Apparently the math says going for two was right

[PghPanther]

The Math had Armstrong ready to set down the Eagle in a boulder laden area of the moon which would have ruined or aborted the mission.......So Armstrong overrode the math and took controls with just enough time to find a clean landing space and saved the mission and possibly the lives of himself and Buzz............

Sometimes the math will tell you what to do but must be overridden in the moment where human assessment can factor more variables specific to the moment......

This was one of them in the game where they should have kicked a PA and if missed could have done the 2 pointer if needed later in the game.

[MHS831]

21 hours ago, TN05 said:

https://predictivefootball.com/late-and-trailing-by-14-points-always-go-for-2-after-a-touchdown/
 

football isn’t math, though

Yes it is.

Statistics, averages, probabilities.  When you call the shots, you better know the odds

 

[MHS831]

analytics are one thing, execution is the other.  Every team should have a play that they can execute most of the time.

[Happy Panther]

I'd hope that Rivera could override the analytics. I also don't like criticizing play calls (much). As @MHS831 says its an execution issue. 

In any case I like the call. Plus he sure looked in from a general sense.

[Happy Panther]

10 hours ago, BrianS said:

I hated the call at the time, but once I sat down and thought about it, it made sense.  There were a lot of factors there.

Think about the entire process.  You need two touchdowns + 2 points total to tie.  That part isn't negotiable.

By trying for two on both TD's, you actually give yourself more opportunities to experience a better outcome.  The conversion odds are 50%, so you have to feel like you're going to get at least one of them.

If you make the first one, then instead of needing a TD + 1 pt to tie, now the TD ties the game and a 1 pt kick wins it.

If you miss the first one, you try the second and either make it and tie, or fail it and lose.  Kicking extra points on both gives you NO chance to win.  Try for 2's gives you the only opportunity to win the game, and even odds on tying versus just kicking extra points.

Yes kicking two XPS gives you a 50% to win in overtime.

Going for 2 gives you a 50% chance to win if you convert. If you don't convert, have another chance to convert sending you into overtime which is a 50/50 coin flip. So its 50% + whatever the odds are you convert the second 2pt conversion and win in overtime  - in this example 12.5%.

So 50% vs 62.5%.

This all assumes you get the second touchdown.

How would the snow affect this? It would lower the odds of kicking but possibly help running backs (internet is split on this). In which case the 12.5% advantage could be higher.

[Happy Panther]

 

10 hours ago, Cavscout said:

Does it take into account Green Bay 48% red zone scoring defense?

 

10 hours ago, Jeremy Igo said:

Does the math take into account all of the variables of the moment? Weather, momentum, lack of run defense, etc?

 

6 hours ago, TN05 said:

The math works with standard conversion and EP rates. I assume you can plug in custom variables and get similar results.

Yeah I don't know if we have on the fly analytics in game. But the whole point of analytics is to devour data and produce scenarios. So in a perfect stats scenario, some geek is crunching numbers real time, has already adjusted for GB specific variables and the weather. If not I would bet the analytics department is doing some retro testing this week. I imagine baseball has stuff like this and probably the NBA.

[Call Me James]

Didn't matter anyway. We lost by 8

 

[OneBadCat]

I thought it made sense because Ron wasn't expecting us to get the ball two more times to win it. So if you convert the 2pt and score another TD with XP you have the lead. If you fail, score one more time, you still have a chance to tie with a 2 pt conversion. With the way the game was going and the officiating he probably didn't want to chance OT with Rodgers. He was taking a chance to control our fate with the ball in our hands.

[Sgt Schultz]

The problem is the numbers assume a 50% success rate.  This year, the Panthers were 1 for 2 prior to yesterday, but that is a very small sample size to conclude the rate is solid. In 2018 they were 40%.  In 2017 they were 33%.  You have to go back to 2016 to find a their latest year at 50%.

So, the 50% success assumption is questionable, at best.  The problem is not only that they have not been a 50% team the last few years, but whatever their success rate has been, the sample size is too small to be useful as a number.

The problem with playing by the numbers is the underlying assumption is that the numbers are good.  At best, on this subject as it pertains to this team, I think that assumption very questionable statistically.

[Happy Panther]

36 minutes ago, Sgt Schultz said:

The problem is the numbers assume a 50% success rate.  This year, the Panthers were 1 for 2 prior to yesterday, but that is a very small sample size to conclude the rate is solid. In 2018 they were 40%.  In 2017 they were 33%.  You have to go back to 2016 to find a their latest year at 50%.

So, the 50% success assumption is questionable, at best.  The problem is not only that they have not been a 50% team the last few years, but whatever their success rate has been, the sample size is too small to be useful as a number.

The problem with playing by the numbers is the underlying assumption is that the numbers are good.  At best, on this subject as it pertains to this team, I think that assumption very questionable statistically.

Yeah you don't rely 100% on the teams numbers necessarily. You would credibility weight the teams estimate with the leagues estimate and maybe with some overall estimate.

The article uses 50% because it resembles a coin flip which is easier to understand. It's a generic NFL coin flip exercise.

A real analysis would use 42.8% for 2pt conversions and say 95% for extra points. I made those numbers up. 

We have no idea what actually happened.

[Diehardpanther02]

Our redzone offense has been suspect. Most of our TDS have been decent yardage or long CMC runs. Our short yardage offense has been bad. Mainly due to OLINE play. We have players who are very good in space on O, but if we have to crunch it through the middle that is our weakness. CMC can get a yard or two but not when 2 opposing DL are in his face and he's getting tackled by our own Oline.

 

 

And I like GVR.

[JARROD]

Almost as nuts as going for a fake punt on the wrong side of the field!

[Sgt Schultz]

49 minutes ago, Happy Panther said:

Yeah you don't rely 100% on the teams numbers necessarily. You would credibility weight the teams estimate with the leagues estimate and maybe with some overall estimate.

The article uses 50% because it resembles a coin flip which is easier to understand. It's a generic NFL coin flip exercise.

A real analysis would use 42.8% for 2pt conversions and say 95% for extra points. I made those numbers up. 

We have no idea what actually happened.

Your 95% number meshes with the league's average for PATs last year, I think.  At least that is the number I have seen used in a lot of analysis.  At one point last year the league's success rate on 2-pointers was about 60%, but I don't know what it was by the end of the year.

The number not only has to be tailored to just the team in question, but then factor in whether the team in question is the team on the field.  For example, the league does better on 2-pointers running the ball than passing (I think the success rates were something like 65% to 45%, respectively).  Let's say the Panthers are consistent with that.  But, do our numbers include Cam Newton in addition to CMC?  If not, well, Houston, we have a problem.  Or, our best OL is sitting on the sidelines with ice on his ankle (using best with our OL is probably a bit much). Or, the opposing defense is well above the center line of the opponents that our number consists of. 

Then there is the 2-ton gorilla of how much of the 2-point data pool is when the game is on the line vs. in a 34-13 game?  The data pool is small enough to start with, but to get a good number you have to reduce it to the point of being insignificant.

Relying on the numbers alone just doesn't capture all the variables, especially with a play as infrequent and situation-dependent as 2-point conversions.  As you said earlier, you'd like to think the coach would have the liberty to effectively say "that's nice" while he trots out the kicker.  Asking the analytics team to admit "here are our numbers, but we think they are useless" is a tall order.