and field position - in the case of an interception, field position would be the opponent's field position. In terms of EPV, a pick six is simply deadly. Luckily, the Cowboys only had one last season, and here is how the pick six from the first Giants game breaks down in EPV:

3-10-DAL 24 (2:56) (Shotgun) T.Romo pass short right intended for P.Crayton INTERCEPTED by B.Johnson at DAL 34. B.Johnson for 34 yards, TOUCHDOWN.

The starting EPV on third-and-ten on Dallas' 24 yard line is -0.38. Not a good position to start with. The interception by Bruce Johnson resulted in -6.3 EPV, as 6.3 is the value we use for a TD. The value of the play was therefore -6.3 minus -0.38 equals -5.92.

In total, the Cowboys attempted 548 passes in 2009 which resulted in an EPV of 171.37, or 0.31 per pass attempt.

Is this number good or bad? To be honest, I don't really know. Until I do the same EPV excercise for all other teams and QBs the only thing I have to compare this with is the EPV of the Cowboys opponents in 2009:  Their 572 pass attempts (excluding one spike) resulted in an EPV of 80.62, or 0.14 per play. 

This result suggests that the Cowboys' passing game generated twice the value of that of our opponents. Here's how those numbers compare with the more traditional passer rating in 2009:

Because the difference in EPV is much bigger than in passer rating, this may initially look like something is wrong with the EPV numbers. But this is actually consistent with how EPV is calculated. EPV values a long pass much more than a short pass, in fact, if you dink and dunk your way up the field with little four and five yards passes, your EPV will likely be negative (Dallas had a passing YPA of 8.2 and allowed only 6.8 YPA).

Versus the passer rating, EPV places a high value on converting a play for a first down (Dallas made 203 passing first downs while allowing only 186). EPV also gives TDs and turnovers a much higher weight than the passer rating ( Dallas scored 29 passing TDs while allowing only 19)

Passing by downs - Trouble on third downs

In terms of EPV, passes on third down generated less value than on any other down situation. We've already analyzed our third down play in a lot of detail ("Cowboys Third Down Offense") and found that

A) No other quarterback in the NFL has had to pass on more third and long situations than Tony Romo (76 pass attempts on third and 8+ yards).

B) The Cowboys last season did not have an effective play/receiver to convert 3rd and medium passing situations.

The official stats show that last season the Cowboys managed to convert 40.5% of their third downs (82/202), and ranked a middling 14th in the NFL. That 40.5% conversion rate is the lowest for a Cowboys team since 2005. The one piece of good news in this stat is that the 202 third downs were actually the sixth fewest in the NFL (the NFL average was 215) - looks like the Cowboys offense was pretty good at avoiding third down situations all together.

Passing by distance required - Give the man some room

The Cowboys passing game generated the most value when working against 8+ yards situations, and was distinctly less effective in shorter yardage situations.

Passing by direction - Money in the middle

Let's take a look at the EPV values by distance and direction (note: long passes are those that travel 15 yards or more through the air):

Two things immediately jump out as I look at this table:

First, the Cowboys liked throwing the ball to the right: they threw the ball to the right side of the field 24% more than they threw the ball to the left side, and a full 61% more than up the middle of the field. Particularly the lack of throws up the middle is a little disconcerting. If a guy sitting Germany and fooling around with an Excel sheet can see this trend, I would assume that opposing defenses knew all about it.

Second, and much more worrying, is that the EPV of short passes to the right is alarmingly low. Think of it this way: short passes to the left generated more than five times the value of a short pass to the right, passes up the middle generated eleven times more value.

I initially thought that this might be because of interceptions and turnovers that mess up the averages, so I looked it up. And yes, of the seven turnovers on short passes, four were to the right, two up the middle and only one to the left. But when I took the turnovers out of the calculation, the EPV didn't change significantly. The turnover adjusted EPV for short passes: right: 0.15, left: 0.30, middle: 0.57.

Now, if you're thinking "Hmmm, didn't Roy Williams line up on the right a lot?" let me disabuse you of that thought right away. We'll look at individual receivers in a separate post, but for now, trust me that Roy Williams had nothing to do with the short right woes, he was one of the few receivers whose numbers were actually better (albeit on a fairly low level) on short passes to the right.

Short passes to the right are something the Cowboys need to look at in training camp, and they also need to figure out a way to get more passes up the middle.

Passes by formation - Shotgun Blues

When I ran the numbers by how Romo lined up before a play, under center or in a shotgun formation, I was expecting what you'd assume for most NFL passing games: the shotgun would be more valuable. Turns out I was wrong:

In 2007, the Patriots became the first NFL team to run the shotgun formation on more than 50 percent of their plays. In theory, the shotgun gives defenses an easy key (especially for d-line and linebackers) to focus on the pass rush and not worry about the run as much. But the Patriots did ride it to a 16-0 record. So much for the theory.

So let's take a look at Garrett's playbook this year and see how it compares to last year. In 2008, Garrett had the shotgun on the field for 43.9% of the Cowboys offensive plays (excl. knees & spikes). It remained basically unchanged at 43.5% this year.

Looking at the numbers in the table, there is no reason why the Cowboys shouldn't pass more from under center. The EPV ic clearly higher, and the more traditional YPA is also higher from under center. Also, passing more from under center would add an extra element of surprise to our offense more often.

Obviously the runs from the shotgun are a small base, but with the type of running attack the Cowboys have, mixing in a little more runs from the shotgun formation might not be the worst strategy either.

Next time, we look at wide receivers. Does Miles Austin really stand head and shoulders above the rest, how big a disappointment was Roy Williams, whatever happened to Jason Witten and is it time to free the Ogletree?

[Special tip of the hat to Brian Burke at advancednflstats.com for providing the EPV data. Follow the link if you want to dig deeper into EPV. TJ Johnson at the MileHighReport here on SB Nation also has a great series running on EPV which I highly recommend.]