With two minutes remaining in regulation in the Cowboys' thrilling overtime victory over the Pittsburgh Steelers, head coach Jason Garrett had to make what I considered to be one of his most difficult decisions all year. In a tie game, the 'Boys faced fourth-and-1 at their own 21-yard line. Nearly all NFL coaches would punt in normal game situations, but Dallas wasn't in a "normal" situation; they surely figured that if they let go of the ball, the next time they'd see it would be overtime, if at all.
If we assume Brian Moorman would normally punt the ball 45 yards in that situation, Pittsburgh would be set up with a first-and-10 at their own 34-yard line, needing around 25 or so yards to attempt a game-winning field goal. Moorman ended up getting a good bounce for a 59-yard punt that really benefited Dallas, but Garrett could have never expected that. At least it certainly wouldn't have been a component of his decision.
With three timeouts and just under two minutes remaining, the Cowboys' chances of stopping Pittsburgh from driving to get into field goal range, based on historical league data, was around 72 percent. The decision to punt gave the Cowboys just around a 30 percent chance to win the game.
Had the Cowboys gone for it and failed, they would have left Pittsburgh with a surefire field goal, at least. However, NFL offenses have converted on fourth-and-1 on 64.4 percent of plays since 2009, and Dallas actually has been surprisingly efficient in short-yardage this year. Based on their field position and the game situation, the break-even expected conversion rate to make going for it the right call was only 33 percent – the 'Boys surely could be expected to rush for a first down on fourth-and-1 on greater than 1-in-3 plays. Even with the risks of failing, the Cowboys' win probability if they went for it would have been 49 percent, just around their expected win probability if they headed straight to overtime.
Intuitively, this should make sense because if Dallas went for it and made it, they'd pretty much guarantee overtime (at worst) with a shot to win it in regulation. If they had failed, they'd still be left with time to at least try to send the game into overtime. The fact that Moorman reeled off a ridiculous punt doesn't change the merits of Garrett's decisions (good or bad).
After the Cowboys' defense sacked Ben Roethlisberger on two occasions, Dallas managed to regain possession in the fourth quarter. Not only that, but a sensational Dwayne Harris punt return set them up in Pittsburgh territory. With a first-and-10 at the Steelers' 49-yard line, Dallas actually had a 72 percent chance to win the game.
After the offense stumbled, Tony Romo & Co. faced a fourth-and-4 at the Steelers' 43-yard line with 32 seconds left to play. With the game still tied, Garrett decided to punt. Was it the right move?
This decision was so difficult for Garrett because in addition to going for it, he also could have attempted a 61-yard field goal. Actually, he initially sent Dan Bailey onto the field to potentially attempt the kick before calling for a timeout.
In punting, Garrett was basically taking a 50 percent chance of winning; the Steelers were unlikely to even attempt a drive with so little time and, if you considered the Cowboys and Steelers to be a coin flip before the game, the same would be true in overtime.
Had Bailey stayed on the field to attempt the long field goal, the percentages wouldn't have been in the Cowboys' favor. Kickers have historically made only around 40 percent of such kicks since 2004, already making it a worse option than punting. When you factor that a miss would leave the Steelers with the ball at midfield, the decision was a no-brainer.
But what about going for it? We know in normal game situations that offenses should stay on the field on fourth-and-4 at the opponent's 43-yard line. In that area, teams have averaged 2.72 points per drive when they've gone for it (compared to losing 0.04 expected points by punting).
This was far from a normal game situation, however, making the two options extremely close. While a punt guaranteed overtime and a successful fourth-down attempt would greatly enhance the Cowboys' ability to win in regulation, a failed try would leave the Steelers around 15 yards out of field goal range with about 25 seconds on the clock.
In the end, I think Garrett made the correct call to punt on fourth-and-4. While I would never advocate punting in that situation in the second or third quarters, the strange situation Dallas was in really altered the choice. Further, there simply aren't enough numbers out there to conclusively determine the merits of going for it in such a situation; it's pretty unlikely that another team has faced fourth-and-4 at the opponent's 43-yard line with 32 seconds to play in a tie game, ever.
Many times, we can use stats to figure out whether or not a particular call increased or decreased a team's win probability; there are all sorts of innovative and useful figures to help coaches in traditional game situations. Once we approach the end of games, however, those choices become far more difficult. Coaches can (and should) still use data to make informed decisions, but it isn't so black-and-white like it is in the fourth quarter.
Whether or not you agree with Garrett's calls, you can certainly appreciate the difficulty of some of his decisions, especially late in games.