When Mike McCarthy took a year off from coaching in 2019, he dove into, among other things, analytics. An old-school coach turned new-school, with a fresh understanding of the numbers and a professed open mind at how to use them.
In the abstract, it was slightly intriguing. To see it play out on the field has been far more interesting.
Last Sunday in the eventual win over the Atlanta Falcons, McCarthy's Dallas Cowboys scored to cut the lead to 39-30 with 4:57 left, then chose to go for two -- rather than cut the lead to eight for a one-score game. Heads exploded across America.
But in the offices of Pro Football Focus, which McCarthy famously visited during his sabbatical, they simply nodded. By the numbers, it was a more than reasonable gamble.
"It's close mathematically," said PFF's Eric Eager, the Executive Director of Research and Development, "but when you put everything together, it's the right choice. And there are a few things that make going for two the right solution."
Here is the math of it, per PFF: With 4:57 left in the game, if the game is 39-30 Falcons, they have a 91.4 percent chance to win. If the game is 39-31 they have a 89.2 percent chance to win and if the game is 39-32 they are at 85.9 percent to win.
Essentially, if the PAT goes in 100 percent of the time, going for two works if they can convert a two-pointer 40 percent of the time.
PFF, which supplies info for all 32 teams and actively consults with many NFL teams, hosted McCarthy in the offseason and Eager was there.
He explained why McCarthy made a sound decision.
The first reason is that if you get the two points, Eager said, you have the opportunity to win the game with a second TD. On the flip side, if you miss the two-pointer and they come back and score, you're still only down two scores.
"Being down nine isn't all that much worse than being down eight if you give up a score," Eager said.
The other aspect -- and this is the low-hanging fruit of it -- it affects how you and your opponent approach the game. At 39-30, your opponent believes they are up two scores and they are less likely to be aggressive. You're going to get the opportunity to get the ball back and maybe get the two scores you need (which is what happened).
With 4:57 to go, the Cowboys were down 15 points and Dak Prescott found tight end Dalton Schultz for a 10-yard touchdown. They now were down by nine -- before the PAT. They could either kick the PAT and almost certainly trail by eight. Or attempt the two-pointer and either go down by seven or stay down by two scores at nine.
They chose the two-point-conversion. It failed. And Dallas still came back to win. After the game, McCarthy responded about it.
"The decision to go for two there is simple mathematics, where you'd rather know if it's a two-score game at the earliest time instead of taking it all the way down to the end," McCarthy said.
In other words, down 15, you know you need a two-pointer at some point, try it earlier and gather facts earlier. The flip side would be, extend the game as long as you can. That's the old-school (normal) way.
"You get rid of uncertainty," Eager said. "If you'd kicked a PAT to go down 39-31, then you drive down and score, all it gets you is the chance to get a 2-point conversion. You think the game comes down to one play, but it's actually a play and a two-point conversion and then overtime."
If you go for 2, the reasoning goes, you have the same info but earlier.
When McCarthy visited PFF, those in the room believed he was committed to learning. Now, he's showing he bought in.
"He's clearly figuring it out," Eager said.