FanPost

The Cowboys Offense is Not Inconsistent

As part of the comment thread from a post analyzing the Cowboys offense over the last 3 years, I saw a theme emerge as a common critique. Specifically, that the Cowboys scored a lot of points, but the offense was inconsistent, and often set the defense up for failure by failing to convert on 3d downs, etc. As a companion theme, some commenters brought up the low scoring games in which we lost (4 games in which we scored 22 points or less against KC, SD, NO, and PHI), implying that our offense was volatile, and often underachieved in a specific game. I thought these were interesting critiques and I didn’t know how to answer. After a little research, I have to conclude that the Cowboys offense, for all its imperfections, was among the most consistent scoring offenses in the league this year. Here’s the proof.

FACTS AND FIGURES

I’m going to have to start with some numbers. I’ll mostly be using averages, a little data sorting, and some standard deviation. The level of commentary on BTB is pretty impressive, and I assume most folks know all about standard deviation, but to avoid leaving anyone behind, here’s a quick definition:

In statistics and probability theory, the standard deviation shows how much variation or dispersion from the average exists.[1] A low standard deviation indicates that the data points tend to be very close to the mean (also called expected value); a high standard deviation indicates that the data points are spread out over a large range of values.

An example helps put this in somewhat simpler terms. Suppose the Cowboys score an average of 25 PPG. A standard deviation of 10 means that 68% of the time the team scores between 15 and 35 points (moving from that average of 27 both up and down), and 95% of the time they score between 5 and 45 points. As the definition explains, the lower the standard deviation, the tighter the PPG range, so that if the standard deviation went down to 5, that means the Cowboys score closer to 25 points more frequently. I think that’s enough math, and look forward to the smarter folks correcting my definitions or improving my examples. So to clarify, lower standard deviation means more consistent and less volatile.

Below is a table with the Cowboys’ PPG, and as we can see, our offense scored an average of 27.4 points a game, and had a standard deviation of 8.1.

Week

1

2

3

4

5

6

7

8

9

10

Dallas Points

36

16

31

21

48

31

17

30

27

17

Week

11

12

13

14

15

16

17

AVG

STDEV

Dallas Points

24

31

28

36

24

22

27.4

8.1

It’s a very pretty table, but let’s establish some context by looking at the rest of the league. I’ve summarized the data into just PPG and Standard Deviation, and hope you will just accept that all the underlying numbers are correct. I’ve highlighted all the playoff teams in green, and broken the league into groups of 8 based on PPG.

Top Quartile

2d Quartile

PPG

STDEV

PPG

STDEV

DEN

37.8

9.3

GB

26.1

9.4

CHI

27.8

9.2

NO

25.9

11.0

NE

27.8

10.9

SF

25.3

11.1

PHI

27.6

12.7

SD

24.8

7.9

CIN

27.5

11.6

DET

24.6

8.6

DAL

27.4

8.1

IND

24.5

8.8

KC

26.9

11.3

MIN

24.4

9.9

SEA

26.1

9.7

ARI

23.7

7.9

Quartile Average

10.4

Quartile Average

9.3

3rd Quartile

Bottom Quartile

PPG

STDEV

PPG

STDEV

PIT

23.7

8.0

BAL

20.0

5.7

CAR

22.9

9.8

MIA

19.8

7.6

TEN

22.6

7.1

CLE

19.3

9.0

ATL

22.1

7.5

NYG

18.4

8.7

STL

21.8

11.4

TB

18.0

8.8

BUF

21.2

7.3

NYJ

17.5

9.1

WAS

20.9

9.4

HOU

17.3

9.3

OAK

20.1

6.1

JAX

15.4

8.6

Quartile Average

8.3

Quartile Average

8.4

League Averages

PPG

STDEV

23.4

9.1

ANALYSIS AND CONCLUSIONS

So what does this table tell us? Here were my key takeaways.

  • While exceeding league average scoring by 4 PPG, Dallas had a lower Standard Deviation than average at 8.1. This was the 10th lowest in the league.

  • Dallas had the lowest Standard Deviation among the top scoring teams, indicating that they scored more consistently than the other teams. If we look at the top 10 scoring offenses, and then remove Dallas, the average Standard Deviation was 10.6, much higher than Dallas’ 8.1.

  • Dallas had a lower Standard Deviation than all the playoff teams, who averaged a 10.3.

  • Interestingly, Standard Deviations went down as PPG went down. I suspect that is because those offense have less firepower, and therefore less upside in scoring. For example, of the top 10 scoring offenses, every team except Green Bay scored 45 points at least once, and Green Bay had a season high of 44. Not a single team in the bottom half of the league scored 45 points even once.

Here are a few more interesting facts about Dallas’ scoring consistency I found lurking in the numbers:

  • Only 8 teams in the league avoided a game in which they scored less than 10 points. Dallas was one of the 8. Five of the top 10 scoring offenses had at least 1 game under 10 points.

  • Dallas had a top 10 scoring week 9 times over the season. By comparison, the Saints only had 7 (yes, one was against us).

  • Dallas finished in the bottom 10 in scoring only 3 times. By comparison, Philly was in the bottom 10 four times.

  • Of the other 9 scoring offenses in the top 10, 8 had a game in which they scored lower than Dallas’ lowest, with Denver as the only exception.

  • Every top offense seemed to have at least one high score outlier, with a roughly equivalent impact on their scoring average if eliminated. So while our Denver game was an outlier in scoring, every team seemed to have one.

What do I conclude? Dallas had some low scoring games which they lost. Absolutely true. Every top offense (Denver again was the exception), and every playoff team had the same problem on occasion. Dallas had the 5th highest PPG, was the 10th most consistent scoring team (judged by standard deviation), and didn’t have as many outliers as most other top offenses. What exactly are our expectations for the offense (other than to outscore the other team every week – and I’m on board with that one)? Are they realistic?

Another user-created commentary provided by a BTB reader.