Down 9 and going for 2

[guffus]

I was trying to remember what New England did when it came back from 25 down in the last Super Bowl. I remember they scored a TD and FG early in the comeback, then got 2 TD's late to tie the game. So did they go for 2 on the last TD or 2nd last TD?

Well I looked it up. Turns out the extra point was missed on their first TD, the kick hit the crossbar, so they needed to go for 2 on both the last 2 TD's, which they got both times.

That was truly a great comeback. So many things had to go right for New England. But in the end, that does not help for this discussion.

 

[guffus]

One other example I remember is in 2002, PSU was playing Iowa, PSU came back from 22 points down to tie the game. PSU went for 2 on the 2nd last TD instead of the last TD. In this case, PSU actually faked the extra point, when they got the 2.

Maybe thats what Iowa should have done against Purdue. Faked going for 1, and then go for 2 on the fake. At least then you have element of surprise.

 

[theboat]

One other example I remember is in 2002, PSU was playing Iowa, PSU came back from 22 points down to tie the game. PSU went for 2 on the 2nd last TD instead of the last TD. In this case, PSU actually faked the extra point, when they got the 2.

Maybe thats what Iowa should have done against Purdue. Faked going for 1, and then go for 2 on the fake. At least then you have element of surprise.

Not the same. There was over 4 minutes left. I get the decision at that point

 

[PCHawk]

The thing is, it adds "almost" zero value to know you need two possessions with one minute left. You could still throw a 15 yard pass, kick a field goal, then go for another onside kick with 40 seconds or so left. The odds are extremely small but not zero. The other way adds zero value. That's why you take extremely small odds over zero.

Then you have to decide if you're a human emotion and momentum guy. If you are, you compare the odds of those benefits vs the odds of adding my extremely unlikely comeback scenario. You will probably decide its best to kick it. But since it's close, it's not stupid either way. Which is my whole point.

 

[PCHawk]

Most coaches care more about human emotion than they do giving their team the best probability. That's why most kick it. Well either that or out of ignorance. Just like the scenario of going for 2 down 14. It helps your chances a lot, but no one does it. They want their best chance to extend the game. Not their best chance to win it. That's the going theme around here too.

 

[theboat]

Most coaches care more about human emotion than they do giving their team the best probability. That's why most kick it. Well either that or out of ignorance. Just like the scenario of going for 2 down 14. It helps your chances a lot, but no one does it. They want their best chance to extend the game. Not their best chance to win it. That's the going theme around here too.

You keep saying this, yet you also say that the probability is the same.

 

[PCHawk]

You keep saying this, yet you also say that the probability is the same.

If the probability was the exact same, you kick it. But it's not the exact same. Going for 2 first gives you a tiny statistical advantage. Its very debatable if that advantage is worth the loss of momentum and human emotion.

 

[DJI9424]

Most coaches care more about human emotion than they do giving their team the best probability. That's why most kick it. Well either that or out of ignorance. Just like the scenario of going for 2 down 14. It helps your chances a lot, but no one does it. They want their best chance to extend the game. Not their best chance to win it. That's the going theme around here too.

You don't seem to understand the concept of conditional probability. If you are 14 points down, the reason you don't go for two is the much higher percentage play is to kick the PAT and do the same if you score another TD. It DOES NOT help your chances to win the game if you FAIL to convert the 2 PAT, either on the first TD or the second.

Your argument is based on the assumption you convert the 2 PAT, and then I agree with you, you have a better chance of winning the game than had you converted the 1 PAT. But that is not a certain outcome, that is a conditional outcome AFTER you have converted the first 2 PAT.

Logically, the best probability is to kick both PATs. Now if a coach wants to play the riskier strategy, that's certainly his prerogative, but don't misconstrue it as the logical play, not unless he KNOWS something about his team or his opponent that makes the standard percentages not apply in the specific scenario. Even so, that would not be a logical decision, it would be based on his own belief for that particular scenario in that particular game.

So the question to be answered is this: Did KF make the logical decision or did he make it based on something else? If he was trying to make the logical decision, he made the wrong call, but if he was basing it on his own belief system, what did he know about his team or opponent that made his decision preferable? That's all any of us want to know, how did he arrive at the decision he made?

 

[PCHawk]

You don't seem to understand the concept of conditional probability. If you are 14 points down, the reason you don't go for two is the much higher percentage play is to kick the PAT and do the same if you score another TD. It DOES NOT help your chances to win the game if you FAIL to convert the 2 PAT, either on the first TD or the second.

Your argument is based on the assumption you convert the 2 PAT, and then I agree with you, you have a better chance of winning the game than had you converted the 1 PAT. But that is not a certain outcome, that is a conditional outcome AFTER you have converted the first 2 PAT.

Logically, the best probability is to kick both PATs. Now if a coach wants to play the riskier strategy, that's certainly his prerogative, but don't misconstrue it as the logical play, not unless he KNOWS something about his team or his opponent that makes the standard percentages not apply in the specific scenario. Even so, that would not be a logical decision, it would be based on his own belief for that particular scenario in that particular game.

So the question to be answered is this: Did KF make the logical decision or did he make it based on something else? If he was trying to make the logical decision, he made the wrong call, but if he was basing it on his own belief system, what did he know about his team or opponent that made his decision preferable? That's all any of want to know, how did he arrive at the decision he made?

If you kick the PAT's, you have a 50% chance to win in ot. If you go for the first 2 pt conversion, you have a just under 50% chance to win in regulation. Its already close to the same odds. Then you add in the chance of making the second one and winning in ot and the math isn't even close. This one is even easier to follow than the one Kirk did.

 

[GoHawks12345]

If you kick the PAT's, you have a 50% chance to win in ot. If you go for the first 2 pt conversion, you have a just under 50% chance to win in regulation. Its already close to the same odds. Then you add in the chance of making the second one and winning in ot and the math isn't even close. This one is even easier to follow than the one Kirk did.

Are we saying a 47% chance to succeed at the 2 pt? 95% for a extra point?


0,0 = 53% * 53% = 28.09% (missing both 2 pt attempts) - LOSS
0,2 = 53% * 47% = 24.91 % (missing first but making 2nd) - TIE
2,1 = 47% * 95% = 44.65% (succeed at 2pt, then kicking extra pt) WIN
2,0 = 47% * 5% = 2.35% (succeed at 2pt, miss 2nd point after) TIE

Your regulation outcomes are all covered:
28.09 + 24.91 + 44.65 + 2.35% = 100%

For the ties, you still have 50/50 chance to win in OT.
24.91% + 2.35% = 27.26% / 2 = 13.63%

So 44.65% chance to win in regulation, plus a 13.63 chance of winning in overtime as a result of the paths above.

44.65 + 13.63 = 58.28

And a reminder, kicking the 2 point after fg, then a 50% chance in OT:
95% * 95% * 50% = 45.125%

So a 58.28 pct chance to a 45.125 pct chance.

 

[hwk23]

You don't seem to understand the concept of conditional probability. If you are 14 points down, the reason you don't go for two is the much higher percentage play is to kick the PAT and do the same if you score another TD. It DOES NOT help your chances to win the game if you FAIL to convert the 2 PAT, either on the first TD or the second. o

Completely different set of facts and not really relevant. There is virtually no difference going for 2 on the 1st or 2nd TD in our case terms of probability. It has nothing to do with KFs belief system.

The only possible argument is intangible differences but if you make that argument you can only be arguing that you have a better chance making the two after the 2nd TD than the 1st. Under either scenario, you have to make the two, so the momentum for the onside kick after making the two or extra point is comparable. Everything comes back to probabilities.

 

[PCHawk]

Are we saying a 47% chance to succeed at the 2 pt? 95% for a extra point?


0,0 = 53% * 53% = 28.09% (missing both 2 pt attempts) - LOSS
0,2 = 53% * 47% = 24.91 % (missing first but making 2nd) - TIE
2,1 = 47% * 95% = 44.65% (succeed at 2pt, then kicking extra pt) WIN
2,0 = 47% * 5% = 2.35% (succeed at 2pt, miss 2nd point after) TIE

Your regulation outcomes are all covered:
28.09 + 24.91 + 44.65 + 2.35% = 100%

For the ties, you still have 50/50 chance to win in OT.
24.91% + 2.35% = 27.26% / 2 = 13.63%

So 44.65% chance to win in regulation, plus a 13.63 chance of winning in overtime as a result of the paths above.

44.65 + 13.63 = 58.28

And a reminder, kicking the 2 point after fg, then a 50% chance in OT:
95% * 95% * 50% = 45.125%

So a 58.28 pct chance to a 45.125 pct chance.

That's a significant advantage to give up for fear of looking stupid because people don't understand.

 

[DJI9424]

If you kick the PAT's, you have a 50% chance to win in ot. If you go for the first 2 pt conversion, you have a just under 50% chance to win in regulation. Its already close to the same odds. Then you add in the chance of making the second one and winning in ot and the math isn't even close. This one is even easier to follow than the one Kirk did.

You are certainly entitled to believe whatever you wish, but you are not allowed to conflate your belief with logic. Again, you're making several conditional assumptions to arrive at your probabilities. Logically, the goal is to score the most points, and this is based on probabilities (expectations), not belief systems.

So, if the expectation of successfully kicking the PAT is 99%, the expectation for points is .99 (99% of 1 point). Similarly, the expectation of converting the 2 PAT is 47% (based on global averages), so the expectation of points is ,94 (47% of 2 points), so logically, the 1 PAT is better.

Now, we're not just talking about 1 TD, we're needing to score 2 TDs. So if we ignore the probabilities of recovering an onside kick and the probability of scoring the second TD, let's calculate the expectation of total points. Also, we need to assume that the first TD and second TD are independent events. Both TDs have to be scored, but we need to see how what was done for the PAT after the first TD affects the end result.

NOTE: The following is based on being 14 points down, not the Iowa/Purdue scenario. Obviously, there is no ability to score fractional points, but that is what is used to determine the highest point expectation, which is what logic is about.

Scenario 1

First 1 PAT successful = total point expectation is 7 + 6.99 = 13.99 (probability 99%)

First 1 PAT fail = total point expectation is 6 + 6 + (.47 * 2) = 12.94 (probability 1%) (MUST go for 2 on second TD in order to tie the game)

So, overall point expectation is (13.99 * .99) + (12.94 * .01) = 13.9795

Scenario 2

First 2 PAT successful = 8 + 6 + (.99 * 1) = 14.99 (probability 47%) (Only need 1 PAT after second TD to win game)

First 2 PAT fail = 6 + 6 + (.47 * 2) = 12.94 (probability 53%)

So, overall point expectation is (14.99 * .47) + (12.94 * .53) = 13.9035

So from a logical perspective, Scenario 1 is preferable before you have made the decision to go for 1 or 2 after the first touchdown. Obviously, if you decide to go for two and are successful, you have increased your point expectation, just as you commented, but before that decision is made, that is not the highest expectation.

Lastly, I would argue (based on my own belief system) that Iowa's offense is not average, it is well below average. Thus the 47% assumption is too high for them, and if that is the case, it just makes it even more logical to kick the PAT after their score in the Purdue game because their likelihood of failure was even greater. And once they failed (as would be expected in most cases), it made it impossible for them to win the game in the time remaining.

 

[PCHawk]

You are certainly entitled to believe whatever you wish, but you are not allowed to conflate your belief with logic. Again, you're making several conditional assumptions to arrive at your probabilities. Logically, the goal is to score the most points, and this is based on probabilities (expectations), not belief systems.

So, if the expectation of successfully kicking the PAT is 99%, the expectation for points is .99 (99% of 1 point). Similarly, the expectation of converting the 2 PAT is 47% (based on global averages), so the expectation of points is ,94 (47% of 2 points), so logically, the 1 PAT is better.

Now, we're not just talking about 1 TD, we're needing to score 2 TDs. So if we ignore the probabilities of recovering an onside kick and the probability of scoring the second TD, let's calculate the expectation of total points. Also, we need to assume that the first TD and second TD are independent events. Both TDs have to be scored, but we need to see how what was done for the PAT after the first TD affects the end result.

NOTE: The following is based on being 14 points down, not the Iowa/Purdue scenario. Obviously, there is no ability to score fractional points, but that is what is used to determine the highest point expectation, which is what logic is about.

Scenario 1

First 1 PAT successful = total point expectation is 7 + 6.99 = 13.99 (probability 99%)

First 1 PAT fail = total point expectation is 6 + 6 + (.47 * 2) = 12.94 (probability 1%) (MUST go for 2 on second TD in order to tie the game)

So, overall point expectation is (13.99 * .99) + (12.94 * .01) = 13.9795

Scenario 2

First 2 PAT successful = 8 + 6 + (.99 * 1) = 14.99 (probability 47%) (Only need 1 PAT after second TD to win game)

First 2 PAT fail = 6 + 6 + (.47 * 2) = 12.94 (probability 53%)

So, overall point expectation is (14.99 * .47) + (12.94 * .53) = 13.9035

So from a logical perspective, Scenario 1 is preferable before you have made the decision to go for 1 or 2 after the first touchdown. Obviously, if you decide to go for two and are successful, you have increased your point expectation, just as you commented, but before that decision is made, that is not the highest expectation.

Lastly, I would argue (based on my own belief system) that Iowa's offense is not average, it is well below average. Thus the 47% assumption is too high for them, and if that is the case, it just makes it even more logical to kick the PAT after their score in the Purdue game because their likelihood of failure was even greater. And once they failed (as would be expected in most cases), it made it impossible for them to win the game in the time remaining.

If I was Iowa's offense, I wouldn't dream of going for 2. How are you not following the math go hawks did. Your math makes sense if you want the best odds of tying the game. Not winning the game.

 

[DJI9424]

If I was Iowa's offense, I wouldn't dream of going for 2. How are you not following the math go hawks did. Your math makes sense if you want the best odds of tying the game. Not winning the game.

I had not read Gohawks' post before I posted mine. Now that I have read it, I will comment. I agree on his calculations regarding the probabilities of winning or losing in regulation (although I believe the % for 1 PAT is higher than 95% and the % for 2 PAT is lower, especially for Iowa), but not on adding the probability of winning in regulation with the probability of winning in overtime (after the game has ended in a tie) to arrive at the overall probability of winning.

Given the game goes to overtime, the chance of winning is stipulated to be 50%, however, the probability of winning in regulation (by going for 2 PAT) is only 44.65%. But if you go for 1 PAT after both touchdowns, the probability of a tie is (.95 * .95) = 90.25% (given that 2 TDs were scored) and when combined with the probability of winning in overtime, the compound probability is 50% of 90.25% or 45.125%, which is higher than the 2 PAT winning strategy. Therefore, your best strategy to winning the game is to take the game into overtime, if you truly believe your chances of winning in overtime are 50/50.

Again, this has a number of conditional probabilities embedded in the assumptions, but given that those assumptions are accepted, the better approach is to go for the tie , based on the overall expectations. Finally, that result makes sense from the perspective that winning in regulation is calculated (based on assumptions) to be below 50% but winning in overtime is stipulated to be 50%. Thus, you improve your overall chances of winning by extending the game into overtime. (Of course, all based on your assumptions).

 

[hwk23]

Are we saying a 47% chance to succeed at the 2 pt? 95% for a extra point?


0,0 = 53% * 53% = 28.09% (missing both 2 pt attempts) - LOSS
0,2 = 53% * 47% = 24.91 % (missing first but making 2nd) - TIE
2,1 = 47% * 95% = 44.65% (succeed at 2pt, then kicking extra pt) WIN
2,0 = 47% * 5% = 2.35% (succeed at 2pt, miss 2nd point after) TIE

Your regulation outcomes are all covered:
28.09 + 24.91 + 44.65 + 2.35% = 100%

For the ties, you still have 50/50 chance to win in OT.
24.91% + 2.35% = 27.26% / 2 = 13.63%

So 44.65% chance to win in regulation, plus a 13.63 chance of winning in overtime as a result of the paths above.

44.65 + 13.63 = 58.28

And a reminder, kicking the 2 point after fg, then a 50% chance in OT:
95% * 95% * 50% = 45.125%

So a 58.28 pct chance to a 45.125 pct chance.

Good post. The 47% might be generous, especially for Iowa, but I think this shows the obvious answer is not always correct.

 

[hwk23]

I had not read Gohawks' post before I posted mine. Now that I have read it, I will comment. I agree on his calculations regarding the probabilities of winning or losing in regulation (although I believe the % for 1 PAT is higher than 95% and the % for 2 PAT is lower, especially for Iowa), but not on adding the probability of winning in regulation with the probability of winning in overtime (after the game has ended in a tie) to arrive at the overall probability of winning.

Given the game goes to overtime, the chance of winning is stipulated to be 50%, however, the probability of winning in regulation (by going for 2 PAT) is only 44.65%. But if you go for 1 PAT after both touchdowns, the probability of a tie is (.95 * .95) = 90.25% (given that 2 TDs were scored) and when combined with the probability of winning in overtime, the compound probability is 50% of 90.25% or 45.125%, which is higher than the 2 PAT winning strategy. Therefore, your best strategy to winning the game is to take the game into overtime, if you truly believe your chances of winning in overtime are 50/50.

Again, this has a number of conditional probabilities embedded in the assumptions, but given that those assumptions are accepted, the better approach is to go for the tie , based on the overall expectations. Finally, that result makes sense from the perspective that winning in regulation is calculated (based on assumptions) to be below 50% but winning in overtime is stipulated to be 50%. Thus, you improve your overall chances of winning by extending the game into overtime. (Of course, all based on your assumptions).

No, you're forgetting the scenarios where you tie when going for two and win in overtime. It's all in Gohawks post.

 

[PCHawk]

No, you're forgetting the scenarios where you tie when going for two and win in overtime. It's all in Gohawks post.

That's right. There are 2 ways to win. One is to make the first one and win in regulation. The other is to miss the first one, make the second one, then win in overtime.

There are 2 ways to lose. Miss both and lose in regulation. Or miss the first, make the second, and lose in ot.

The odds of one of the two ways of winning happening are quite a bit better than the odds of one of the two ways of loosing happening.

 

[PCHawk]

I will add that the chances of Iowa making a 2 point conversion are about 5%. So if I was Kirk and down 14, I wouldn't try it.

 

[SamBrownlee48]

Except, for the 39th time, all things weren’t equal. Knowing you need two possessions adds ZERO value, when you only have time for one. I’m still waiting for one example of a coach that has done the same. So far I got “chip Kelly probably would even though I’ve never actually watched his games.”

What things were not equal?

And you are incorrect. While knowing you need two scores with 1 minute left is a pretty hopeless situation, it is not as hopeless as knowing you needed two scores with 0 minutes left.

With a minute left, you now know you are going to need to convert 2 onside kicks, get a TD and a field goal. This gives you a chance to take the FG really quick if the opportunity presents itself. It also allows you to know that you can't use the whole minute in the next drive if you convert onside kick.

If you don't know that you are going to need two possessions until after the 2nd TD, you won't have the information you will need to decide to take the FG right away, and you won't worry as much about using the whole minute to get the TD.

So there is value in taking the 2pt right away. The only value in not doing so is really to the fans, who get to keep their hopes up a couple minutes longer before the 2pt conversion is missed.