Who Wins Games For The Cowboys: Offense or Defense?

Heading into last season, there was relatively little concern about the Cowboys' offense, but the naysayers, scaremongers and doom merchants were having the time of their lives predicting catastrophe for the Cowboys defense.

• In a series of Tweets in August 2014, Ross Tucker wrote that the "Cowboys have worst defensive depth chart (1st 4 games) I've seen since I entered NFL in 2001. I'm serious. I think we have to consider possibility that Cowboys will have worst defense in NFL history."
• Also in August, Nathan Jahnke of Pro Football Focus ranked the Cowboys as the worst defense in the league heading into 2014 in an ESPN In$ider article.
• The Washington Post triumphantly added that the defense was "comically bare" and that "Dallas very well could be looking at the worst defense in the history of football in 2014."
• At the end of August 2014, Gregg Rosenthal of NFL.com predicted that the Cowboys would barely scrape out six wins with a defense that "could go from worst to collegiate."
• In September 2014, Rick Gosselin of the Dallas Morning News wrote that "on paper, the 2014 Dallas defense is the worst collection of talent by this franchise since 1960."

The offense certainly lived up to expectations, ranking fifth overall in scoring with 467 points, and the defense made many pre-season prognosticators look foolish by finishing the season ranked as the 15th-best scoring defense with 352 points allowed. Still, when we talk about the 2014 Cowboys defense, somebody will inevitably find a way to credit the Cowboys offense as the real reason behind the defensive improvement, just as somebody else will vehemently insist that Rod Marinelli is a miracle worker.

One way (among many) to get a better understanding of the contributions of each unit is to look at 'wins over average' for each unit.

'Wins over average' is a metric developed by Doug Drinen at Pro-Football-Reference.com (PFR) in a post on adjusting QB records, back in the days before PFR closed down their blog. I've taken his approach and modified it slightly to look at total offense and total defense.

Think back to last year's 20-17 OT win against the Texans in week 5. Did the offense or the defense win that game for the Cowboys? The two teams entered overtime tied at 17 points each, and Houston got the ball first, but their drive stalled at the Dallas 48 when a pass on 3rd-and-2 fell incomplete. That's a stop for the defense, right? After Houston punted, the offense then marched up the field and Dan Bailey eventually hit a 49-yard field goal to win the game. So that's a win for the offense, right?

You could go through many recent Cowboys games and find arguments for or against each unit. But instead of subjectively evaluating each game, I'll approach this with a more objective, stat-based approach, and I'll start by looking at the last four years before drilling down on 2014. This is how it works:

Wins over Average: As our measure of difficulty for the offense, we'll use points allowed, because it's a lot harder for an offense to win a game if their defense gives up 30 plus points than if the defense gives up 10. For the defense, we'll look at points scored by their offense, as it's a lot harder to hold opposing offenses in check when your own offense is only scoring 10 points rather than 30 points.

In the table below I've crunched the regular season number for all NFL teams for the last four years, excluded tied games, formed five clusters by points allowed/points scored, and looked at the winning percentages for each.

Obviously, one team's points scored is the other team's points allowed, so the table below contains the same numbers for offense and defense, just reversed:

How to read the table: An NFL offense almost always (258-7) scores more points than the other team if the defense allows ten points or less. The NFL average winning percentage is .974 in those games. At the other end of the spectrum, when your offense scores 32 points or more, your defense is almost always (326-36) able to hold the opposing team to fewer points for a winning percentage of .901

Cowboys offense and wins over average: The Cowboys have a record of 7-0 over the last four years when the defense allowed ten points or less. The average NFL team would be expected to win 6.8 of those games (7 games x .974), so the Cowboys have 0.2 wins over average in this bracket. In the 11-17 points bracket, the Cowboys have a 10-1 record, where an average NFL team would be expected to win 8.6 games (11 games x .779). The Cowboys offense therefore gets the credit for 1.4 wins over average. Across all brackets, the wins over average for the offense total +6.2 games over the last four seasons.

Cowboys defense: The Cowboys have a record of 0-4 over the last four years when the offense scored ten points or less. The average NFL team would be expected to win 0.1 games (4 games x .026), so the Cowboys scored -0.1 wins over average in this bracket. Across all brackets, the wins over average total -1.4 games for the defense as per the table below.

So statistically, the Cowboys offense generated slightly more wins over average than the defense, but not by a wide margin. And given that the Cowboys went 8-8 for three seasons before climbing to 12-4 last year, that shouldn't come as much of a surprise. But just how much is 2014 skewing the overall results? Here's how each individual year of the period looked in terms of wins over average:

What you can see here is a defense that was slipping for three years before last year's turnaround. The Cowboys were losing slightly more games than could have been expected given the points they were scoring on offense. And while the defense was slipping, the offense was okay, but didn't really win that much more games than one would have expected given the points allowed by the defense. All of that changed in 2014 of course.

In 2014, the Cowboys offense recorded 3.9 wins over average, in part because they won four of six games where the defense allowed between 25 and 31 points (league average would have been just 1.5 wins).

Week 3: 34-31 over Rams
Week 12: 31-28 over Giants
Week 14: 41-28 over Bears
Week 15: 38-27 over Eagles

And those 3.9 wins over average were the best value in the league last year.

Here is the full list of projected wins, with some other data points added in. For your convenience, the table is sortable (just click on the blue column headers).

Wins over average is not a perfect stat. For one thing, it ignores the impact of special teams completely. But it does give you an interesting way to look at the contribution of offense and defense to a team's record. If we look at the Cowboys in 2014 we see that they had a very strong offense, and surprisingly also had a good defense (at times) - their 1.6 wins over average rank the defense eighth overall, not bad for the "worst collection of talent" in franchise history. Of course, none of that is news for Cowboys fans.

But a look at some of the other teams here may provide some new perspective. Despite Matthew Stafford, Reggie Bush and Megatron, the Lions offense was merely average in terms of WOA, while their defense tops the ranking here with 4.1 WOA. The Bills and Chiefs also had a large disparity between a strong defense and a sub par offense, but unlike the Lions they improved their defense and have reasonable hope for some improvement on offense.

In the NFC East, the Redskins rank near the bottom of the table with big issues on both sides of the ball. The Giants were slightly below average on offense, but were uncharacteristically weak on defense. The Eagles had the fourth-best offense in terms of WOA (no doubt aided by an exceptional special teams effort), but their defense only ranked 23rd overall, exactly the rank they had in points allowed. So the next time a Kelly acolyte wants to tell you a story about how great the Eagles defense was, tell him to lay off the smoothies.

For the Cowboys, the data here is a double-edged sword. Sure, it's nice to be ranked on top of a table like the one above, and it's also nice to see the defense ranked as high as it is here. But with those rankings come risks: there's a good chance the Cowboys will regress towards the mean on both measures next season. That doesn't mean the 2014 season was a fluke, but it does mean that in games where the Cowboys defense allows 25 or more points in 2015, the team is unlikely to repeat its 6-3 overall record in such games from 2014. Instead, a 2-7 record would be much more likely.

Of course, the best way to avoid that regression to the mean is to simply hold every opponent to 24 points or less - or beat every opponent 42-7.