Sitting in a three-way tie atop the NFC East with an 8-6 record, the Cowboys now control their own destiny to the postseason. To take down the division, Dallas will need more of what we saw from the offense in Sunday's 27-24 overtime victory over Pittsburgh.
- You have to give head coach Jason Garrett credit for the recent upswing in play-action passes. Play-action passes are underutilized throughout the league; of the 28 quarterbacks who have taken at least half of their team's snaps, only five have worse efficiency on play-action passes than straight dropbacks. Coming into the game, the 'Boys were ranked last in the NFL in play-action pass rate with only four per contest. Garrett seems to have noticed that play-action passes can be a valuable tool for Dallas, as the 'Boys have now run 18 of them in the last two games alone.
- Against the Steelers, Garrett dialed up 10 play-action passes. One of them resulted in a sack – the first of the year given up on a play-action look – and another was the failed bootleg attempt on third-and-1. That left Romo with eight actual passes on play-action looks, and he made the most of them. Romo went 7-for-8 for 121 yards and two touchdowns on his play-action passes, recording 15.1 yards per attempt (YPA) and a perfect passer rating of 158.3. The lone incompletion was a drop by Lawrence Vickers.
- The most obvious reason for the increase in play-action looks is the return of DeMarco Murray, although I'm not sure Murray's presence has much to do with the success of the looks. While a solid rushing game is always a good thing, it actually isn't a prerequisite for play-action passes. Of the 16 teams ranked in the bottom half of the NFL in yards per carry (YPC), 13 have higher YPA on play-action passes than all other attempts. That includes Dallas.
- Along with the play-action passes came more blocking from Jason Witten. Whereas the tight end had stayed in to block on only seven passes in the last three games combined, Witten was used in pass protection on 11 occasions against Pittsburgh. One resulted in the lone sack on Romo. Romo completed seven of the other 10 passes for 122 yards and a touchdown. Even though it's generally optimal to have Witten in a route, using him in pass protection from time to time can make it more difficult for defenses to decipher runs and passes, and that seemed to work out for Dallas on Sunday.
- It's tough to sell a negative play being a good call, but Garrett's decision to call a naked bootleg on third-and-1 at midfield wasn't necessarily a poor one. It isn't like the Cowboys call that play often; it was just the second naked bootleg for Romo all year, and the other one was a 1-yard touchdown. The play was there to be had, but the 'Boys simply didn't execute one block. Nonetheless, if the job of an offensive coordinator is to maximize his team's chances of succeeding, Garrett did his job with that call. The fact that it didn't work out shouldn't be used to retroactively grade the decision.
- So how can you judge the merits of a particular call if you can't use the result? That's one reason it's beneficial to understand advanced stats. Sure, a punt on fourth-and-1 at the opponent's 35-yard line might work out in favor of the punting team from time to time, but historic game data tells us that punting is generally a poor choice in that situation no matter how it turns out. Over large sample sizes, the percentages always win out. Sure, stats can be misleading, but that's really an error on the part of how we interpret them, not something inherent to the numbers. The ultimate goal of any stat is to be predictive, and some stats are more predictive of future success than others. That's why I often cite numbers like net-YPA and run success rate; they're highly predictive and thus far more useful than things like bulk stats. It's also why I told you before the game that the Cowboys were going to be able to run on the Steelers; Pittsburgh was ranked fourth in the NFL in YPC allowed, but 22nd in run success rate, a stat that accounts for game situations and is thus far more reflective of a defense's true ability to stop the run than YPC. The Cowboys' running backs ended up averaging 5.35 YPC.
- The Cowboys came out victorious in this game because they were able to win first downs. Contrary to popular belief, that doesn't mean setting up "manageable" second and third downs; it means gaining as many yards as possible on a down on which defenses typically still play the run. As they should generally do, Dallas passed the ball on first down quite a bit. Actually, Garrett called a pass on 75.8 percent of the Cowboys' first-down plays, including 11 straight to start the second half. Romo was sacked on one of those and ran for four yards on another. On his 23 first-down passes, Romo was 17-for-23 for 231 yards (10.0 YPA), two touchdowns and no picks.
- One of the reasons first-down passes tend to be successful is that they're somewhat unexpected; NFL teams as a whole still run more than they pass on first down. Similarly, third-down runs are generally successful (even up to third-and-5) because they possess the element of surprise. That's why Garrett's decision to run on third-and-Goal from the 3-yard line in the fourth quarter was so smart, and Murray ended up punching the ball into the end zone for the score. On top of that, the Cowboys were down by a touchdown and likely in four-down territory, so a failed Murray run could still inch them closer to the goal line for a fourth-down attempt. They didn't need it, however, because Garrett displayed a trait that defines all of the league's top play-callers: being unpredictable.
- In addition to passing on first down, Dallas was so efficient through the air because they did a decent job of picking up Pittsburgh's blitzes. I counted the Steelers as sending five or more rushers on 16 of Romo's passes. The quarterback completed 11 of them for 144 yards and a touchdown.
- It appeared as though Garrett made a halftime adjustment to attack the perimeter of Pittsburgh's offense through the air. The 'Boys didn't throw any screens in the first half, but Garrett called four of them in the second half, all to wide receivers. They worked wonderfully, going for gains of 7, 18, 13, and 18 yards.
- Because the Cowboys were having success on the ground and with the short passing game, they didn't take too many shots downfield. Only three of Romo's passes traveled at least 20 yards in the air; two fell incomplete and the other was Bryant's 24-yard score.
- Don't we all just love Brandon Carr right about now?
