Cowboys Poll: How many games must Jason Garrett win to keep his job?

Things have not been going the Cowboys’ way this year, and that’s not just a knee-jerk reaction to the last games.

No, things have been out of kilter for one reason or another since Week 1. In fact, things have been out of kilter since the Atlanta game in Week 10 last year. How much out of kilter? Just look at these stats over the last 13 games:

• Points scored: 211 (NFL rank: 31st)
• Pass attempts: 376 (30th)
• Passing Yards per Attempt: 5.91 (29th)
• Offensive Passer Rating: 73.9 (27th)

And if things remain the same as they’ve been for those 13 games, the team will be lucky to finish this year with an 8-8 record.

The Pythagorean Formula (for more details, go here) was developed to measure overall team strength on the hypothesis that a team’s true strength could be measured more accurately by looking at points scored and points allowed, rather than by looking at wins and losses.

The interesting thing about the formula, at least for our purposes today, is you can calculate a projected season win total after every single game. Which is exactly what I did with the cumulative points differentials after each game:

Obviously, the results here get more robust with every additional data point. After some wild swings in the first two game, the win projection has settled in at around six to seven wins since.

And this far into the season, the data is robust enough to suggest that the Cowboys are not suddenly going to go on a six-game win streak.

Another method to calculate projected wins is via passer rating differential (PRD). We know that PRD is one of the stats most closely linked to winning in the NFL. And PRD can be plugged into a very simple formula to predict win totals in the NFL. The PRD formula (Projected Wins = PRD*0.16+8) has shown a fairly close correlation with the Cowboys’ actual wins over the last decade or so, but has been oddly off the last few years:

Overall, the PRD formula generates a mean average error of 1.9 games over the last four seasons, which isn’t great, and is off significantly more in the last two years. This suggests that unlike previous years, the performance in the passing game for the Cowboys is not as big a determinant in their W/L loss as in the previous year.

Coincidentally, with the arrival of Ezekiel Elliott in Dallas and the departure of Tony Romo in 2016, the Cowboys became a much more run-oriented team, so I figured that I had to find a way to account for the running game in the PRD formula.

So I played around withe the data a bit came up with a modified PRD Formula that includes a team’s rushing yards differential (RYD) as follows: Projected Wins = (PRD + RYD/50) * 0.16 + 8.

In statistics, the relationship between two variables is called a correlation, and the strength of that correlation is measured by the “correlation coefficient”. This coefficient (r²) is expressed as a number between 1 and -1. The closer the r² number is to 1 or -1, the stronger the relationship between the two variables. The closer it is to zero, the weaker the relationship.

I ran the original PRD formula against the 2017 regular season results for all 32 teams, and r² between Passer Rating Differential (PRD) and wins in 2017 was 0.62. Not bad, but not great either. Rushing yards differential for 2017 was a decidedly unimpressive 0.17, which is one reason why so many stat heads poo-poo the running game and suggest teams need to pass more. But after a little tinkering, something interesting happened. The modified PRD, which now includes rushing yards differential, came back with an r² of 0.77 which is one of the strongest correlations you’ll find outside of points differential. Here’s a graph that illustrates the correlation between the modified PRD formula and wins for all 32 NFL teams in 2017:

In terms of Mean Average Error, the original PRD Formula was off by 1.6 games on average in 2017, the modified PRD reduced that 1.3 games.

And once applied to the last few Cowboys season, the modified PRD also delivered much better results:

With the modified PRD, the mean average error for the Cowboys improved from 1.9 to a very impressive 0.9 wins.

So what does the modified PRD project as the final record for the 2-3 Cowboys? With a PRD after five games of -22.5 and a RYD of +200, the formula projects 5.5 wins for the Cowboys this year.

So the Pythagorean Formula has the 2018 Cowboys at 6.6 wins, the modified PRD sees them at 5.5 wins (both projections already factor in the Cowboys’ two wins). The modified PRD tells basically the same story as the Pythagorean formula, even if the projected win totals are a little lower.

Want more projection? Fivethirtyeight.com has the Cowboys at 7.5 wins. Football Outsiders project the Cowboys for 7.4 wins.

Looks like the 2018 Cowboys are likely headed to a six- or seven-win season, with eight wins as a clear ceiling.

Which brings us to the titular question of this post. How many games must Jason Garrett win to keep his job?

Many fans would like to fire Garrett regardless of his record this year, but that’s not the way the Cowboys operate, as Tom Ryle pointed out the end of last season:

Of course, there is also the fact that the most invulnerable general manager in the NFL is Jerry Jones, the only owner to serve as his own GM (along with a sizable amount of help from his son Stephen, particularly in cap management and personnel decisions). That eliminates one major way to shake an organization up that other teams can use. And Jones has clearly shown he is fond of continuity, particularly with Jason Garrett, who is far and away the best fit as far as working with his boss since Jones bought the Cowboys.

Jason Garrett is in his ninth season as a head coach with the Cowboys, and has had plenty of time to mold the team to his wishes. Yet all he has to show for it is one playoff win, a 65-59 career record, and another season that that looks to be his fifth non-winning season.

Of course, none of this would have been possible without the ineptness of the father-son duo playing GM. Yet those two will be the final arbiters of Garrett’s fate, and as Tom points out, they might wimp out at firing Garret just as Garrett wimped out on the 4th-and-1 in Houston.

No such wimping in Vegas, where Garrett is now listed as the mostly likely head coach to be fired first.

And Jerry Jones wasn’t too happy about his coach on Sunday either. RJ Ochoa pointed out that Jones publicly contradicting Garrett’s decision to punt doesn’t bode well for the coach.

So how many wins will it take for Jerry Jones to hang on to Garrett for yet another season?

Give your answer in the attached poll or in the comments section below.