Doug Pederson followed the numbers, bucked convention and got aggressive on fourth down during last year’s run to the Super Bowl LII title.
On Sunday, the Philadelphia Eagles coach made another unconventional choice that is backed by data and has long been touted by those in the analytics community as an obvious move: going for a two-point conversion while down by eight points late in the game. And though his team lost, it was still the correct decision.
Let's set up the situation: The Eagles trailed the Minnesota Vikings by 14 points early in the fourth quarter before Carson Wentz threw a touchdown pass to Wendell Smallwood with 12:05 remaining. That cut the lead to eight, then the Eagles went for (and succeeded) in going for two.
The numbers indicate the decision was a no-brainer.
In order for any of this to matter, the Eagles were going to have score another touchdown, so let’s assume for the sake of argument the only other score in regulation was going to be another Philadelphia touchdown. And to set this up, let’s use league-average success rates since 2016 -- 94 percent on PAT conversions and 48.9 percent on two-point conversions -- as reasonable expectations for the Eagles. Jake Elliott's PAT rate during the same time frame is also 94 percent.
Going for it, like Pederson did, presents four possible scenarios -- as listed below, along with their likelihood of occurring based on those league-average percentages:
Convert on the two-point attempt, successfully kick a PAT after the second touchdown and win: 46 percent
Convert on the two-point attempt, but miss the PAT and tie: 3 percent
Fail on the first two-point attempt, but succeed on the second and tie: 25 percent
Fail on both two-point attempts and lose: 26 percent
For those keeping score at home, that’s 46 percent to win, 28 percent to tie and 26 percent to lose. Since we’re assuming that’s the only other score in the game, let’s just split the 28 percent evenly between the two teams in overtime for simplicity -- leaving us with the Eagles having a 60 percent chance to win and a 40 percent chance to lose.
Let’s compare that to the odds with a decision to kick the PAT while down by eight points. There also are four scenarios here:
Convert both PATs and tie: 88 percent
Convert first PAT, but miss second and lose: 6 percent
Miss first PAT, convert a two-point conversion and tie: 3 percent
Miss first PAT, fail at a two-point conversion and lose: 3 percent
In this case, it’s 0 percent to win, 91 percent to tie and 9 percent to lose. Or, after factoring in overtime, 45.5 percent to win and 55.5 percent to lose. Much worse.
The key here is that by going for two the first time, a coach knows the result of the first attempt before deciding how to proceed on the second.
Now at the time, the Eagles couldn’t bank on being the only team left to score -- and that their score would be a touchdown. The difference between kicking a field goal while down by seven points as compared to down by eight, however, is minimal late in the game. And while the Vikings could have scored again and muddled the path, any situation in which the Vikings extended their lead would have made it very unlikely for the Eagles to win.
None of that is enough to deviate from Pederson’s controversial but (most likely) data-driven decision to go for two while down eight.