Buffalo Bills coach Sean McDermott twice erred in passing up opportunities to go for two in the fourth quarter of the Bills' 33-27 overtime loss to the Tampa Bay Buccaneers on Sunday, according to ESPN's win probability model.
The decisions were costly. While going for two would not have guaranteed victory, a different choice following either touchdown would have pushed the odds in Buffalo's favor and could have been the difference between a win and a loss.
Let's get into each of those decisions and explain why McDermott made incorrect choices:
Decision No. 1
Bills score touchdown with 9:07 remaining, putting them down 11 points (27-16)
This is essentially the same situation as the "down-eight-go-for-two" strategy that many teams have begun to employ, just with Buffalo needing an additional field goal.
The idea is that the Bills were going to have to score at least a touchdown and a field goal to have a chance to win the game no matter what. Had they gone for two in this situation and converted, a touchdown (and PAT) plus a field goal would have won them the game in regulation. Had they failed on the 2-point try, they could have gone for two again on the second touchdown and still been in position to tie with a field goal. Only if they failed twice would the Bills have lost in regulation.
To put it simply: It's much more likely to convert once than fail twice.
At the time, the win probability sacrificed appeared minimal: Buffalo's chance to win the game dropped from 6.6% to 5.8% by incorrectly opting for a PAT. That win probability difference, however, is muted by the fact that the Bills were down big late in the game. If we knew they were going to score a touchdown and field goal again -- the only scenario that mattered, and thus should have been planned for -- the difference would be large.
Here's another way to show how clear-cut this choice was: Our model estimated that the Bills needed to have a 28% chance of conversion to justify going for two. Based on league averages, their offense and the Bucs' defense, we estimated Buffalo's chance of conversion at 48%. That's a huge difference that indicates there is no reasonable justification for kicking the PAT.
Decision No. 2
Bills score touchdown with 4:53 remaining, putting them down four points (27-23)
After making the incorrect choice on the first touchdown, McDermott faced another 2-point decision on the next score.
Even live, not knowing Buffalo would definitely have another opportunity to score, the win probability heavily favored going for 2. It gave the Bills a 25.4% chance to win by going for two, and 23.3% kicking the PAT. The break-even percentage was a 33% conversion rate so again, this should have been an easy choice to go for two.
Why? I like to think of this as, in a way, finding out the result of overtime in advance. Assume the Bills kick a PAT, like they did, and then kick a field goal, like they did. That puts them into overtime, where they have a slightly less than 50-50 chance to win the game. If they could have known the overtime result in advance that would have affected their decision-making.
Let's say the Bills actually went for two here and failed, so they were down 27-23. Later, when they faced fourth-and-2 from the Tampa Bay 7-yard line with 22 seconds remaining, they would have known to go for it on fourth down because they would know they would have needed a touchdown.
Conversely, had they converted and been down 27-25, they would have known a field goal would win them the game and kicked.
Because a 2-point conversion is almost a 50-50, just like overtime, the Bills finding out with five minutes remaining whether they needed a touchdown or field goal to win effectively cost them nothing and let them optimize their decisions for the rest of the game.
Should Josh Allen's health have been a factor?
No. Allen was limping around and wore a walking boot on his left leg after the game, but that shouldn't have come into play here. While context around matchup and health should play a role in decision-making, these choices were not at the margin. There's no normal scenario in which an NFL team should expect to convert a 2-point conversion less than 28% of the time.
If anything, Allen's health was all the more reason to try to decide the game in regulation rather than play for overtime, which only extended the disadvantage of playing with a banged-up quarterback and increased the time he spent playing hurt.