Hint
McCarthy and the NGS Decision Guide were in full agreement on all four of the Cowboys' fourth-down situations against the Patriots in the first half, three of which were strong
go-for-it recommendations, according to our model. The second half, however, was a different story.
FIRST QUARTER: With 12:04 remaining, the Cowboys (tied 0-0) faced a fourth-and-1 from their own 34-yard line. Ezekiel Elliott was stuffed for no gain, and the Patriots took over.
We care more about the process than outcome when it comes to evaluating fourth-down decisions. In this case, McCarthy was right to go for it, despite the offense not converting on the play. The NGS Decision Guide recommended Dallas go for it instead of punt by a difference of 2.6 percentage points in win probability value. This was driven heavily by the Cowboys' chances of converting if they went for it (73.9 percent), which increases even more if they called a run play (77.2 percent).
FIRST QUARTER: With 2:30 remaining, the Cowboys (trailing 14-7) faced a fourth-and-1 from their own 47-yard line and successfully converted on an Elliott run that bounced outside the tackles.
Another fourth-and-short, another Elliott run call. With offensive lineman
Connor McGovern lined up as a fullback, Elliott took the handoff and cut sharply to the left edge to pick up 3 yards and earn a new set of downs. We had that play as a 5.7 percent decision in favor of going for it.
SECOND QUARTER: With 1:32 remaining and the Cowboys (trailing 14-10) facing a fourth-and-goal from the Patriots' 1-yard line, Dak Prescott fumbles prior to crossing the goal line on a QB sneak attempt.
After failing to score on three previous tries from the 1-yard line, McCarthy once again opted to go for it. Our model strongly supported this decision, with a 10.4 percentage-point difference in win probability value. Although the decision meshed with the analytics, the failed attempt did result in a loss of 24 percentage points in expected win probability value.
FOURTH QUARTER: With 2:47 remaining and the Cowboys (trailing 21-20) facing a fourth-and-2 from the Patriots' 33-yard line, Dallas’ Greg Zuerlein misses a 51-yard field-goal attempt.
McCarthy's fourth-down aggressiveness hit an abrupt wall as the game reached its final moments. Per the NGS Decision Guide, going for it here would've yielded 4.6 percentage points in win probability value. Part of the model’s recommendation accounted for the conversion probabilities of going for it (59%) versus kicking a field goal (55%). Furthermore, Dallas’ win probability would have increased to 70 percent with a successful conversion. A conversion also would have allowed Dallas to burn more clock to set up a potential game-winning field goal with significantly less time remaining.
FOURTH QUARTER: With 24 seconds remaining and the Cowboys (trailing 29-26) facing a fourth-and-1 from the Patriots' 31-yard line, Zuerlein makes a 49-yard field goal.
Let's ignore the fact that McCarthy oddly called one of his two remaining timeouts to kick a field goal with 24 seconds left instead of letting the clock wind down to sub-5 seconds first. Given the situation, and the timeout, the optimal call was to go for it by 16 percentage points of win probability value.
The NGS Decision Guide estimates the Cowboys had a 74 percent probability of converting the fourth-and-1, compared to a 62 percent probability of making the 49-yard field-goal try. Ultimately, Zuerlein effectively sent the game into overtime, though it was still a risky decision, given that the guide pegged a make as giving the Cowboys -- with one timeout left -- a 58 percent chance (roughly the chances of winning in overtime) to win. If the Cowboys had converted on the fourth-and-1, however, they would have been able to either take their chances at the end zone or try a closer, more high-percentage field goal.
