See that title? I’m a big believer in the Oxford comma. I always have been, and I’ll never stray away from it. I’ve been told 1,000 times to leave out the last comma in a list, and I’ll never do it.
- I’m writing this article for my fans, Jason Garrett and Bill Callahan.
- I’m writing this article for my fans, Jason Garrett, and Bill Callahan.
Which makes more sense to you?
Okay, so here are a few articles I’ve published lately. Click the links to read more.
At NBC, a projection for DeMarco Murray in 2013:
The key to projecting Murray in 2013 is determining his snap count. That’s also the hard part. Most will argue that you can’t project Murray to play all 16 games because he’s injury prone, but that’s really difficult to argue. Yes, he’s suffered injuries in his first two seasons in the league, but he could just be unlucky, too. Two seasons is hardly a significant enough sample for us to declare with any sort of confidence if Murray is injury prone or not.
That doesn’t mean we should project Murray to play every snap because 1) running backs as a whole get injured often and 2) the Cowboys drafted Joseph Randle. If the Cowboys run 1,100 offensive plays this year, Murray will probably be on the field for around 660 of them—60 percent. That number factors in the risk of injury.
In his rookie year, Murray touched the ball on 51.0 percent of his snaps, and that number dropped to 40.7 percent in 2012. If it evens out at around, say, 44.0 percent in 2013, Murray would touch the ball 290 times.
And a Barry Church projection:
Church is an interesting player because he has such limited experience. He’s played only 399 defensive snaps during his three-year career; that’s the equivalent of less than half of one full season for a starter. That means that his stats aren’t as important as those for other players, and we almost need to treat him as though he’s never played.
Prior to the 2012 season, Church actually led all Cowboys in tackle rate at 10.5 percent. Again, it came in limited time, but Church seemed to use his short-area quickness to make plays in tight areas. He also looked improved in pass coverage in 2012; he appeared lighter on his feet and yielded only three receptions on seven attempts.
Since we don’t have too much of a basis to predict how Church will play in 2013, perhaps the best way to project his stats is to look at comparable players in similar schemes. Church’s closest comp is Seahawks safety Kam Chancellor. New Cowboys defensive coordinator Monte Kiffin has stated that he wants to play a lot of Cover 3 like Seattle did in 2012. In that defense, the strong safety has “curl-to-flat” responsibility, meaning he’s down in the box. That’s where Church excels.
A look at the Cowboys’ late-game heroics (or poor starts, depending on your viewpoint):
Coming Out Cold
As you examine the Cowboys’ 2012 win probability graphs, one of the main things that stands out is exactly how poorly the team started games. I calculated the Cowboys’ average win probability after each quarter and charted it below.
In the typical game, the Cowboys owned only a 43.0 percent probability to win after the first quarter. By the time halftime rolled around, the Cowboys had just a 36.9 percent chance to win, on average. With the average team’s halftime win probability obviously 50 percent, it’s clear that the Cowboys were pretty poor to begin games in 2012. How bad? Well, if the Cowboys have the same amount of first half success in 2013, an overall 36.9 percent halftime win probability, their most likely final season record will be 6-10.
It’s pretty amazing to see the staggering jump in win probability after the third quarter. At the end of a game, the Cowboys have obviously either won or lost, meaning their “win probability” is either zero or 100 percent. It’s pretty remarkably that they increased their win probability from 35.0 percent at the end of the third quarter to 50 percent, an 8-8 record, at the end of games.
Why passing more often on first down will help the Cowboys’ offense:
In 2012, the Cowboys ranked 15th in the NFL in points scored and 11th in yards per play, yet they managed to rank third in an important offensive category: three-and-out percentage. The Cowboys went three-and-out on just 17.2 percent of their offensive possessions—the third-best mark in the NFL, behind only New England and Green Bay. Nice company. If you’re curious, the league average was 22.7 percent, while the average for the playoff teams was just 20.6 percent, per STATS, Inc.
So why were the Cowboys so good at starting drives? One reason is that Jason Garrett called a lot of passes on first down and early in drives. The Cowboys threw the ball on 58.7 percent of first downs—including 54.0 percent in the first half—when the overall league average was only 47.8 percent.
And believe it or not, the team should be passing way more on first down. On most areas of the field, a four-yard gain on first down is considered a “neutral” play, i.e. it doesn’t help the offense. Only gains of five yards or more are really beneficial on first down. The ‘Boys averaged 7.06 yards on first down passes and 3.33 yards on first down runs.
The Cowboys’ use of analytics needs to change:
Analytics in the NFL
All of this data isn’t meant simply for amusement. Analytics can have profound effects on team success if properly embraced. Numbers are the primary catalyst behind the surge in pass rates and the rise in running back by committee, for example, but there’s still plenty of work to be done. The Cowboys have not been one of the teams to fully embrace the data. The Patriots, Packers, Saints, and Ravens are among those teams that have. That’s not a coincidence.
The next generation of data-driven coaches could be right around the corner. Actually, we’re probably going to see a new approach to NFL decision making in Philadelphia with Chip Kelly. If Kelly and the Eagles succeed, others will follow.
The Cowboys have too often been behind the curve, adopting trends only after too many other teams have done the same such that there’s no longer a competitive advantage to be had. If Dallas wants to be a perennial contender, they need to become a leader, embracing the wave of big data.