The Cowboys ended the season at 8-8 for the second consecutive year. As the emotional impression from the campaign wears off, a more objective view of the team starts to come into focus. As I mentioned on my recent videos, for every second half comeback that showed true grit and determination there had to be an epic fail of a first half, or first three quarters that put Dallas in the position to need a fantastic finish. In all ways, shapes and forms, this was an average Dallas Cowboys team.
Or were they?
In 2011, I began to educate myself on the various advance stats methodologies that try to glean information from the small sample size that is the 16 game NFL schedule. We know that Dallas went 8-8 on the 2012 season; in a broad sense, that's all that really matters. But do the advanced stats say that Dallas was this average team, or did they just suffer from bad breaks along the way (injury pun intended)? Do the advanced stats show that they deserved a .500 record or do they paint a picture of luck entering into play more often than it should have?
Football Outsiders creates one of my favorite metrics called variance. FO compiles DVOA data (Defense-adjusted Value Over Average) percentages which measure how well a team did against that specific opponent based on how the average team would do in the exact same down/distance/score/time scenarios. In tracking that, they measure variance, which shows how consistent said team is week in, week out; how much does their weekly performance vary.
On defense, Dallas ranked 21st in the league in variance; it was very rare for Dallas to put up a stretch of similar performances, good or bad. However with their Defensive DVOA rating sitting at 23rd on the season, it's apparent which way things leaned.
On offense, Dallas' DVOA percentage ranked 11th in the league, yet their variance ranked 20th. So again, the offense was just as up and down from week to week as the defense. You could however say that the O was markedly better performing than the D.
You can actually see the "lack of stability" in Dallas' final scores throughout the year. When you take out the four non-offensive scores the team had (strangely all against Philadelphia) in 2012, here are their offensive point totals: 24, 7, 16, 18, 29, 19, 24, 13, 17, 23, 31, 31, 20, 27, 31, 18. There are nine weeks when the swing up or down was 7 points or more.
Circling back to DVOA, Dallas' rankings (O:11th / D:23rd / ST:15th) show that the Cowboys ended up very near to where they should be, tied for the 14th best record in the league.
But what about on a game-by-game basis? You can't look at a season without considering the schedule. Based on various advanced stat measurements, how did Dallas fair against superior or inferior opponents?
The following charts look at the 16 games on Dallas' schedule based on the following advanced metrics (and I'm including PFF's cumulative grades here even though they aren't "stats"): ANY/A, DVOA, EPA, Passer Rating Differential, PFF Cumulative Grades. Unfortunately, I don't have a way to include home-field advantage, so the numbers will be a little off. Also, Only DVOA and PFF's Cumulative Grades will include Special Teams ratings.
Based on season-ending rankings, should Dallas have come out on top or lost the head-to-head encounters? Do these rankings give us accurate insight into overall performance? Take a look. When comparing a rating head-to-head, a negative result would indicate a "loss" while a positive result would mean a "victory".
ADJUSTED NET YARDS PER PASSING ATTEMPT
ANY/A is one of the better metrics that measures the overall productivity of a team's offense. (Passing Yards - Sack Yards + 20*PassTDs-45*Interceptions) / (Passing Attempts + Times Sacked)
|Week||Opponent||Real Result||Dallas ANY/A||Opp ANY/A||Adv. Stat Result|
TOTAL EXPECTED POINTS ADDED
EPA is the difference between the Expected Points (EP) at the start of a play and the EP at the end of the play. EP is the value of the current down, distance, and field position situation in terms of future expected net point advantage. Total EPA basically measures the grand total of individual play production of a team.
|Week||Opponent||Real Result||Dallas Total EPA||Opp Total EPA||Result|
PASSER RATING DIFFERENTIAL
Standard NFL Passer Rating stats. Passer Rating differential is known as a solid descriptive factor between wins and losses.
|Week||Opponent||Real Result||Dallas P/R Diff||Opp P/R Diff||Result|
TOTAL DEFENSE-ADJUSTED VALUE OVER AVERAGE
See article for definition. Total DVOA includes offense, defense and special teams "value over average".
|Week||Opponent||Real Result||Dallas DVOA||Opp DVOA||Result|
PRO FOOTBALL FOCUS' CUMULATIVE GAME GRADES
PFF grades each player on every play of every game on the same scale, with positive plays scoring as high as +2.0 and bad plays as low as -2.0. This score provides the cumulative grades of all offense, defense and special teams players on a team.
|Week||Opponent||Real Result||Dall PFF||Opp PFF||Result|
According to the results, Dallas was pretty fortunate to win some of the games that it did. There isn't a single metric that would indicate Dallas beating Cincinnati, and paying attention to the way that game went you can kind of see why. Cincinnati had some uncharacteristic occurrences of wide receiver drops and untimely penalties that kept the game close where Dallas could pull off a late comeback. The Cowboys win over the Giants to start the season also seems to be a win they grabbed against a better performing opponent.
So what, if anything, can be gleaned from all of this? Can we definitively say that Dallas overachieved? Probably not. What it does say, in my humble opinion, is that Dallas' schedule was littered with teams that were better performers than they were on a regular basis, but they still managed to achieve a .500 record. What does that mean moving forward? Not sure. One could look at it as Jason Garrett was able to get more out of his players then what should have been expected. I think it would have been troubling if we had conducted this exercise and saw that Dallas had 10-6 or 11-5 records in some of the extrapolations.