The math behind going for 2 when down by 14 late in game and scoring a TD.

[CbusLion]

This was already covered earlier in the thread. While it will certainly change the math, it is unlikely to overcome the 12.5% edge.

You're assuming acceptance of the simplified math that yields the 12.5%, which is fine for this conversation, but an individual team's historical conversion success rate can absolutely be enough below 50% that this doesn't work. With your math, a 42% or worse 2pt success rate would tell you to kick it. Look at some actual success rates by team, that alone shows that you can't use broad averages and apply them to your specific scenario.
I agree that 62.5% > 50%, if it's hypothetically that simple sure go for 2. But it isn't. I ultimately don't have a problem going for 2 under a lot of circumstances, but there's a lot that goes into actual analytics and decision science and the danger is making poor decisions assuming your data and logic is better than it really is.

 

[NoSoup4U]

As I said, I’m surprised that many do have that same belief…it goes back to announcers/coaches/fans thinking the most important part is “extending” and not “winning” (and using the logic that you can’t win it if you don’t extend it).

My basketball version of it is coaches/fans that are anti-fouling when up 3, since “worst case, you’d still go to OT if they make a 3, whereas with fouling, you bring up the chance of losing in regulation”. Me, I foul even if it means a chance of losing in regulation, since it increases my odds of winning at a macro-level.

Sorry I call ******** on the logic

You obviously havent been in the moment what it means to win Game 6 and move on-- and what it takes to do so

not some flippant answer well we had a better chance to win game 7 this way

Um you couldn't even win game 6 ---- how's that for a macro level view

It has everything to do with playing one game at a time not factoring in 1 game that might not even get to be played......this is not about what fans think the fans are not playing the game, nor what stat boys are computing, they are not playing the game

so I call ******** on your strawman argument that is strictly statistically in nature---- you wouldn't be around long as manager purely playing the odds

by the way you really want to sacrifice game 6 over a 3.5 % delta gain on winning the series??? I would listen more intently if you said you had 20% more chance in winning------ 3.5% in what world is that even statistically significant??? maybe if you have that % to walk again versus 0% after a bad auto accident

sorry fellas I guarantee my teams win more World Series by winning game 6 down 3 games to 2 ---than losing game 6

 

[leinbacker]

That simple math ignores the team-specific variables. Like not being able to convert 2pt attempts at the NCAA averages.

Having more Intel on the opponent's short yardage defense increases your success rate.

Not when they repeatedly stuff you in short yardage situations throughout the game

but math discounts momentum and a gut feel of your players energy

 

[SleepyLion]

And if you lose game 7 then game 6 was meaningless.

Under this thought would not all the games (pre-season, regular season, playoffs) be meaningless if you lose game 7?
If game 6 is meaning less why not game 5 or 4, 3, 2, 1 etc.?

 

[s1uggo72]

He has a point. You never have to draw up plays on a chalk board anymore. Hell, you don't even need paper.
Why draw up plays anyway? Stop thinking and just go out there and beat the snot out of the other team, you know, on the field.

The Eagles tried that this year, I think

 

[CbusLion]

Not when they repeatedly stuff you in short yardage situations throughout the game

but math discounts momentum and a gut feel of your players energy

Exactly, if they're kicking your *** in short yardage, kick the xp's and play for OT.

Math discounts momentum if your math doesn't account for it. I believe the best analytics groups can accommodate for these dynamics in their modeling. Not sure Penn State has that, or not yet.

 

[Grant Green]

You're assuming acceptance of the simplified math that yields the 12.5%, which is fine for this conversation, but an individual team's historical conversion success rate can absolutely be enough below 50% that this doesn't work. With your math, a 42% or worse 2pt success rate would tell you to kick it. Look at some actual success rates by team, that alone shows that you can't use broad averages and apply them to your specific scenario.
I agree that 62.5% > 50%, if it's hypothetically that simple sure go for 2. But it isn't. I ultimately don't have a problem going for 2 under a lot of circumstances, but there's a lot that goes into actual analytics and decision science and the danger is making poor decisions assuming your data and logic is better than it really is.

Ok. Let's say your chances of making 2pt conv are 42%. Here's the new math for going for the 2pt conv. assuming you have a 50% chance of winning in OT (which is probably lower since you likely aren't a very good team relative to your opp).

Chance of making the 2pt (and therefore winning the game) = 42%.
Chance of missing 2pt, but making 2nd 2pt and then winning in OT = 58% X 42% X 50% = 12%
Total probability of winning = 42% + 12% = 54%

Your 2pt rate would have to be as low as ~38% to give yourself only a 50% chance of winning the game. And again, if your 2pt rate is that low, you probably have less than a 50% chance of winning in OT, so the equation must change.

 

[Grant Green]

The Eagles tried that this year, I think

Ha. True. Maybe they needed to play smarter and not harder.

 

[Grant Green]

Under this thought would not all the games (pre-season, regular season, playoffs) be meaningless if you lose game 7?
If game 6 is meaning less why not game 5 or 4, 3, 2, 1 etc.?

I think you are reading my post out of context. It was a reply to "you have to win game 6 to get to game 7".
My point is that you have to win them both. I don't care about winning Game 6 if it gives me an overall worse probability of winning both.

 

[Erial_Lion]

Michigan was 15 points, not 14. PSU scores and rather than making it a one score game, Franklin went for two and kept it as a score game.

If he kicked it instead, it means we only need one drive versus two. Kind of a deflating moment.

That’s what we were talking about in the “scoring to go down 9” scenario… you should go for 2, so that you know if you need one more score or two.

 

[razpsu]

The analytics community will also universally tell you that you should...but it blows people's minds when coaches do it, since "extending the game" seems to be more important than "increasing your chances of winning the game".

Thankfully, Franklin went for 2 in both of these situations this season, so he's being fed the right information...though unfortunately both of them failed.

Didn’t he do it though in the 2nd quarter as opposed to the 4th quarter like Tampa did. 2nd quarter that analytics of going for two makes no sense when the other team has plenty of time to score more points .

 

[Erial_Lion]

Didn’t he do it though in the 2nd quarter as opposed to the 4th quarter like Tampa did. 2nd quarter that analytics of going for two makes no sense when the other team has plenty of time to score more points .

He did do it once early that wasn’t a +EV spot to do it (maybe it was the Michigan game, but not positive)…but he did it late in those games in correct spots to do so.

 

[Nits74]

What the Math doesnt assume is the subsequent psychological mindset when the 2 pt is missed and or made

Mindset they miss:

Doesnt the offense now feel the need to press/sense of urgency only to just tie the game at best??? therefore playing tight making errors in judgement.... AKA Baker on that INT??
Lions playing loose knowing that chewing clock and punting is a viable option once they get the ball up by 8


Ok mindset that they make it:

Feeling confident forcing the Lions to play tight on their possession and hope your D makes a Stop( Bowles D mind at work) yet not forgone that a TD is on the horizon for your offense(as you say its an assumption it could be done-- but they took almost a full qtr to score a TD in Q2 and Q3 and Q4 what makes them think all of a sudden they can score so quickly???).
Lions knowing they can hold the ball methodically and HAVE to play the game they have been playing all day and maybe get a FG game over...punting may not be an option with a lot of time left

why play with a carrot dangling from the stick in either case??? you forced your hand on the turn instead of playing to the river to use Hold em parlance

You beat me to it. Agree 100%

 

[CbusLion]

Ok. Let's say your chances of making 2pt conv are 42%. Here's the new math for going for the 2pt conv. assuming you have a 50% chance of winning in OT (which is probably lower since you likely aren't a very good team relative to your opp).

Chance of making the 2pt (and therefore winning the game) = 42%.
Chance of missing 2pt, but making 2nd 2pt and then winning in OT = 58% X 42% X 50% = 12%
Total probability of winning = 42% + 12% = 54%

Your 2pt rate would have to be as low as ~38% to give yourself only a 50% chance of winning the game. And again, if your 2pt rate is that low, you probably have less than a 50% chance of winning in OT, so the equation must change.

Welp I deleted my math but yours looks good so let's say 38% it is. It is perfectly reasonable for a coach to assume a 38% or lower 2pt success rate factoring in your short yardage offense is weak, home/away crowd, your strong defense helping you more in OT, etc. So down 14 points, you score and then kick the extra point.

Now you're down 7, score again, and reassess your likelihood of 2pt success vs OT success. I believe that it changes throughout the game, and if it's still 38% or anything under your OT chances, kick it again.

If you hypothetically aren't allowed to kick twice and must attempt at least 1 2pt conversion one could argue that your 2pt success rate is higher the later in the game you get, e.g. opposing DL wears down, more game intel on the safeties' tendencies, etc. And that each subsequent 2pt attempt has a lower success rate, where you don't get to assume a success rate of 50% (or 38%) twice, the first is 50% but the second play is only 20%. E.g. you only have 1 golden play you keep in your back pocket, and can't rely on a 2nd 2pt attempt with a lower success rate.

 

[Erial_Lion]

Welp I deleted my math but yours looks good so let's say 38% it is. It is perfectly reasonable for a coach to assume a 38% or lower 2pt success rate factoring in your short yardage offense is weak, home/away crowd, your strong defense helping you more in OT, etc. So down 14 points, you score and then kick the extra point.

Now you're down 7, score again, and reassess your likelihood of 2pt success vs OT success. I believe that it changes throughout the game, and if it's still 38% or anything under your OT chances, kick it again.

If you hypothetically aren't allowed to kick twice and must attempt at least 1 2pt conversion one could argue that your 2pt success rate is higher the later in the game you get, e.g. opposing DL wears down, more game intel on the safeties' tendencies, etc. And that each subsequent 2pt attempt has a lower success rate, where you don't get to assume a success rate of 50% (or 38%) twice, the first is 50% but the second play is only 20%. E.g. you only have 1 golden play you keep in your back pocket, and can't rely on a 2nd 2pt attempt with a lower success rate.

If you’re going to go for 2 after one of your scores, the first is the no-brainer. Going for it after the first score and missing means you’d still have the opportunity to go for 2 after h to e second score to force OT. You wait until the second one, and you’ve lost that other path to victory.

You’re grossly overthinking it if you’re thinking the crowd, time of game, “golden play”, etc will swing your odds of converting a two point conversion that much. People overreact to small sample sizes (ie, thinking that our 0 for 3 on two pointers against Mich/OSU this season would be at all significant in any calculations)…especially when taking into account that going to OT means winning a game against a defense that you apparently think you can’t move the ball against.

 

[Grant Green]

Welp I deleted my math but yours looks good so let's say 38% it is. It is perfectly reasonable for a coach to assume a 38% or lower 2pt success rate factoring in your short yardage offense is weak, home/away crowd, your strong defense helping you more in OT, etc. So down 14 points, you score and then kick the extra point.

Now you're down 7, score again, and reassess your likelihood of 2pt success vs OT success. I believe that it changes throughout the game, and if it's still 38% or anything under your OT chances, kick it again.

If you hypothetically aren't allowed to kick twice and must attempt at least 1 2pt conversion one could argue that your 2pt success rate is higher the later in the game you get, e.g. opposing DL wears down, more game intel on the safeties' tendencies, etc. And that each subsequent 2pt attempt has a lower success rate, where you don't get to assume a success rate of 50% (or 38%) twice, the first is 50% but the second play is only 20%. E.g. you only have 1 golden play you keep in your back pocket, and can't rely on a 2nd 2pt attempt with a lower success rate.

In theory, I agree with you. The best way to do the analysis is to use the actual data produced by the matchup. It would be pure and perfect if you could somehow get close to the actual percentages of one team vs another. Can't argue at all with that logic. However, I think there are not going to be a lot of instances where a team's 2 pt rate is going to be that much lower than their OT win rate (it's really all about comparing the two rates). Maybe PSU was one of those instances vs OSU. Who knows. On the road, at OSU, I have no problem with Franklin trying to win that game in regulation (as improbable as it would have been). I don't think their chance in OT was better than 40%.

 

[CbusLion]

If you’re going to go for 2 after one of your scores, the first is the no-brainer. Going for it after the first score and missing means you’d still have the opportunity to go for 2 after h to e second score to force OT. You wait until the second one, and you’ve lost that other path to victory.

You’re grossly overthinking it if you’re thinking the crowd, time of game, “golden play”, etc will swing your odds of converting a two point conversion that much. People overreact to small sample sizes (ie, thinking that our 0 for 3 on two pointers against Mich/OSU this season would be at all significant in any calculations)…especially when taking into account that going to OT means winning a game against a defense that you apparently think you can’t move the ball against.

We're talking analytics, every dynamic probability and outcome gets layered in, no such thing as overthinking. PSU going 0/3 is just an example that assuming 50% can be off, or way off.

I think the point is that, down 14 pts in the 4th, you hypothetically don't know if you're going to go for a 2pt conv after one/both of your scores, it's possible that kicking the extra points and playing for OT is a viable or potentially preferable path but once you go for 2 on your first TD you've removed the 2 kick scenario.

 

[CbusLion]

In theory, I agree with you. The best way to do the analysis is to use the actual data produced by the matchup. It would be pure and perfect if you could somehow get close to the actual percentages of one team vs another. Can't argue at all with that logic. However, I think there are not going to be a lot of instances where a team's 2 pt rate is going to be that much lower than their OT win rate (it's really all about comparing the two rates). Maybe PSU was one of those instances vs OSU. Who knows. On the road, at OSU, I have no problem with Franklin trying to win that game in regulation (as improbable as it would have been). I don't think their chance in OT was better than 40%.

With how shaky the offense was, I dont either. We had a tough path no matter what we did. The examples where your OT chances are significantly better than your 2pt conversion success is when your defense has been killing the opposing offense or they have a bad kicking game. Doesn't happen often when playing top5 teams, so you could argue as an underdog you should be very aggressive to give yourself the best chance at winning.

Against Michigan, our defense played tough as hell but that OL is legit. Being aggressive after our scores was reasonable although disappointing. Disappointing results doesn't mean the decision was dumb, but don't tell that to the internet.

 

[ODShowtime]

Maybe there's a lot of angst about 2 pt conversions in this fan base because we were forced to watch numerous unsuccessful OT efforts vs. an vastly inferior team? Or a phantom 2 pointer a few years ago?

Sort of 2PTSD for 2 point stress disorder?

I, for one, will always have Penix trauma since that time the Penix came up short.

 

[NewEra 2014]

This is an interesting discussion.

A team’s odds of making a 2 pt conversion go down the more times they attempt a 2 pt conversion. Each time they run a play from that distance when attempting to score a TD during the game also lowers their chances from 2pt range later in the game.

Football analytics are not even close to being good enough to figure out odds in the scenario where a team empties its playbook trying to score from around the 3 yard line during a game.

 

[Bkmtnittany1]

I, for one, will always have Penix trauma since that time the Penix came up short.

Penix stopped short! That’s my move!!!

 

[s1uggo72]

This is an interesting discussion.

A team’s odds of making a 2 pt conversion go down the more times they attempt a 2 pt conversion. Each time they run a play from that distance when attempting to score a TD during the game also lowers their chances from 2pt range later in the game.

Football analytics are not even close to being good enough to figure out odds in the scenario where a team empties its playbook trying to score from around the 3 yard line during a game.

What you talking about? Analytics has it covered from all ends?? lol

 

[Erial_Lion]

 

[Erial_Lion]

This is an interesting discussion.

A team’s odds of making a 2 pt conversion go down the more times they attempt a 2 pt conversion. Each time they run a play from that distance when attempting to score a TD during the game also lowers their chances from 2pt range later in the game.

Is this actually a fact, or an assumption that you are making? I've never seen any data that would point us towards this conclusion, so I'm interested in seeing if it exists.

 

[StamfordBridgeVikingatLeftOffensiveTackle]

This could end up as the McAndrew Board's "Monty Hall Problem."

Why You Should Always Switch: The Monty Hall Problem (Finally) Explained - By Steven Pinker - Behavioral Scientist

The infamous problem even professors and mathematicians got wrong comes down to one unintuitive inference—in the Monty Hall problem, Monty Hall is God.

behavioralscientist.org

 

[Erial_Lion]

This could end up as the McAndrew Board's "Monty Hall Problem."

Why You Should Always Switch: The Monty Hall Problem (Finally) Explained - By Steven Pinker - Behavioral Scientist

The infamous problem even professors and mathematicians got wrong comes down to one unintuitive inference—in the Monty Hall problem, Monty Hall is God.

behavioralscientist.org

Ugh, you’re gonna really put me on tilt if we go down this rabbit hole…my wife still doesn’t get it and says “always trust your gut”.

 

[Grant Green]

This could end up as the McAndrew Board's "Monty Hall Problem."

Why You Should Always Switch: The Monty Hall Problem (Finally) Explained - By Steven Pinker - Behavioral Scientist

The infamous problem even professors and mathematicians got wrong comes down to one unintuitive inference—in the Monty Hall problem, Monty Hall is God.

behavioralscientist.org

That's funny. The show that discussed the 2pt conv yesterday also made reference to the monty problem. I was going to make a post about it today and I'm sure there will be some that don't believe it.

 

[CbusLion]

This is an interesting discussion.

A team’s odds of making a 2 pt conversion go down the more times they attempt a 2 pt conversion. Each time they run a play from that distance when attempting to score a TD during the game also lowers their chances from 2pt range later in the game.

Football analytics are not even close to being good enough to figure out odds in the scenario where a team empties its playbook trying to score from around the 3 yard line during a game.

Sounds like you watched the 9-OT Illinois game from 2021.

 

[LionJim]

Sounds like you watched the 9-OT Illinois game from 2021.

dang.

 

[StamfordBridgeVikingatLeftOffensiveTackle]

Sounds like you watched the 9-OT Illinois game from 2021.

Over and over.

 

[Lanz]

Is this actually a fact, or an assumption that you are making? I've never seen any data that would point us towards this conclusion, so I'm interested in seeing if it exists.

Well teams typically have a few favorite plays that they are good at in the 2pt situations. Once they have run those plays in a game they are forced to either use plays that they are not as good at or re-use plays that the defense has now seen. So logically they will be less successful. However, as with this whole discussion, there are way more factors in play.

 

[Grant Green]

Well teams typically have a few favorite plays that they are good at in the 2pt situations. Once they have run those plays in a game they are forced to either use plays that they are not as good at or re-use plays that the defense has now seen. So logically they will be less successful. However, as with this whole discussion, there are way more factors in play.

I would guess that the probability decreases a bit after the "favorite plays" are run, but then hits a floor. At some point, it becomes a game of rock paper scissors, especially if you can run multiple plays from one formation. Agreed - way more factors at play.
I think Erial's point is that sometimes things seem intuitive (2 pt probability goes down as more are run), but the actual results do not follow intuition.
The Monty Hall problem that Stamford brought up is a great example.

 

[Erial_Lion]

I think Erial's point is that sometimes things seem intuitive (2 pt probability goes down as more are run), but the actual results do not follow intuition.
The Monty Hall problem that Stamford brought up is a great example.

Pretty much exactly where I was going...everyone is focused on the offense having a "favorite play", but defenses also might want to throw a new look out there that, once an offense has seen it, won't be as effective in a subsequent situation. Offenses also get the advantage of seeing how the other team is defending certain situations, which can give them an advantage in subsequent tries.

I've never seen data around the chances of version going up/down as the number of attempts in a game increases...but I'm guessing that the actual results would show much less change that some would expect.

 

[SleepyLion]

Does it make a difference if you only have one shot?

If I'm on Let's Make a Deal, I'm never going to be on it again. So, I get I should switch, if I played the game 3 times I would win twice. But if I only play once does it really matter? All the probabilities are based on large numbers, but I would not have a large number of tries.

Similarly with the lottery. The chances of winning the megapowerballmillions is 1 in a large number, but if you only buy one ticket ever and you win did the odds matter to you? If the risk is small (compared to your total wealth) and not correlated (to other uses of the ticket cost) would it still be a good investment?

 

[NewEra 2014]

Does it make a difference if you only have one shot?

If I'm on Let's Make a Deal, I'm never going to be on it again. So, I get I should switch, if I played the game 3 times I would win twice. But if I only play once does it really matter? All the probabilities are based on large numbers, but I would not have a large number of tries.

Similarly with the lottery. The chances of winning the megapowerballmillions is 1 in a large number, but if you only buy one ticket ever and you win did the odds matter to you? If the risk is small (compared to your total wealth) and not correlated (to other uses of the ticket cost) would it still be a good investment?

Good post. In my view, weak coaches rely solely on analytics to insulate themselves from criticism. There are times that the flow of a game and the specific circumstances of a game dictate a departure from analytics. Analytics cannot capture specific circumstances—they apply to a large sample size and present the “average” result based on the large sample size.

 

[Erial_Lion]

Does it make a difference if you only have one shot?

If I'm on Let's Make a Deal, I'm never going to be on it again. So, I get I should switch, if I played the game 3 times I would win twice. But if I only play once does it really matter? All the probabilities are based on large numbers, but I would not have a large number of tries.

Similarly with the lottery. The chances of winning the megapowerballmillions is 1 in a large number, but if you only buy one ticket ever and you win did the odds matter to you? If the risk is small (compared to your total wealth) and not correlated (to other uses of the ticket cost) would it still be a good investment?

It’s the difference between having a 33% chance of winning the prize of a 67% chance. It doesn’t guarantee anything so you could stick and win or switch and lose, but why ever turn down the opportunity to double you chances of winning the prize?

 

[Erial_Lion]

Good post. In my view, weak coaches rely solely on analytics to insulate themselves from criticism. There are times that the flow of a game and the specific circumstances of a game dictate a departure from analytics. Analytics cannot capture specific circumstances—they apply to a large sample size and present the “average” result based on the large sample size.

No coach would rely “solely” on analytics…but once upon a time, the thought of abandoning mid-range jumpers or shifting 3 players to one side of the infield was looked at as crazy…and when it didn’t work, fans/media members likely said the coach is wrong to implement these new analytics-based strategies, focusing on the small sample sizes.

Coaches that use analytics in their approach are “smart”, not “weak”.

 

[Grant Green]

Does it make a difference if you only have one shot?

If I'm on Let's Make a Deal, I'm never going to be on it again. So, I get I should switch, if I played the game 3 times I would win twice. But if I only play once does it really matter? All the probabilities are based on large numbers, but I would not have a large number of tries.

Yes it matters. If give you one shot to roll a dice to win $1000 and I give you the following options to choose from, which do you take?

Option 1: roll a 1 or a 2 to win the money
Option 2: roll a 3, 4, 5, or 6 to win the money.

 

[CbusLion]

Good post. In my view, weak coaches rely solely on analytics to insulate themselves from criticism. There are times that the flow of a game and the specific circumstances of a game dictate a departure from analytics. Analytics cannot capture specific circumstances—they apply to a large sample size and present the “average” result based on the large sample size.

They can cut up the analytics into very specific splits if they want, there's an enormous amount of data that has been mined and scrubbed. I agree you can't use analytics alone.

A lot of these decisions on 2pt conversions, 4th down play calling, etc. are rarely "dumb". Fans like to complain pretending like they know better or have perfect intuition from their zero playcalling experience. "Why would you run the ball there?!?!??" always makes me laugh.

And sports journalists like to question failed calls after the game to elicit a good response from the coach.

 

[BW Lion]

A lot of people freaked out a few years ago when Doug Pederson started doing this. Now, it's becoming more common as coaches accept it (TB did it on Saturday), but I still see folks protesting. I heard a good explanation of the math today that hopefully makes more sense.

First, it's only valid if your team is able to get a 2nd TD without the other team scoring, so that is the assumed scenario. Also, for simplification purposes, assume that an XP is 100% probability and 2Pt conv is 50% (both are slightly less but that isn't too critical). Also must assume that winning in OT is close to 50%, so this may not be applicable if your team is a heavy favorite.

Option 1: Kick the 2 XPs, go to OT. Probably of winning = 50%

Option 2: Go for 2 on the first TD. There is a 50% chance that you win the game right then and there (remember, we assumed that you will score again and kick an XP).
If you miss it, you still have a 50% chance to make the 2Pt conv on the second TD and then a 50% chance to win in OT. The probably of this scenario playing out is 50% (that you miss the 1st 2pt conv) X 50% (that you make the 2nd 2pt conv) X 50% (that you win in OT) = 12.5%.

Going for 2pt conversion leading to a total probability of winning the game = 50% (make the 1st 2pt conv) + 12.5% (miss 1st 2pt conv, but make 2nd and win in OT) = 62.5%

Your understandably simplistic math excludes the concept of momentum

Much like B1G officiating (t’s not what they call but when they call it) …. the same principle applies. PSU has frequently lost momentum because “what” was called “when”

If a team has the “momentum” they should opt for the most certain and highest probabilistic outcome…perhaps a tie involving high percentage extra points via kicking.