[stupac2]

There are a lot of reasons to have expected Green Bay to be able to score even if Montgomery had taken a knee, but on balance, an abstract return would have increased the teams ability to score; the downside of a below average return was small compared to the upside of a big return.

If you only look at yardage, maybe*, but you can't ignore time. As I already explained having the first play of the drive be before the 2-minute warning was really big because it expanded the Packers' options, and effectively game them another time out, which is very valuable (a full set of TOs is worth nearly as much as turnover, apparently). Losing that stoppage means the Rams can focus more attention on guarding the boundary because they know most plays will be going there. I think that's more valuable than whatever "variance" gains you're citing.


*I still think you're wrong about this, but less wrong than ignoring time. I booted up PFR's win probability calculator to check out some scenarios (and factoring that he was a few yards deep in the end zone when he caught the punt):

1) 1st and 10 from the Rams 25, 2:05 to go: 0.3868 WP.
2) 1st and 10 from the Rams 40, 1:55 to go: 0.4388 WP.
3) 1st and 10 from the Rams 20, 1:55 to go: 0.3625 WP.

I'm not sure if this model includes the value of the timeout, and if it doesn't, well there you go, even his best return isn't going to get this done. So let's assume it does. The question becomes "What is the chance of getting that long return, and does it outweigh the odds of something bad happening by roughly 8X?" (8X because you're trying to increase your odds of winning by 0.05, vs going from 0.37 to roughly 0.) As far as I can tell Montgomery hadn't had a return longer than 30 yards since 2016, he didn't have any returns at all last year. Being more generous, of his 32 returns before yesterday it looks like ~3 went for distance, so let's say his chance of hitting a good return is somewhere between 5-10%. His fumble rate before yesterday was right around 1%, though that has a small sample size issue.

But you see the issue right away, the odds of him fumbling times the WP lost is right in the range of the odds of him busting a good one vs the WP gained. I can understand making that trade if you're QBed by, say, Blake Bortles, but with Aaron Rodgers? It's suspect, to say the least.

[stupac2]

There are a lot of reasons to have expected Green Bay to be able to score even if Montgomery had taken a knee, but on balance, an abstract return would have increased the teams ability to score; the downside of a below average return was small compared to the upside of a big return.

If you only look at yardage, maybe*, but you can't ignore time. As I already explained having the first play of the drive be before the 2-minute warning was really big because it expanded the Packers' options, and effectively game them another time out, which is very valuable (a full set of TOs is worth nearly as much as turnover, apparently). Losing that stoppage means the Rams can focus more attention on guarding the boundary because they know most plays will be going there. I think that's more valuable than whatever "variance" gains you're citing.


*I still think you're wrong about this, but less wrong than ignoring time. I booted up PFR's win probability calculator to check out some scenarios (and factoring that he was a few yards deep in the end zone when he caught the punt):

1) 1st and 10 from the Rams 25, 2:05 to go: 0.3868 WP.
2) 1st and 10 from the Rams 40, 1:55 to go: 0.4388 WP.
3) 1st and 10 from the Rams 20, 1:55 to go: 0.3625 WP.

I'm not sure if this model includes the value of the timeout, and if it doesn't, well there you go, even his best return isn't going to get this done. So let's assume it does. The question becomes "What is the chance of getting that long return, and does it outweigh the odds of something bad happening by roughly 8X?" (8X because you're trying to increase your odds of winning by 0.05, vs going from 0.37 to roughly 0.) As far as I can tell Montgomery hadn't had a return longer than 30 yards since 2016, he didn't have any returns at all last year. Being more generous, of his 32 returns before yesterday it looks like ~3 went for distance, so let's say his chance of hitting a good return is somewhere between 5-10%. His fumble rate before yesterday was right around 1%, though that has a small sample size issue.

But you see the issue right away, the odds of him fumbling times the WP lost is right in the range of the odds of him busting a good one vs the WP gained. I can understand making that trade if you're QBed by, say, Blake Bortles, but with Aaron Rodgers? It's suspect, to say the least.