Ravens @ Patriots - AFC Championship Game (1 Viewer)

[TobiasFunke]

I had no problem with it. The Ravens have been very poor in short yardage all year - last week they were stuffed at the goal line by the Texans, and this week Wilfork and the Pats shoved the Ravens around up front.Harbaugh's coaching had them in position to win, or at worst tie, the game in the final seconds. What about Bellichick's decision to kick the FG on the opening drive of the second half on 4th-and-1 from the Ravens 6?

Also a bad decision, although not quite as bad since they were a little further away from the goal line if the converted and I believe they also had further to go.Do you have any data on the Ravens conversion rates this year, and the Patriots? Absent that, the standard conversion rates for the NFL say you should go for it in that situation and it's not even close. Short yardage conversion rate NFL-wide is about 68% I believe. Assuming that, the EV of going for it is obviously gonna be higher than kicking the FG. And that's before you even get into the psychological victory you hand the other team when you wimp out.The team, with Harbaugh's coaching, was in a position to tie in the final seconds. But Harbaugh and Harbaugh alone made a decision that may have cost them four points and the AFC championship, any way you slice it.

 

[Modog814]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

 

[humpback]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

Here we go again.You could argue that one quarter's worth of data is more applicable than all of your "huge amounts of evidence" combined. This incorrect decision from a statistical standpoint is your opinion based on estimates, not a mathematical fact. For the record, I think I would have gone for it as well, but it is a subjective decision and certainly defensible.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

Here we go again.You could argue that one quarter's worth of data is more applicable than all of your "huge amounts of evidence" combined. This incorrect decision from a statistical standpoint is your opinion based on estimates, not a mathematical fact. For the record, I think I would have gone for it as well, but it is a subjective decision and certainly defensible.

Turns out there actually were some media folks who brought this up. LinkIf it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

 

[Ghost Rider]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I completely disagree. Flacco is not good at QB sneaks, and the Patriots were stuffing the middle of the field at that point in regards to the run. Taking the 3 was the smart move. After getting the INT of Brady and then hitting the big play to Torrey Smith, to come out of that with nothing would have been demoralizing to the offense and the team.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I completely disagree. Flacco is not good at QB sneaks, and the Patriots were stuffing the middle of the field at that point in regards to the run. Taking the 3 was the smart move. After getting the INT of Brady and then hitting the big play to Torrey Smith, to come out of that with nothing would have been demoralizing to the offense and the team.

I see. That must be why the Ravens converted their only short yardage attempt of the game.Like I said, I'd be open to psychobabble arguments if it was close call, but it wasn't.

 

[Ghost Rider]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I completely disagree. Flacco is not good at QB sneaks, and the Patriots were stuffing the middle of the field at that point in regards to the run. Taking the 3 was the smart move. After getting the INT of Brady and then hitting the big play to Torrey Smith, to come out of that with nothing would have been demoralizing to the offense and the team.

I see. That must be why the Ravens converted their only short yardage attempt of the game.Like I said, I'd be open to psychobabble arguments if it was close call, but it wasn't.

Oh. So that means that you will convert every short yardage attempt because you converted one? And IIRC the short yardage attempt they converted was late in the game, when the offense was rolling and had tons of momentum. Early on, after several 3 and outs, the Ravens offense had finally gotten going after a Brady INT and, again, to come out of that with no points would have been brutal. It's a long game, and it was still early. You take the points. Had Evans caught that pass at the end, would we even be having this discussion?

 

[humpback]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

Here we go again.You could argue that one quarter's worth of data is more applicable than all of your "huge amounts of evidence" combined. This incorrect decision from a statistical standpoint is your opinion based on estimates, not a mathematical fact. For the record, I think I would have gone for it as well, but it is a subjective decision and certainly defensible.

Turns out there actually were some media folks who brought this up. LinkIf it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

Here we go again.You could argue that one quarter's worth of data is more applicable than all of your "huge amounts of evidence" combined. This incorrect decision from a statistical standpoint is your opinion based on estimates, not a mathematical fact. For the record, I think I would have gone for it as well, but it is a subjective decision and certainly defensible.

Turns out there actually were some media folks who brought this up. LinkIf it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

It's in the link. EV of going for it is 4.1, plus if you fail the opposition has the ball on the 4 yard line instead of wherever they'd end up with the ball after your field goal attempt. No-brainer.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I completely disagree. Flacco is not good at QB sneaks, and the Patriots were stuffing the middle of the field at that point in regards to the run. Taking the 3 was the smart move. After getting the INT of Brady and then hitting the big play to Torrey Smith, to come out of that with nothing would have been demoralizing to the offense and the team.

I see. That must be why the Ravens converted their only short yardage attempt of the game.Like I said, I'd be open to psychobabble arguments if it was close call, but it wasn't.

Oh. So that means that you will convert every short yardage attempt because you converted one? And IIRC the short yardage attempt they converted was late in the game, when the offense was rolling and had tons of momentum. Early on, after several 3 and outs, the Ravens offense had finally gotten going after a Brady INT and, again, to come out of that with no points would have been brutal. It's a long game, and it was still early. You take the points. Had Evans caught that pass at the end, would we even be having this discussion?

No, it simply counters your argument that the Pats were stuffing the middle of the field. I was countering your team-specific argument.No, we wouldn't be having this discussion if Evans had caught that ball (although I had it during the game with my friends). We also wouldn't be having a discussion about Cundiff's missed FG. You think people should stop discussing that mistake based on the fact that we wouldn't be having that conversation if Evans had caught the pass?

 

[humpback]

If it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

It's in the link. EV of going for it is 4.1, plus if you fail the opposition has the ball on the 4 yard line instead of wherever they'd end up with the ball after your field goal attempt. No-brainer.

I'm not seeing the EV calculation, just the "results", which it says used "history" to calculate. Right there is the first major subjective component which makes it absolutely not a mathematical fact.

 

[TobiasFunke]

If it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

It's in the link. EV of going for it is 4.1, plus if you fail the opposition has the ball on the 4 yard line instead of wherever they'd end up with the ball after your field goal attempt. No-brainer.

I'm not seeing the EV calculation, just the "results", which it says used "history" to calculate. Right there is the first major subjective component which makes it absolutely not a mathematical fact.

"History" = data from all NFL games. I don't see how that's subjective.If the subjectivity you're referring to is that the Ravens and Patriots are somehow so far from typical that the data isn't useful, I would say that (1) you're wrong, since the Ravens were 14th in the league in yards per rush attempt and the Pats were 8th in yards per attempt against, which is more than close enough to average/even for the data to be relevant; and (2) you're wrong because the calculation isn't even close. 4.1 vs. 2.4, plus the yardage thing. It's a HUGE difference. As the article says, even if you assume the FG is a sure thing and the chance of success of a conversion attempt are barely above 50% instead of 69%, you should still go for it from an EV perspective.

If you click through the link the Grantland article it takes you to advancednflstats.com, where you can run the numbers yourself.

 

[humpback]

If it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

It's in the link. EV of going for it is 4.1, plus if you fail the opposition has the ball on the 4 yard line instead of wherever they'd end up with the ball after your field goal attempt. No-brainer.

I'm not seeing the EV calculation, just the "results", which it says used "history" to calculate. Right there is the first major subjective component which makes it absolutely not a mathematical fact.

"History" = data from all NFL games. I don't see how that's subjective.If the subjectivity you're referring to is that the Ravens and Patriots are somehow so far from typical that the data isn't useful, I would say that (1) you're wrong, since the Ravens were 14th in the league in yards per rush attempt and the Pats were 8th in yards per attempt against, which is more than close enough to average/even for the data to be relevant; and (2) you're wrong because the calculation isn't even close. 4.1 vs. 2.4, plus the yardage thing. It's a HUGE difference. As the article says, even if you assume the FG is a sure thing and the chance of success of a conversion attempt are barely above 50% instead of 69%, you should still go for it from an EV perspective.

Using the data from all NFL games is a subjective input. I have no idea how far back they are going to get the data, but why would the odds of that specific attempt, under those unique set of circumstances, be affected by what 2 entirely different teams, under entirely different circumstances, did? Why wouldn't it more heavily weight the odds of just the Ravens and Patriots? More recent attempts by those two teams? More recent attempts by all teams under much more similar circumstances? Etc, etc, etc.That EV calculation is only estimate (as they all are). I have no problem with thinking they should have gone for it, but you're just plain wrong to say it's a mathematical fact. There are far too many subjective criteria and values to say that.

 

[TobiasFunke]

If it were close, you could say that it was an opinion based on estimates and not a mathematical fact. But given the numbers there, it was absolutely a mathematical fact. Even if you assume only a 50% chance of success on the 4th down conversion, which is impossibly low, it still would have been the right move.

When a decision is close based on the average NFL team, you can say it's subjective and depends on the teams involved and psychological factors and all that mumbo-jumbo. This decision was not remotely close.

Let's see the math.

It's in the link. EV of going for it is 4.1, plus if you fail the opposition has the ball on the 4 yard line instead of wherever they'd end up with the ball after your field goal attempt. No-brainer.

I'm not seeing the EV calculation, just the "results", which it says used "history" to calculate. Right there is the first major subjective component which makes it absolutely not a mathematical fact.

"History" = data from all NFL games. I don't see how that's subjective.If the subjectivity you're referring to is that the Ravens and Patriots are somehow so far from typical that the data isn't useful, I would say that (1) you're wrong, since the Ravens were 14th in the league in yards per rush attempt and the Pats were 8th in yards per attempt against, which is more than close enough to average/even for the data to be relevant; and (2) you're wrong because the calculation isn't even close. 4.1 vs. 2.4, plus the yardage thing. It's a HUGE difference. As the article says, even if you assume the FG is a sure thing and the chance of success of a conversion attempt are barely above 50% instead of 69%, you should still go for it from an EV perspective.

Using the data from all NFL games is a subjective input. I have no idea how far back they are going to get the data, but why would the odds of that specific attempt, under those unique set of circumstances, be affected by what 2 entirely different teams, under entirely different circumstances, did? Why wouldn't it more heavily weight the odds of just the Ravens and Patriots? More recent attempts by those two teams? More recent attempts by all teams under much more similar circumstances? Etc, etc, etc.That EV calculation is only estimate (as they all are). I have no problem with thinking they should have gone for it, but you're just plain wrong to say it's a mathematical fact. There are far too many subjective criteria and values to say that.

Maybe "mathematical fact" was incorrect phrasing, but the numbers are so skewed that it's impossible to imagine factors like the teams involved (especially since, as I've shown, Baltimore rushing O and NE rushing D were pretty even) would change the math from the data for "average" teams enough for it to matter.

Consider this: if you score a TD to tie the game with thirty seconds left you should kick the extra point to take the lead instead of trying for a two point conversion. We all know that. Sure, it's possible that you have the world's worst kicker and the world's best red zone offense and your opponent employs Dwight Howard and Yao Ming for the sole purpose of blocking FG attempts, and that those factors would change the calculation for you. But the difference is so stark that it's almost impossible to imagine circumstances that make going for two the correct play.

This obviously isn't that extreme, but the point is the same. The difference between going for it and trying the FG for the average NFL team playing the average NFL team is SO huge that it's almost impossible to imagine the personnel making a big enough difference to change the outcome.

 

[Modog814]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

No you're not disregarding huge amounts of evidence. A couple things are wrong with the "huge amounts of evidence"...1) It's been compiled through many different teams playing various opponents in various situations. It probably has very little relevance to those 2 teams in that exact situation. You don't get to play the game on paper. 2) You're making assumptions that may or may not be valid that this particular situation to be a positive EV play. What if he made the judgement that he thought he had a 65% chance of making that 4th down. And then a 65% chance of scoring a TD. And he determined that since he went for it on 4th and 1 there, he was most likely going to go for it on 4th down again if they didn't get it. So his EV is now negative. So maybe he did make the right decision.

3) And Most importantly, even if it were all 100% accurate, the positive EV applies to a scenario that could get played over and over and over again. John Harbaugh doesn't get that opportunity. He gets one shot and one shot only. For whatever reason he determined the right play was to kick it there.

Why do you think it was the wrong call based on the ultimate result? You're putting him in a no win situation. If he goes for it and doesn't get it and they still lose, then he made the wrong call based on your reasoning. If they don't score there, and Brady marches down the field and puts up a TD, Baltimore would be in a world of hurt down 10-0.

 

[TobiasFunke]

I haven't heard anyone criticize Harbaugh's horrendous decision to kick the field goal at fourth and 1 for the three yard line early in the second quarter.

Just an indefensible decision. There is no argument whatsoever in that situation, when a conversion gives you a first and goal within the five against a relatively weak defense. You go for it without hesitation. There's a pretty good chance that the call cost his team four points and a place in the Super Bowl.

I didn't have a problem with it either. It was a long yard, not just a couple inches. If it were the latter, I believe Harbaugh would have gone for it. But NE was doing a great job of stopping the run to that point. Personally, I would have gone for it because 1) their Defense was playing pretty well and at worst you'd make them have to drive a long way to get points, 2) It's mighty tough to beat an offensive team like the Pats by settling for FG's. Also if they were down 7-0 I think going for the FG is wrong. But at 3-0, I don't have a problem playing conservatively there to take the tie.

So you would disregard huge amounts of evidence suggesting that the obvious play from an EV perspective is to go for it based on one quarter's worth of data? I think it's crazy how this is being overlooked. If he'd gone for it and failed and his team ended up just short it would have been a huge part of the postgame story. Why do we nail coaches to the wall when they make the supposedly "risky" play and it turns out to be the wrong call based on the ultimate result, but completely ignore it when they make the supposedly "safe" play and it turns out to be the wrong call based on the ultimate result?

Harbaugh's a great coach and I wish he was the coach of my team, but make no mistake about it- he made the wrong decision from a statistical standpoint, and it just might have cost Baltimore a place in the Super Bowl.

No you're not disregarding huge amounts of evidence. A couple things are wrong with the "huge amounts of evidence"...1) It's been compiled through many different teams playing various opponents in various situations. It probably has very little relevance to those 2 teams in that exact situation. You don't get to play the game on paper. 2) You're making assumptions that may or may not be valid that this particular situation to be a positive EV play. What if he made the judgement that he thought he had a 65% chance of making that 4th down. And then a 65% chance of scoring a TD. And he determined that since he went for it on 4th and 1 there, he was most likely going to go for it on 4th down again if they didn't get it. So his EV is now negative. So maybe he did make the right decision.

3) And Most importantly, even if it were all 100% accurate, the positive EV applies to a scenario that could get played over and over and over again. John Harbaugh doesn't get that opportunity. He gets one shot and one shot only. For whatever reason he determined the right play was to kick it there.

Why do you think it was the wrong call based on the ultimate result? You're putting him in a no win situation. If he goes for it and doesn't get it and they still lose, then he made the wrong call based on your reasoning. If they don't score there, and Brady marches down the field and puts up a TD, Baltimore would be in a world of hurt down 10-0.

You're basically just deferring to the coach and saying "trust his judgment." The logic you give here could be used to defend any coaching decision ever made.No offense, but I don't think you're really following the argument at all. The bolded is one of the things that tells me that. If he went for it and failed, it would still be the right call. A call is right or wrong when it's made, regardless of the outcome. Nothing is guaranteed to succeed or fail, best you can do is play the percentages. Harbough played them wrongly, and clearly so.

 

[humpback]

Maybe "mathematical fact" was incorrect phrasing, but the numbers are so skewed that it's impossible to imagine factors like the teams involved (especially since, as I've shown, Baltimore rushing O and NE rushing D were pretty even) would change the math from the data for "average" teams enough for it to matter. Consider this: if you score a TD to tie the game with thirty seconds left you should kick the extra point to take the lead instead of trying for a two point conversion. We all know that. Sure, it's possible that you have the world's worst kicker and the world's best red zone offense and your opponent employs Dwight Howard and Yao Ming for the sole purpose of blocking FG attempts, and that those factors would change the calculation for you. But the difference is so stark that it's almost impossible to imagine circumstances that make going for two the correct play.This obviously isn't that extreme, but the point is the same. The difference between going for it and trying the FG for the average NFL team playing the average NFL team is SO huge that it's almost impossible to imagine the personnel making a big enough difference to change the outcome.

This isn't the average NFL team playing the average NFL team under average NFL conditions. This is a very specific, unique set of circumstances that you just can't accurately assign a probability to. This exact situation has never happened before and will never happen again, and it's either going to be a success or a failure- it can't be 68% (or whatever probability you come up with) successful.Again, I don't even know what they are using in their calculation of EV- I clicked on the link, can input numbers, but all I get is the output, not the formula. Are the odds of winning and losing under those scenarios based on the results from all games where teams had the same lead with the same amount of time left? Do they consider who got the ball to start the game? Do they factor score or just lead/deficit? Do they factor home/away? Do they factor in how the outcome of that attempt can change the rest of the game (playcalling)? I know they aren't factoring in specific teams, coaches, players, etc.I have no idea how much data is being used or how relevant it is. I will say that I would certainly factor in more recent data more heavily, more team specific data much more heavily, more situation specific data much more heavily (4th and 1 from the 3 early in the 2nd qtr down 3-0 in a championship game is very different from 4th and 1 from the 45 in the 4th qtr. down 28-7 in a meaningless game), etc. Even then, I really don't think past data is all that relevant to this particular attempt, it's all just an unknowable estimate. It's all so subjective, there isn't one correct EV.

 

[TobiasFunke]

Maybe "mathematical fact" was incorrect phrasing, but the numbers are so skewed that it's impossible to imagine factors like the teams involved (especially since, as I've shown, Baltimore rushing O and NE rushing D were pretty even) would change the math from the data for "average" teams enough for it to matter. Consider this: if you score a TD to tie the game with thirty seconds left you should kick the extra point to take the lead instead of trying for a two point conversion. We all know that. Sure, it's possible that you have the world's worst kicker and the world's best red zone offense and your opponent employs Dwight Howard and Yao Ming for the sole purpose of blocking FG attempts, and that those factors would change the calculation for you. But the difference is so stark that it's almost impossible to imagine circumstances that make going for two the correct play.This obviously isn't that extreme, but the point is the same. The difference between going for it and trying the FG for the average NFL team playing the average NFL team is SO huge that it's almost impossible to imagine the personnel making a big enough difference to change the outcome.

This isn't the average NFL team playing the average NFL team under average NFL conditions. This is a very specific, unique set of circumstances that you just can't accurately assign a probability to. This exact situation has never happened before and will never happen again, and it's either going to be a success or a failure- it can't be 68% (or whatever probability you come up with) successful.Again, I don't even know what they are using in their calculation of EV- I clicked on the link, can input numbers, but all I get is the output, not the formula. Are the odds of winning and losing under those scenarios based on the results from all games where teams had the same lead with the same amount of time left? Do they consider who got the ball to start the game? Do they factor score or just lead/deficit? Do they factor home/away? Do they factor in how the outcome of that attempt can change the rest of the game (playcalling)? I know they aren't factoring in specific teams, coaches, players, etc.I have no idea how much data is being used or how relevant it is. I will say that I would certainly factor in more recent data more heavily, more team specific data much more heavily, more situation specific data much more heavily (4th and 1 from the 3 early in the 2nd qtr down 3-0 in a championship game is very different from 4th and 1 from the 45 in the 4th qtr. down 28-7 in a meaningless game), etc. Even then, I really don't think past data is all that relevant to this particular attempt, it's all just an unknowable estimate. It's all so subjective, there isn't one correct EV.

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

 

[Modog814]

You're basically just deferring to the coach and saying "trust his judgment." The logic you give here could be used to defend any coaching decision ever made.No offense, but I don't think you're really following the argument at all. The bolded is one of the things that tells me that. If he went for it and failed, it would still be the right call. A call is right or wrong when it's made, regardless of the outcome. Nothing is guaranteed to succeed or fail, best you can do is play the percentages. Harbough played them wrongly, and clearly so.

I'm following the argument. You're coming up with percentages based on situations that may or may not be related to the situation they were in. NE was stopping the run in that first half, and Harbaugh apparently felt that lowered his chances of not only converting the 4th down, but perhaps lowered his chances of scoring a TD even if they got the first down. The fact that historically a random team vs another random team converts 68% in short yardage situation, imo, doesn't necessarily apply. We don't have one random team vs. another random team. We have one team who had up to that point struggled to run the ball against another team that seemed to be selling out to stop the run up the middle. Wilfork was playing great and Harbaugh didn't believe the risk was worth the reward in that situation. What you don't seem to get is that these percentages, even if accurate, apply to the long-run, and over thousands of attempts it may turn out in their favor. But they don't get to play those thousands of attempts. They get 1 attempt. Best example I can give you, we flip a coin, if it's head I'll give you $5 Million dollars, if it's tails, you give me $4 Million, one flip only. Would you take that deal? Mathematically it's a no-brainer. But putting $4 Million on the line to win $5 Million is a pretty risky thing, and to most, unless they have $4 Million burning a hole in their pocket, it's not worth the risk because you only get one flip. If this was the regular season, I agree with you 100%, you take the chance and go for it. But it's not, there may not be a next week for these things to balance out. Trust me, I'm a math guy, I do it for a living, and I get what you're saying mathematically. But you're not considering another side of the equation that is purely judgemental and can't be put into mathematical terms.

 

[Modog814]

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.

As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

See this is the problem. These aren't average teams (you have arguable the 2 best teams in the AFC), this is not an average situation (this is playoff game, win or go home, not really an average situation) and you don't get the average result (you can't score 4.1 points. It's 0, 3 or 7).

 

[humpback]

Maybe "mathematical fact" was incorrect phrasing, but the numbers are so skewed that it's impossible to imagine factors like the teams involved (especially since, as I've shown, Baltimore rushing O and NE rushing D were pretty even) would change the math from the data for "average" teams enough for it to matter. Consider this: if you score a TD to tie the game with thirty seconds left you should kick the extra point to take the lead instead of trying for a two point conversion. We all know that. Sure, it's possible that you have the world's worst kicker and the world's best red zone offense and your opponent employs Dwight Howard and Yao Ming for the sole purpose of blocking FG attempts, and that those factors would change the calculation for you. But the difference is so stark that it's almost impossible to imagine circumstances that make going for two the correct play.This obviously isn't that extreme, but the point is the same. The difference between going for it and trying the FG for the average NFL team playing the average NFL team is SO huge that it's almost impossible to imagine the personnel making a big enough difference to change the outcome.

This isn't the average NFL team playing the average NFL team under average NFL conditions. This is a very specific, unique set of circumstances that you just can't accurately assign a probability to. This exact situation has never happened before and will never happen again, and it's either going to be a success or a failure- it can't be 68% (or whatever probability you come up with) successful.Again, I don't even know what they are using in their calculation of EV- I clicked on the link, can input numbers, but all I get is the output, not the formula. Are the odds of winning and losing under those scenarios based on the results from all games where teams had the same lead with the same amount of time left? Do they consider who got the ball to start the game? Do they factor score or just lead/deficit? Do they factor home/away? Do they factor in how the outcome of that attempt can change the rest of the game (playcalling)? I know they aren't factoring in specific teams, coaches, players, etc.I have no idea how much data is being used or how relevant it is. I will say that I would certainly factor in more recent data more heavily, more team specific data much more heavily, more situation specific data much more heavily (4th and 1 from the 3 early in the 2nd qtr down 3-0 in a championship game is very different from 4th and 1 from the 45 in the 4th qtr. down 28-7 in a meaningless game), etc. Even then, I really don't think past data is all that relevant to this particular attempt, it's all just an unknowable estimate. It's all so subjective, there isn't one correct EV.

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

And as I've said several times now, that is purely your opinion. I don't agree with your opinion. In my opinion, most of that data should be basically irrelevant because most of those situations were very different from that one. Do you really think what happened between the Bucs and the Rams in a 34-7 game with 1 minute to go in the 4th qtr in a meaningless game in 2003 has any bearing on the Ravens chances of converting on that one particular play this weekend? Again, I don't even know what data they used or how the used it, but I certainly would not weight every 4th and 1 the same if I were to come up with an estimate. Even then, it would be a very crude estimate that we have no way of knowing how accurate it was, even in hindsight.There isn't one "correct" probability here, and there isn't one "correct" EV. Just like the "go for two" debate, this can not be proven by math.

 

[TobiasFunke]

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.

As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

See this is the problem. These aren't average teams (you have arguable the 2 best teams in the AFC), this is not an average situation (this is playoff game, win or go home, not really an average situation) and you don't get the average result (you can't score 4.1 points. It's 0, 3 or 7).

The teams weren't on the field. The Baltimore offense and the Patriots defense were on the field, and those units are close enough to average to make the data relevant.I don't understand what you're saying about not getting the average result. The EV isn't meant to predict exactly how many points you'll get, it's an average expected number of points. When the average expected points of Option A for an average team is WAY more than the average expected points for Option B for an average team, you need to show me something that tells me these teams are really really different from average for you to convince me that Option B should even be considered. I don't see that here.

And I think you're underestimating just how much variance from the norm you'd have to show. As the article said- even if you assume only a 50% conversion rate from going for it, which is impossibly low, going for it still has a higher EV.

 

[Modog814]

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.

As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

See this is the problem. These aren't average teams (you have arguable the 2 best teams in the AFC), this is not an average situation (this is playoff game, win or go home, not really an average situation) and you don't get the average result (you can't score 4.1 points. It's 0, 3 or 7).

The teams weren't on the field. The Baltimore offense and the Patriots defense were on the field, and those units are close enough to average to make the data relevant.I don't understand what you're saying about not getting the average result. The EV isn't meant to predict exactly how many points you'll get, it's an average expected number of points. When the average expected points of Option A for an average team is WAY more than the average expected points for Option B for an average team, you need to show me something that tells me these teams are really really different from average for you to convince me that Option B should even be considered. I don't see that here.

And I think you're underestimating just how much variance from the norm you'd have to show. As the article said- even if you assume only a 50% conversion rate from going for it, which is impossibly low, going for it still has a higher EV.

They are average compared to what? The Pats have an average passing defense? An average rushing defense? An average pass rush? An average ability to stop opponents in short yardage situation? Average compared to this years teams? Average compared to historical teams? Same goes for Baltimore offense. There are just too many variables in play to label a team average simply because they rank mid pack in points score or points allowed (or some other metric) in a particular season. And don't get me wrong, I'm not saying that they aren't average...they very well might be, but I think in this case its a very over simplifying assumption.I don't need to show you that they are significantly different than average. The average points refers to playing out the scenario a large number of times. That isn't the case, they get one shot. In a single one trial situation it's not always about what has the higher EV, it's about giving yourself the best chance to win that game. Harbaugh felt in that situation coming away with 3 points and a tie game gave them a better chance of winning than risking it and coming away with 0. Was it the correct decision? I don't know, like I said, I probably would have gone for it, but it was a long yard not just inches and the run game wasn't going well and the pats run def was playing great. He felt like he didn't have to push his chips into the middle at that particular point in time, and it looked like he was right because he put the Ravens in a situation to win/tie the game as the end.

I'd need to see what the formula is that they are using to get the 4.1 points. But it looks like it's roughly saying 68% chance of converting and a 86% chance of scoring a TD after converting. That seems a touch high to me considering Baltimore ranked 18th in Red Zone TD scoring. But I'm not necessarily saying that the math is incorrect. What I'm saying is that even with the EV+ it's a gamble. And if it were a best of 7 series, I'd say yes, you go for it. But it's not, it's one game, that you MUST win. You can't gamble every play just because it's EV+. I'm guessing that passing the ball on 1st and 10 probably has a higher EV than running the ball, does that mean you should always throw the ball on first down?

 

[Modog814]

The teams weren't on the field. The Baltimore offense and the Patriots defense were on the field, and those units are close enough to average to make the data relevant.I don't understand what you're saying about not getting the average result. The EV isn't meant to predict exactly how many points you'll get, it's an average expected number of points. When the average expected points of Option A for an average team is WAY more than the average expected points for Option B for an average team, you need to show me something that tells me these teams are really really different from average for you to convince me that Option B should even be considered. I don't see that here. And I think you're underestimating just how much variance from the norm you'd have to show. As the article said- even if you assume only a 50% conversion rate from going for it, which is impossibly low, going for it still has a higher EV.

Also, just as an aside, through week 14 this year (that's all the data I have right now) there were 26 instances of teams going for it on 4th and 1 inside their opponents 10 yard line. 17 times they failed to convert.

 

[TobiasFunke]

The data that's being used is every NFL play. I don't know from how many years, but it's a lot of data. That means all the factors of teams involved, situation, etc. balance out so you get the average result for the average team against the average team in the average situation.

As I've said several times now, your argument would have some merit if these two teams were wildly different from the league average and the league average data showed that it was a close call. But the relevant units were pretty close to average, and the league average data showed it was nowhere near a close call.

See this is the problem. These aren't average teams (you have arguable the 2 best teams in the AFC), this is not an average situation (this is playoff game, win or go home, not really an average situation) and you don't get the average result (you can't score 4.1 points. It's 0, 3 or 7).

The teams weren't on the field. The Baltimore offense and the Patriots defense were on the field, and those units are close enough to average to make the data relevant.I don't understand what you're saying about not getting the average result. The EV isn't meant to predict exactly how many points you'll get, it's an average expected number of points. When the average expected points of Option A for an average team is WAY more than the average expected points for Option B for an average team, you need to show me something that tells me these teams are really really different from average for you to convince me that Option B should even be considered. I don't see that here.

And I think you're underestimating just how much variance from the norm you'd have to show. As the article said- even if you assume only a 50% conversion rate from going for it, which is impossibly low, going for it still has a higher EV.

They are average compared to what? The Pats have an average passing defense? An average rushing defense? An average pass rush? An average ability to stop opponents in short yardage situation? Average compared to this years teams? Average compared to historical teams? Same goes for Baltimore offense. There are just too many variables in play to label a team average simply because they rank mid pack in points score or points allowed (or some other metric) in a particular season. And don't get me wrong, I'm not saying that they aren't average...they very well might be, but I think in this case its a very over simplifying assumption.I don't need to show you that they are significantly different than average. The average points refers to playing out the scenario a large number of times. That isn't the case, they get one shot. In a single one trial situation it's not always about what has the higher EV, it's about giving yourself the best chance to win that game. Harbaugh felt in that situation coming away with 3 points and a tie game gave them a better chance of winning than risking it and coming away with 0. Was it the correct decision? I don't know, like I said, I probably would have gone for it, but it was a long yard not just inches and the run game wasn't going well and the pats run def was playing great. He felt like he didn't have to push his chips into the middle at that particular point in time, and it looked like he was right because he put the Ravens in a situation to win/tie the game as the end.

I'd need to see what the formula is that they are using to get the 4.1 points. But it looks like it's roughly saying 68% chance of converting and a 86% chance of scoring a TD after converting. That seems a touch high to me considering Baltimore ranked 18th in Red Zone TD scoring. But I'm not necessarily saying that the math is incorrect. What I'm saying is that even with the EV+ it's a gamble. And if it were a best of 7 series, I'd say yes, you go for it. But it's not, it's one game, that you MUST win. You can't gamble every play just because it's EV+. I'm guessing that passing the ball on 1st and 10 probably has a higher EV than running the ball, does that mean you should always throw the ball on first down?

We're never gonna agree here, so let's just say we see it differently.

That said, I agree with a lot of what you say here about the variables, but couldn't possibly disagree more with the notion that your decisionmaking process should be different for a one time event as opposed to if you're playing seven games or 16 games or a million games. If it's EV+, it increases your chances of winning, so you do it. End of story. Only exception would be a scoreboard-influenced decision late in the game. But early in the second quarter, it's a no-brainer (assuming it's EV+, obviously).

 

[Modog814]

We're never gonna agree here, so let's just say we see it differently.

That said, I agree with a lot of what you say here about the variables, but couldn't possibly disagree more with the notion that your decisionmaking process should be different for a one time event as opposed to if you're playing seven games or 16 games or a million games. If it's EV+, it increases your chances of winning, so you do it. End of story. Only exception would be a scoreboard-influenced decision late in the game. But early in the second quarter, it's a no-brainer (assuming it's EV+, obviously).

It absolutely makes a difference. You have the luxury of time, it's another variable in the ultimate equation, and that affords you the opportunity to take more risks.

Explain to me why a late game scoreboard influenced decision is an exception? Is it because you no longer have time to make up for the failures of that gamble? According to your reasoning it's positive-EV so you should go for it.

One final example, say you $1 Million and I offered you two options, 1) we can flip a coin once, if you win I give you $1.5 Million, if you lose you give me your one million. EV of $250K Option 2) We flip a coin 1 million times, and each time you win I give you $1.45, each time you lose you give me $1. EV of $225K.

Which option would you take? Is the extra $25K EV worth the 50% likelihood that you walk away down $1 Million? Or would you take the nearly certain (or something close to it) $225K?

 

[TobiasFunke]

We're never gonna agree here, so let's just say we see it differently.

That said, I agree with a lot of what you say here about the variables, but couldn't possibly disagree more with the notion that your decisionmaking process should be different for a one time event as opposed to if you're playing seven games or 16 games or a million games. If it's EV+, it increases your chances of winning, so you do it. End of story. Only exception would be a scoreboard-influenced decision late in the game. But early in the second quarter, it's a no-brainer (assuming it's EV+, obviously).

It absolutely makes a difference. You have the luxury of time, it's another variable in the ultimate equation, and that affords you the opportunity to take more risks.

Explain to me why a late game scoreboard influenced decision is an exception? Is it because you no longer have time to make up for the failures of that gamble? According to your reasoning it's positive-EV so you should go for it.

One final example, say you $1 Million and I offered you two options, 1) we can flip a coin once, if you win I give you $1.5 Million, if you lose you give me your one million. EV of $250K Option 2) We flip a coin 1 million times, and each time you win I give you $1.45, each time you lose you give me $1. EV of $225K.

Which option would you take? Is the extra $25K EV worth the 50% likelihood that you walk away down $1 Million? Or would you take the nearly certain (or something close to it) $225K?

Regarding the bolded- late game situations can make a difference because maximizing points isn't always the best way to win. For example, if I have a terrible offense and I'm playing a great defense, a two-point conversion attempt will have a lower EV than an XP attempt, but if I'm trailing by two points with 30 seconds left I'm still gonna try for the two point conversion because it gives me the best chance to win. You don't have the same calculations early in the second quarter- the only goals are to score as many points as possible while also limiting the opponent to the fewest points possible (remember, even a failed 4th and 1 attempt would have left the Patriots 95 yards from the end zone instead of 75 or so). That's all I was saying- I didn't want to make a blanket statement, because there are a very small handful of situations where the highest EV decision isn't the correct one to give you the best chance of winning.As for the rest of it- your example is irrelevant because my answer would be influenced by (1) diminishing marginal returns on $ in terms of utility for me; and to a lesser degree (2) risk aversion. Neither of things should be a factor for coaches (although I think the latter often is). Their only consideration should be making the decision that gives the team the best chance to win.

 

[Modog814]

We're never gonna agree here, so let's just say we see it differently.

That said, I agree with a lot of what you say here about the variables, but couldn't possibly disagree more with the notion that your decisionmaking process should be different for a one time event as opposed to if you're playing seven games or 16 games or a million games. If it's EV+, it increases your chances of winning, so you do it. End of story. Only exception would be a scoreboard-influenced decision late in the game. But early in the second quarter, it's a no-brainer (assuming it's EV+, obviously).

It absolutely makes a difference. You have the luxury of time, it's another variable in the ultimate equation, and that affords you the opportunity to take more risks.

Explain to me why a late game scoreboard influenced decision is an exception? Is it because you no longer have time to make up for the failures of that gamble? According to your reasoning it's positive-EV so you should go for it.

One final example, say you $1 Million and I offered you two options, 1) we can flip a coin once, if you win I give you $1.5 Million, if you lose you give me your one million. EV of $250K Option 2) We flip a coin 1 million times, and each time you win I give you $1.45, each time you lose you give me $1. EV of $225K.

Which option would you take? Is the extra $25K EV worth the 50% likelihood that you walk away down $1 Million? Or would you take the nearly certain (or something close to it) $225K?

Regarding the bolded- late game situations can make a difference because maximizing points isn't always the best way to win. For example, if I have a terrible offense and I'm playing a great defense, a two-point conversion attempt will have a lower EV than an XP attempt, but if I'm trailing by two points with 30 seconds left I'm still gonna try for the two point conversion because it gives me the best chance to win. You don't have the same calculations early in the second quarter- the only goals are to score as many points as possible while also limiting the opponent to the fewest points possible (remember, even a failed 4th and 1 attempt would have left the Patriots 95 yards from the end zone instead of 75 or so). That's all I was saying- I didn't want to make a blanket statement, because there are a very small handful of situations where the highest EV decision isn't the correct one to give you the best chance of winning.As for the rest of it- your example is irrelevant because my answer would be influenced by (1) diminishing marginal returns on $ in terms of utility for me; and to a lesser degree (2) risk aversion. Neither of things should be a factor for coaches (although I think the latter often is). Their only consideration should be making the decision that gives the team the best chance to win.

And for what reason doesn't it give you the best chance of winning?

And the example is very relevant. Believe it or not, points have a diminishing utility too. Earlier in this thread we've already discussed the impact of reaching 30 points in a game and it's almost 90%+ win percentage when reaching that mark. You maximize points to win the game, but when you're win% is already in the 90's, each additional point increases it less. And risk aversion also has a role in decision making. Let me ask you, if Harbaugh thought his team was flat-out better than the Pats, why take a relatively risk-free situation (of 3 points) and make it risky for a marginal increase in EV? Theres no need to take unnecesary gambles when you believe you're the favorite or close to it. It's not like he disregarded a positive EV play for a negative one. He believed, for whatever reason, that taking the 3 points there gave him the best chance to win. You can disagree if you want, but it's impossible to prove mathematically that he was wrong.

 

[TobiasFunke]

We're never gonna agree here, so let's just say we see it differently.

That said, I agree with a lot of what you say here about the variables, but couldn't possibly disagree more with the notion that your decisionmaking process should be different for a one time event as opposed to if you're playing seven games or 16 games or a million games. If it's EV+, it increases your chances of winning, so you do it. End of story. Only exception would be a scoreboard-influenced decision late in the game. But early in the second quarter, it's a no-brainer (assuming it's EV+, obviously).

It absolutely makes a difference. You have the luxury of time, it's another variable in the ultimate equation, and that affords you the opportunity to take more risks.

Explain to me why a late game scoreboard influenced decision is an exception? Is it because you no longer have time to make up for the failures of that gamble? According to your reasoning it's positive-EV so you should go for it.

One final example, say you $1 Million and I offered you two options, 1) we can flip a coin once, if you win I give you $1.5 Million, if you lose you give me your one million. EV of $250K Option 2) We flip a coin 1 million times, and each time you win I give you $1.45, each time you lose you give me $1. EV of $225K.

Which option would you take? Is the extra $25K EV worth the 50% likelihood that you walk away down $1 Million? Or would you take the nearly certain (or something close to it) $225K?

Regarding the bolded- late game situations can make a difference because maximizing points isn't always the best way to win. For example, if I have a terrible offense and I'm playing a great defense, a two-point conversion attempt will have a lower EV than an XP attempt, but if I'm trailing by two points with 30 seconds left I'm still gonna try for the two point conversion because it gives me the best chance to win. You don't have the same calculations early in the second quarter- the only goals are to score as many points as possible while also limiting the opponent to the fewest points possible (remember, even a failed 4th and 1 attempt would have left the Patriots 95 yards from the end zone instead of 75 or so). That's all I was saying- I didn't want to make a blanket statement, because there are a very small handful of situations where the highest EV decision isn't the correct one to give you the best chance of winning.As for the rest of it- your example is irrelevant because my answer would be influenced by (1) diminishing marginal returns on $ in terms of utility for me; and to a lesser degree (2) risk aversion. Neither of things should be a factor for coaches (although I think the latter often is). Their only consideration should be making the decision that gives the team the best chance to win.

And for what reason doesn't it give you the best chance of winning?

And the example is very relevant. Believe it or not, points have a diminishing utility too. Earlier in this thread we've already discussed the impact of reaching 30 points in a game and it's almost 90%+ win percentage when reaching that mark. You maximize points to win the game, but when you're win% is already in the 90's, each additional point increases it less. And risk aversion also has a role in decision making. Let me ask you, if Harbaugh thought his team was flat-out better than the Pats, why take a relatively risk-free situation (of 3 points) and make it risky for a marginal increase in EV? Theres no need to take unnecesary gambles when you believe you're the favorite or close to it. It's not like he disregarded a positive EV play for a negative one. He believed, for whatever reason, that taking the 3 points there gave him the best chance to win. You can disagree if you want, but it's impossible to prove mathematically that he was wrong.

It doesn't give you the best chance of winning because there's not enough time left for additional possessions. That's not the case early in the second quarter. Pretty straightforward.I think the fact that your example of points having diminishing marginal returns required a blowout scenario. That wasn't the case when this decision we're talking about was made.

For example, take your $ hypothetical and divide the dollar figures by 1000 or 10,000. If that's the case I take the higher EV option every time, and so do most people. That hypo is far more relevant to a 3-0 game early in the second quarter than your example.

I don't follow your "unnecessary gambles" logic at all. Every play is a gamble, you never know what could happen when the ball is snapped. Using phrases like "unnecessary gambles" sounds like risk aversion to me, and that's "keep my job" coaching, not "win the game" coaching.

 

[Modog814]

It doesn't give you the best chance of winning because there's not enough time left for additional possessions. That's not the case early in the second quarter. Pretty straightforward.

Exactly. So number of trials remaining is a factor. So why do you think number of games in a series wouldn't be a factor?

I think the fact that your example of points having diminishing marginal returns required a blowout scenario. That wasn't the case when this decision we're talking about was made.

You pointed out in my example that you didn't think it was relevent because you're decision would be influenced by the possibility of diminishing returns on the amount of money you won. I was simply refuting your reasoning because it exists in the game scenario as well.

For example, take your $ hypothetical and divide the dollar figures by 1000 or 10,000. If that's the case I take the higher EV option every time, and so do most people. That hypo is far more relevant to a 3-0 game early in the second quarter than your example.

Yes, if lower the value of the bet, it becomes easier. Why? Because not much as at stake. In a one game scenario, the stakes are high, I'm arguing that at any point in a Conference Championship game the stakes are much closer to my hypothetical than your alternate hypothical of dividing the values by 1000 or 10,0000. It's all or nothing. It's not, "oh i lost, I'll get em next time." There is no next time, every single play, ever single possession is important. There is no need there to risk 3 points for an average possiblility of gaining and additional 1.1

I don't follow your "unnecessary gambles" logic at all. Every play is a gamble, you never know what could happen when the ball is snapped. Using phrases like "unnecessary gambles" sounds like risk aversion to me, and that's "keep my job" coaching, not "win the game" coaching.

No it's not "keep my job" coaching. The reasoning goes like this...I like my chances in this game, especially at 3-3. Sure, 7-3 would be better, but 0-3 could be disasterous if with the momentum swing Brady marches down the field and puts me in a 10-0 hole. I don't think I'd like my chances as much then against this high powered offense. Hmmmm, it is a tough one though, 7-3 would sure be sweet. And TobiasFunke tells me my EV on such a play is 4.1 compared to 3 for a FG. So on average, if I go for it, I can walk away ahead 4.1 to 3. That's really not much better than 3-3, especially since it's unlikely I keep the Pats from scoring again this game. Plus its a long yard, not just inches, and my QB isn't great at sneaks, and their Defense has been blowing up the run so far. I tried this last week too and failed, it could have cost me the game. If I go for it and don't score, I'm giving a not so great defense who's already playing well a real confidence boost. And my offense has struggled a bit, so getting any points would be a nice boost for them. Plus, if I kick the fg, the worst that can happen is the Pats score, and I'm still only down 1 score, still in the game. It's just not worth the risking of falling behind a team that thrives on having a lead. Besides, I liked my chances at 0-0, I still like my chances at 3-3. Now, the same (reversed) agrument could be made for going for it. The point is, there are many other factors involved that you can't possible put a number on get an accurate EV for that exact situation. Look, I'm not saying the decision was right, and I'm not saying it's wrong. All I'm saying is there is too many variables to even come up with an accurate EV calculation for that situation. And even if you could, it's possible (read that, possible, I'm not saying always, just that its possible) that selecting the lower EV can actually give you the better chance at winning. Look back to my hypothetical, multiply or divide the amounts by however much you want. The truth is, option 1 gives you a 50% chance of losing money. Option 2, you're almost certain to end up winning. Obviously, football is more complicated than that, but that's the point, it's too complex to be able to accurately calculate an EV of a play. We could probably go back and forth like this all day, and if we haven't changed each others mind by now, we probably never will.

 

[TobiasFunke]

It doesn't give you the best chance of winning because there's not enough time left for additional possessions. That's not the case early in the second quarter. Pretty straightforward.

Exactly. So number of trials remaining is a factor. So why do you think number of games in a series wouldn't be a factor?

I think the fact that your example of points having diminishing marginal returns required a blowout scenario. That wasn't the case when this decision we're talking about was made.

You pointed out in my example that you didn't think it was relevent because you're decision would be influenced by the possibility of diminishing returns on the amount of money you won. I was simply refuting your reasoning because it exists in the game scenario as well.

For example, take your $ hypothetical and divide the dollar figures by 1000 or 10,000. If that's the case I take the higher EV option every time, and so do most people. That hypo is far more relevant to a 3-0 game early in the second quarter than your example.

Yes, if lower the value of the bet, it becomes easier. Why? Because not much as at stake. In a one game scenario, the stakes are high, I'm arguing that at any point in a Conference Championship game the stakes are much closer to my hypothetical than your alternate hypothical of dividing the values by 1000 or 10,0000. It's all or nothing. It's not, "oh i lost, I'll get em next time." There is no next time, every single play, ever single possession is important. There is no need there to risk 3 points for an average possiblility of gaining and additional 1.1

I don't follow your "unnecessary gambles" logic at all. Every play is a gamble, you never know what could happen when the ball is snapped. Using phrases like "unnecessary gambles" sounds like risk aversion to me, and that's "keep my job" coaching, not "win the game" coaching.

No it's not "keep my job" coaching. The reasoning goes like this...I like my chances in this game, especially at 3-3. Sure, 7-3 would be better, but 0-3 could be disasterous if with the momentum swing Brady marches down the field and puts me in a 10-0 hole. I don't think I'd like my chances as much then against this high powered offense. Hmmmm, it is a tough one though, 7-3 would sure be sweet. And TobiasFunke tells me my EV on such a play is 4.1 compared to 3 for a FG. So on average, if I go for it, I can walk away ahead 4.1 to 3. That's really not much better than 3-3, especially since it's unlikely I keep the Pats from scoring again this game. Plus its a long yard, not just inches, and my QB isn't great at sneaks, and their Defense has been blowing up the run so far. I tried this last week too and failed, it could have cost me the game. If I go for it and don't score, I'm giving a not so great defense who's already playing well a real confidence boost. And my offense has struggled a bit, so getting any points would be a nice boost for them. Plus, if I kick the fg, the worst that can happen is the Pats score, and I'm still only down 1 score, still in the game. It's just not worth the risking of falling behind a team that thrives on having a lead. Besides, I liked my chances at 0-0, I still like my chances at 3-3.

Now, the same (reversed) agrument could be made for going for it. The point is, there are many other factors involved that you can't possible put a number on get an accurate EV for that exact situation.

Look, I'm not saying the decision was right, and I'm not saying it's wrong. All I'm saying is there is too many variables to even come up with an accurate EV calculation for that situation. And even if you could, it's possible (read that, possible, I'm not saying always, just that its possible) that selecting the lower EV can actually give you the better chance at winning. Look back to my hypothetical, multiply or divide the amounts by however much you want. The truth is, option 1 gives you a 50% chance of losing money. Option 2, you're almost certain to end up winning. Obviously, football is more complicated than that, but that's the point, it's too complex to be able to accurately calculate an EV of a play.

We could probably go back and forth like this all day, and if we haven't changed each others mind by now, we probably never will.

Yup. My only issue here is with you saying that 4.1 EV and 2.8 EV or whatever it would be for the FG attempt is remotely close. That's a HUGE difference for a single play. But otherwise, I think we've talked it out as much as we can.

BTW, enjoyed it greatly. See you for the next coaching decision analysis thread