Around the NHL 2023-2024

[MissouriMook]

Right, but they’re 1 for 2 now. What do you think the success rate is? I’m not even sure it’s .500 over time.

Points-wise, they are now in the same position as if they had lost both games. I don’t know what their combined expected points would have been at the beginning of OT for those two games, but I would bet my house it would be greater than 2.

 

[Bye Bye Blueston]

Points-wise, they are now in the same position as if they had lost both games. I don’t know what their combined expected points would have been at the beginning of OT for those two games, but I would bet my house it would be greater than 2.

Yeah, it’s dumb. Would have made more sense to pull goalie in regulation to try to deny any points to opponent. But this far out that is probably not smart either.

 

[Brian39]

Points-wise, they are now in the same position as if they had lost both games. I don’t know what their combined expected points would have been at the beginning of OT for those two games, but I would bet my house it would be greater than 2.

The math goes beyond just the raw total of expected points earned vs not earned in those specific games. Minnesota is far enough back in the standings that the difference between 1 point and 0 points is fairly irrelevant compared to the opportunity to get 2 points. They needed 2 points much more than they needed to salvage 1 point. That game was 'must win' not 'must at least get a point.'

The analytics aren't just about the expected point totals. They also include the variable of what those point totals mean to your chances of earning enough points to reach the playoff cut line.

Let's say the playoff cutline is gong to be 97 points (LA's current pace rounding down). Any end-of-year point total below 97 is a failure. Gaining a couple extra OT points and ending the year with 94 points instead of 92 is a failure. Those 'extra' points you earn are meaningless. The relevant math is how you reach 97 points. The only scenarios where that extra .5 expected point via the loser point is relevant are the handful of scenarios where you subsequently earn enough points in your other games to still push your total up to the playoff cut line.

I'm not sure if it was the correct decision, but I'd also bet my house that the analytics referenced are regarding their resulting playoff chances based on earning 2 vs 1 vs 0 points in that specific game, not simply the expected point totals over a growing sample size. I'd bet good money that the analytics referred to basically said "we have to go undefeated in any game that is tied down the stretch in order to have any chance of making the playoffs. Any loss, even with a loser point in OT, ends the season." At that point, increasing your chances of earning 2 points at the expense of a worthless loser point is the correct move.

If I have $20 to my name and I'm going to lose my house next week unless I can get the bank $50,000, the analytics don't tell me to invest my $20 in the S+P. They tell me to play the lottery with that $20. My expected ROI is substantially worse with the lottery vs a safe investment, but a week's worth of safe returns is meaningless in that situation. The odds are overwhelmingly high that I will lose the house either way and putting that $20 into the S+P means that I will likely end next week with an extra $20 and change. But that extra money is meaningless compared to the ever so slight chance that I could win the lottery and save my house.

That's the reality an NHL team is living under when they are 8 points back of the playoff cut line with 10 games to got. It's hail mary time.

 

[Brian39]

Are you actually trying to say that Forsberg and Nyquist are carrying ROR because I don't think it's a coincidence that they are both having career years while playing with ROR, pretty sure ROR is the one carrying them. So saying Forseberg is better than Buch or Nyquist is better than Perron is funny to me, the best you could say is that they fit ROR's style better than anyone we currently have. I think if we still had ROR and Perron that Neighbors would be a great winger to go with them and that entire line would work great together.

No.

The question was whether we could offer ROR the type of supporting cast he has in Nashville. That answer is overwhelmingly no. There is a ton of space between "is the supporting cast as strong" and "is the supporting cast carrying a player."

Forsberg is a better player than Buch. There is a very good reason that Nashville gave him $8.5M x 8 years two summers ago. He is now a two-time 40 goal scorer and a two-time 84 point player. He played at a 34.4 goal and 74.5 point pace over the 6 seasons leading into this one, which included 42 goals and 84 points in 69 games in 2021/22. He is a noticeably better goal scorer than Buch. Kyrou may or may not ever become the player Forsberg is, but he's not there yet and had no chemistry with ROR last year.

Josi is a substantially better player than any D we have. This shouldn't even need to be a discussion. He is obviously a better player than any D we have. He has a Norris trophy, a runner-up Norris season and four other top 10 Norris vote finishes. He's been a clear #1 D for a decade and played at a 78 point pace heading into this season. ROR is very obviously not carrying him.

Josi and Forsberg are the two best skaters on the Preds and are better players than anyone we have at their positions. The difference between those two and the best W/D he'd get to play with on our team is a hell of a lot more pronounced than just a better stylistic fit. I don't think there is even a remotely serious argument to the contrary.

Watching Perron this year, I think it is crystal clear that he has lost several steps and isn't the player Nyquist is. But even if you say they are equal, he would have to do a massive amount more heavy lifting than Nyquist currently is to make up for the gap between Forsberg/Josi and what we could offer ROR on the top line and top PP unit.

I'm baffled at an argument that Neighbours-ROR-Perron with any combinaiton of our D group is a better group than Forsberg-ROR-Nyquist and Josi. None of that means that they are 'carrying' ROR. But playing with better players is overwhelmingly better than playing with worse players.

 

[Xerloris]

No.

The question was whether we could offer ROR the type of supporting cast he has in Nashville. That answer is overwhelmingly no. There is a ton of space between "is the supporting cast as strong" and "is the supporting cast carrying a player."

Forsberg is a better player than Buch. There is a very good reason that Nashville gave him $8.5M x 8 years two summers ago. He is now a two-time 40 goal scorer and a two-time 84 point player. He played at a 34.4 goal and 74.5 point pace over the 6 seasons leading into this one, which included 42 goals and 84 points in 69 games in 2021/22. He is a noticeably better goal scorer than Buch. Kyrou may or may not ever become the player Forsberg is, but he's not there yet and had no chemistry with ROR last year.

Josi is a substantially better player than any D we have. This shouldn't even need to be a discussion. He is obviously a better player than any D we have. He has a Norris trophy, a runner-up Norris season and four other top 10 Norris vote finishes. He's been a clear #1 D for a decade and played at a 78 point pace heading into this season. ROR is very obviously not carrying him.

Josi and Forsberg are the two best skaters on the Preds and are better players than anyone we have at their positions. The difference between those two and the best W/D he'd get to play with on our team is a hell of a lot more pronounced than just a better stylistic fit. I don't think there is even a remotely serious argument to the contrary.

Watching Perron this year, I think it is crystal clear that he has lost several steps and isn't the player Nyquist is. But even if you say they are equal, he would have to do a massive amount more heavy lifting than Nyquist currently is to make up for the gap between Forsberg/Josi and what we could offer ROR on the top line and top PP unit.

I'm baffled at an argument that Neighbours-ROR-Perron with any combinaiton of our D group is a better group than Forsberg-ROR-Nyquist and Josi. None of that means that they are 'carrying' ROR. But playing with better players is overwhelmingly better than playing with worse players.

My point was that I do not think ROR needs to be carried, he simply needs a good line to go with him stylistically. Which is why I said if we still had ROR and Perron I think Neighbors would go good with them. I was not saying Perron and Neighbors would be better. I think how ever that if we still had ROR and Perron that adding Neighbors would give us an extremely strong 2nd line and I don't think it would hardly effect ROR's stats he's having this year. Also, it's kind of weird you keep referring to Josi as if he's on a line with ROR.


Just to add, this has nothing to do with you but I think it was dumb as f*** how Berube kept putting Kyrou with ROR.

 

[MissouriMook]

The math goes beyond just the raw total of expected points earned vs not earned in those specific games. Minnesota is far enough back in the standings that the difference between 1 point and 0 points is fairly irrelevant compared to the opportunity to get 2 points. They needed 2 points much more than they needed to salvage 1 point. That game was 'must win' not 'must at least get a point.'

The analytics aren't just about the expected point totals. They also include the variable of what those point totals mean to your chances of earning enough points to reach the playoff cut line.

Let's say the playoff cutline is gong to be 97 points (LA's current pace rounding down). Any end-of-year point total below 97 is a failure. Gaining a couple extra OT points and ending the year with 94 points instead of 92 is a failure. Those 'extra' points you earn are meaningless. The relevant math is how you reach 97 points. The only scenarios where that extra .5 expected point via the loser point is relevant are the handful of scenarios where you subsequently earn enough points in your other games to still push your total up to the playoff cut line.

I'm not sure if it was the correct decision, but I'd also bet my house that the analytics referenced are regarding their resulting playoff chances based on earning 2 vs 1 vs 0 points in that specific game, not simply the expected point totals over a growing sample size. I'd bet good money that the analytics referred to basically said "we have to go undefeated in any game that is tied down the stretch in order to have any chance of making the playoffs. Any loss, even with a loser point in OT, ends the season." At that point, increasing your chances of earning 2 points at the expense of a worthless loser point is the correct move.

If I have $20 to my name and I'm going to lose my house next week unless I can get the bank $50,000, the analytics don't tell me to invest my $20 in the S+P. They tell me to play the lottery with that $20. My expected ROI is substantially worse with the lottery vs a safe investment, but a week's worth of safe returns is meaningless in that situation. The odds are overwhelmingly high that I will lose the house either way and putting that $20 into the S+P means that I will likely end next week with an extra $20 and change. But that extra money is meaningless compared to the ever so slight chance that I could win the lottery and save my house.

That's the reality an NHL team is living under when they are 8 points back of the playoff cut line with 10 games to got. It's hail mary time.

I think you're overcomplicating this in your analysis and its leading you to a bad conclusion.

First, the strategy takes the expected points entering overtime from a range of 1.0 to 2.0 and stretches it to 0.0 to 2.0 and there is no extraneous value from being able to earn more than the maximum two points. Let's say two fairly evenly matched teams enter overtime with identical 1.5 points expected. When you pull your goalie, you'll need to expect to score 75% of the time in that situation to reach the same 1.5 expected points. I don't know that there is a team in the league that would be that efficient at 4-on-3 with the goalie pulled, and even if there were I certainly wouldn't count Minnesota among the likely candidates.

Finally, as I alluded to before, there is no ability to earn more than the two points that are already available if you keep the goalie in the net, so your analgy to using $20 on the lottery as a Hail Mary doesn't really translate here. I think the biggest point you're missing here is that the point you spit up by not retaining the "loser point" is worth just as much as the extra point you gain winning the game. To me, the only way this makes sense is a situation where it is Game 82 and you have to have two points to make the playoffs and one point leaves you on the outside looking in. At that point, fire away, but until then respect the point you've earned getting the game to overtime and work hard to get the second one.

 

[Ranksu]

Joshua beasting on Vancouver. Another Bad move Army let him go.

Btw sick move that between legs goal.

 

[Brian39]

My point was that I do not think ROR needs to be carried, he simply needs a good line to go with him stylistically. Which is why I said if we still had ROR and Perron I think Neighbors would go good with them. I was not saying Perron and Neighbors would be better. I think how ever that if we still had ROR and Perron that adding Neighbors would give us an extremely strong 2nd line and I don't think it would hardly effect ROR's stats he's having this year. Also, it's kind of weird you keep referring to Josi as if he's on a line with ROR.


Just to add, this has nothing to do with you but I think it was dumb as f*** how Berube kept putting Kyrou with ROR.

I don't know what to tell you if you don't think that a Norris-caliber offensive D man impacts how everyone on the ice plays. They have played a ton together this season and ROR has been a beneficiary (Just like every forward who is out there with an elite scoring D man who can also competently defend).

ROR has been on the ice with Josi for 52.6% of his total TOI this season and those situations account for 40 of his 61 points. 17 of those 40 points have been secondary assists. They have played together a ton this year, which includes the top PP unit where ROR has racked up 14 of his 25 goals. Who do you think is running that PP? That offense starts and runs through Josi when he is on the ice and that includes the large amount of time he spends out there with ROR and the rest of their top forward line.

Josi drives the bus offensively on that team and he has for years. ROR has played over half his minutes with Josi and that is where he has scored two thirds of his goals and points. ROR has absolutely benefited from playing with Josi and it is bizarre to suggest otherwise. We don't have a D man who drives play or runs a PP anything like the way Josi does.

None of this means that ROR is getting carried and at no point have I said he is being carried. But it absolutely means that he almost certainly wouldn't be putting up the same numbers if he was playing with worse players.

Getting to play a bunch of your minutes with a Norris-caliber D and all of your minutes with an elite winger puts you in better position to succeed than playing them with Torey Krug and a lesser winger. With Perron or not, we do not have the personnel to put him in the same position to succeed as he is in Nashville.

 

[Brian39]

I think the biggest point you're missing here is that the point you spit up by not retaining the "loser point" is worth just as much as the extra point you gain winning the game.

I'm not missing that point, I'm disagreeing with it. Let's look at some of their playoff odds with different records:

Going 7-1-1 down the stretch gives them a 9.1% chance of making the playoffs. Now, let's focus on the impact of the game where they could get that loser point (for the sake of this, assume that they can/will go 7-1 in the rest of their games besides that game):

Pulling the goalie and winning that game instead of taking the loser point (so going 8-1) gives them a 24.1% chance at the playoffs. Winning the game with an empty net thus makes it 2.65 times more likely that they will make the playoffs than if they had held on to the loser point.

Pulling the goalie, losing, and forfeiting the loser point (going 7-2) gives them a 3.8% chance of making the playoffs. That makes it 2.39 times less likely that they will make the playoffs than if they had held on to the loser point).

With such a high percentage of their games being 'must win' games, the difference between successfully winning the game is greater than the difference between taking a loser point and taking no points in a loss. There is absolutely more to 'gain' by adding a point than by losing a point.

Again, I'm not sold that Minnesota made the right decision. I'm not sold that they actually increased their odds of winning enough by doing this. But the 2nd point is worth more than the 1st point when they need to earn 1.75+ points in each of their remaining games.

 

[BlueDream]

Joshua beasting on Vancouver. Another Bad move Army let him go.

Btw sick move that between legs goal.

I remember when I expressed my disappointment about his departure I was told on here that he was just a 4th liner and it didn’t matter. Typical.

 

[Xerloris]

I don't know what to tell you if you don't think that a Norris-caliber offensive D man impacts how everyone on the ice plays. They have played a ton together this season and ROR has been a beneficiary (Just like every forward who is out there with an elite scoring D man who can also competently defend).

ROR has been on the ice with Josi for 52.6% of his total TOI this season and those situations account for 40 of his 61 points. 17 of those 40 points have been secondary assists. They have played together a ton this year, which includes the top PP unit where ROR has racked up 14 of his 25 goals. Who do you think is running that PP? That offense starts and runs through Josi when he is on the ice and that includes the large amount of time he spends out there with ROR and the rest of their top forward line.

Josi drives the bus offensively on that team and he has for years. ROR has played over half his minutes with Josi and that is where he has scored two thirds of his goals and points. ROR has absolutely benefited from playing with Josi and it is bizarre to suggest otherwise. We don't have a D man who drives play or runs a PP anything like the way Josi does.

None of this means that ROR is getting carried and at no point have I said he is being carried. But it absolutely means that he almost certainly wouldn't be putting up the same numbers if he was playing with worse players.

Getting to play a bunch of your minutes with a Norris-caliber D and all of your minutes with an elite winger puts you in better position to succeed than playing them with Torey Krug and a lesser winger. With Perron or not, we do not have the personnel to put him in the same position to succeed as he is in Nashville.

He would be a 2nd line center here behind Thomas, which would make his opponent difficulty easier, which in theory could negate any down turn his production would take with lesser partners.

Also, Sure Josi contributes to his points but it's not like Josi is his winger, Josi has his own partner he plays with.

 

[stl76]

I think you're overcomplicating this in your analysis and its leading you to a bad conclusion.

First, the strategy takes the expected points entering overtime from a range of 1.0 to 2.0 and stretches it to 0.0 to 2.0 and there is no extraneous value from being able to earn more than the maximum two points. Let's say two fairly evenly matched teams enter overtime with identical 1.5 points expected. When you pull your goalie, you'll need to expect to score 75% of the time in that situation to reach the same 1.5 expected points. I don't know that there is a team in the league that would be that efficient at 4-on-3 with the goalie pulled, and even if there were I certainly wouldn't count Minnesota among the likely candidates.

Finally, as I alluded to before, there is no ability to earn more than the two points that are already available if you keep the goalie in the net, so your analgy to using $20 on the lottery as a Hail Mary doesn't really translate here. I think the biggest point you're missing here is that the point you spit up by not retaining the "loser point" is worth just as much as the extra point you gain winning the game. To me, the only way this makes sense is a situation where it is Game 82 and you have to have two points to make the playoffs and one point leaves you on the outside looking in. At that point, fire away, but until then respect the point you've earned getting the game to overtime and work hard to get the second one.

It’s not about points or expected points. It’s about expected utility of the point(s).

Expected utility hypothesis - Wikipedia

en.m.wikipedia.org

An example: if there is a 50% of getting 1 point, 40% chance of getting 0 points, and a 10% chance of getting 2 points, it may still make sense to go for two points if the utility you get from 0 or 1 point(s) is low compared to the utility of getting 2 points.

I don’t know what the analytics say or how the math looks for pulling the goalie in OT. But as a fan, I love it. Super entertaining IMO, hope it happens more often.

 

[MissouriMook]

It’s not about points or expected points. It’s about expected utility of the point(s).

Expected utility hypothesis - Wikipedia

en.m.wikipedia.org

An example: if there is a 50% of getting 1 point, 40% chance of getting 0 points, and a 10% chance of getting 2 points, it may still make sense to go for two points if the utility you get from 0 or 1 point(s) is low compared to the utility of getting 2 points.

I don’t know what the analytics say or how the math looks for pulling the goalie in OT. But as a fan, I love it. Super entertaining IMO, hope it happens more often.

My argument about this being a bad long term strategy was that you have to compound these probable outcomes over multiple games, thus diluting the benefit when/if it works if there is a future failure. Applying this strategy twice and winning one game in OT with the goalie pulled and losing the other leaves you in the same position as you would have been in if you had lost both games in OT with the goalie in the net.

Further, I was arguing that this was unlikely to be a 50/50 proposition - it probably fails closer to twice for each time it works, but that is just my guess. To me, the only time this makes any sense is if you're down to the last game (maybe two) and failing to get two points mathematically eliminates you. Until you reach that point, I feel that the points you save outweigh the potential for getting the second point based on the pure statistical likelihood of success or failure.

 

[Stupendous Yappi]

The math goes beyond just the raw total of expected points earned vs not earned in those specific games. Minnesota is far enough back in the standings that the difference between 1 point and 0 points is fairly irrelevant compared to the opportunity to get 2 points. They needed 2 points much more than they needed to salvage 1 point. That game was 'must win' not 'must at least get a point.'

The analytics aren't just about the expected point totals. They also include the variable of what those point totals mean to your chances of earning enough points to reach the playoff cut line.

Let's say the playoff cutline is gong to be 97 points (LA's current pace rounding down). Any end-of-year point total below 97 is a failure. Gaining a couple extra OT points and ending the year with 94 points instead of 92 is a failure. Those 'extra' points you earn are meaningless. The relevant math is how you reach 97 points. The only scenarios where that extra .5 expected point via the loser point is relevant are the handful of scenarios where you subsequently earn enough points in your other games to still push your total up to the playoff cut line.

I'm not sure if it was the correct decision, but I'd also bet my house that the analytics referenced are regarding their resulting playoff chances based on earning 2 vs 1 vs 0 points in that specific game, not simply the expected point totals over a growing sample size. I'd bet good money that the analytics referred to basically said "we have to go undefeated in any game that is tied down the stretch in order to have any chance of making the playoffs. Any loss, even with a loser point in OT, ends the season." At that point, increasing your chances of earning 2 points at the expense of a worthless loser point is the correct move.

If I have $20 to my name and I'm going to lose my house next week unless I can get the bank $50,000, the analytics don't tell me to invest my $20 in the S+P. They tell me to play the lottery with that $20. My expected ROI is substantially worse with the lottery vs a safe investment, but a week's worth of safe returns is meaningless in that situation. The odds are overwhelmingly high that I will lose the house either way and putting that $20 into the S+P means that I will likely end next week with an extra $20 and change. But that extra money is meaningless compared to the ever so slight chance that I could win the lottery and save my house.

That's the reality an NHL team is living under when they are 8 points back of the playoff cut line with 10 games to got. It's hail mary time.

I am saying I suspect the team defending the 4 in 3 gets 2 points more often than the other way around. I think we all understand the urgency about 2 points vs 1.

If pulling your goalie DECREASES your chance to get 2 points, there are no analytics that will compute that is the smart choice.

 

[MissouriMook]

I'm not missing that point, I'm disagreeing with it. Let's look at some of their playoff odds with different records:

Going 7-1-1 down the stretch gives them a 9.1% chance of making the playoffs. Now, let's focus on the impact of the game where they could get that loser point (for the sake of this, assume that they can/will go 7-1 in the rest of their games besides that game):

Pulling the goalie and winning that game instead of taking the loser point (so going 8-1) gives them a 24.1% chance at the playoffs. Winning the game with an empty net thus makes it 2.65 times more likely that they will make the playoffs than if they had held on to the loser point.

Pulling the goalie, losing, and forfeiting the loser point (going 7-2) gives them a 3.8% chance of making the playoffs. That makes it 2.39 times less likely that they will make the playoffs than if they had held on to the loser point).

With such a high percentage of their games being 'must win' games, the difference between successfully winning the game is greater than the difference between taking a loser point and taking no points in a loss. There is absolutely more to 'gain' by adding a point than by losing a point.

Again, I'm not sold that Minnesota made the right decision. I'm not sold that they actually increased their odds of winning enough by doing this. But the 2nd point is worth more than the 1st point when they need to earn 1.75+ points in each of their remaining games.

In the long view you're taking here, you're using probabilities of making the playoffs as if these games occur in a vacuum. You don't need to win a certain number of games, you need to accumulate a certain number of points more than the team(s) you are chasing. My point is based on the assumption (admittedly unproven) that this is likely worse than a 50/50 proposition and in a longer time frame you are likely to surrender more points than you gain. The only time I would agree that this is a worthwhile strategy is if you would be mathematically eliminated by gaining any less than the two available points. The longer out from the end of the season you start implementing this strategy, the less likely it is to actually help you, unless you can make a case for a positive expectation of points gained over leaving the goalie in. I would argue against that notion.

 

[MissouriMook]

I am saying I suspect the team defending the 4 in 3 gets 2 points more often than the other way around. I think we all understand the urgency about 2 points vs 1.

If pulling your goalie DECREASES your chance to get 2 points, there are no analytics that will compute that we the smart choice.

I would argue the opposite of the bolded, but I suspect the data to this point would be unreliable given how rarely it occurs. Using 6 on 5 as a guide, I suspect (without any data in front of me) that this probably has a success rate of 25%-35% over a much, much larger sample size. I would concede that the odds would improve at 4 on 3 over 6 on 5, but I don't think it is enough to get you to 50%. I think the most interesting variable would be what the expectation would be that the OT goes the remainder of the OT with neither team scoring. I'd say that is probably less than 10% (depending on when the goalie is pulled), but I think it mutes some of the value of the strategy as the positive expectation comes from scoring, not from not getting scored on.

 

[Stupendous Yappi]

I would argue the opposite of the bolded, but I suspect the data to this point would be unreliable given how rarely it occurs. Using 6 on 5 as a guide, I suspect (without any data in front of me) that this probably has a success rate of 25%-35% over a much, much larger sample size. I would concede that the odds would improve at 4 on 3 over 6 on 5, but I don't think it is enough to get you to 50%. I think the most interesting variable would be what the expectation would be that the OT goes the remainder of the OT with neither team scoring. I'd say that is probably less than 10% (depending on when the goalie is pulled), but I think it mutes some of the value of the strategy as the positive expectation comes from scoring, not from not getting scored on.

You sound like you’re agreeing with me, but saying you aren’t. I think the 4 on 3 succeeds less than 50% of attempts.

I am not sure solid data really exist since this has been attempted so infrequently.

 

[stl76]

My argument about this being a bad long term strategy was that you have to compound these probable outcomes over multiple games, thus diluting the benefit when/if it works if there is a future failure. Applying this strategy twice and winning one game in OT with the goalie pulled and losing the other leaves you in the same position as you would have been in if you had lost both games in OT with the goalie in the net.

Further, I was arguing that this was unlikely to be a 50/50 proposition - it probably fails closer to twice for each time it works, but that is just my guess. To me, the only time this makes any sense is if you're down to the last game (maybe two) and failing to get two points mathematically eliminates you. Until you reach that point, I feel that the points you save outweigh the potential for getting the second point based on the pure statistical likelihood of success or failure.

To the first paragraph, unless I am misunderstanding you - that’s just not correct. Two games are two distinct events. The outcome of one game has no direct calculable effect on the outcome of another game that I am aware of. If the calculations show that the best option is to pull the goalie, then you pull the goalie. Doesn’t matter what happened in a previous game - other than the caveat that part of the calculations may include previous events and/or conditional probabilities.

To the second paragraph, you’re best off using tools like expected utility theory and optimal decision theory to arrive at a solution to pull the goalie or not. I’m (poorly) trying to lay out the process for calculating the optimal decision. What you or I feel/guess is frankly irrelevant.

Personally, I don’t have the data necessary to have a strong opinion here on what the best decision is. I hope more coaches pull the goalie in OT because it’s exciting IMO, that’s about all I know.

 

[MissouriMook]

You sound like you’re agreeing with me, but saying you aren’t. I think the 4 on 3 succeeds less than 50% of attempts.

I am not sure solid data really exist since this has been attempted so infrequently.

Yes, I misread the bolded part. I thought you were saying the strategy worked more than half the time.

 

[MissouriMook]

To the first paragraph, unless I am misunderstanding you - that’s just not correct. Two games are two distinct events. The outcome of one game has no direct calculable effect on the outcome of another game that I am aware of. If the calculations show that the best option is to pull the goalie, then you pull the goalie. Doesn’t matter what happened in a previous game - other than the caveat that part of the calculations may include previous events and/or conditional probabilities.

To the second paragraph, you’re best off using tools like expected utility theory and optimal decision theory to arrive at a solution to pull the goalie or not. I’m (poorly) trying to lay out the process for calculating the optimal decision. What you or I feel/guess is frankly irrelevant.

Personally, I don’t have the data necessary to have a strong opinion here on what the best decision is. I hope more coaches pull the goalie in OT because it’s exciting IMO, that’s about all I know.

On the first part, I'm not suggesting that they're connected, just that over a longer time frame than just a game or two the law of averages will catch up with you and you will do what the Wild did - win one and lose one - essentially neutering the purpose of the strategy when the outcome of two events is worse on an expected basis than if you had left the goalie in.

I understand what you're saying about the utility theory (at least I think I do) but that doesn't change the fact that the underlying mathematics don't lie. If you can't sustain a success rate well in excess of 50% (probably more like 75%) this strategy is a losing one over time because you can never get more than the two points. It would change my opinion (not to mention be an awesome topic for discussion) if pulling the goalie would allow you to capture all three points when you win since you get none when you lose.

As I have explained in other posts, it seems like (to use @Brian39 's term) a Hail Mary that you should only pull out when not getting both points will officially and mathematically eliminate you. Otherwise, if you pull it out too soon and have to repeat the trick, you're almost certainly going to squander valuable points along the way. Like they did.

 

[Stupendous Yappi]

I could see some sense in pulling the goalie if you're in a playoff race that is likely to come down to the tie-breaker for ROW points, and you can't tolerate getting the extra point in the shoot-out. Or if your goalie has diarrhea.

 

[Ranksu]

 

[Ted Hoffman]

I get the idea behind pulling the goalie if you're in a playoff race, but like pulling the goalie in any game: there's a point where it makes sense, and there's a point where it doesn't. I think a very non-trivial part of the strategy is nothing more than "pull the goalie, ????????, get points!" which sounds great, but as we saw this team do last season, if you don't have a strategy for what you're doing when you have the extra attacker there's a high chance of the opposition just firing it down into the empty net. Which, yes, if you're taking a loss it doesn't matter if you lose by 1 or 4 and so theoretically the gain from going with the extra attacker outweighs any downside [which is what teh analytics on pulling the goalie argue] but it's irrelevant if you don't have a strategy to increase your chances of success.

And what made Minnesota's strategy "look good" is the same fallacy people fall for way too often: high confidence in low-probability events repeating. Oh, look at what they did, what a great idea! Yeah, until as @MissouriMook pointed out: it worked the first time, it didn't work the 2nd. You're in the same spot as if you'd taken 2 OTLs, and you don't know if you might have squeezed out a win in the shootout in the 2nd game to get 4 points instead of the 2 you realized. That extra point or two could have Minnesota still within spitting distance of a playoff spot, with 2 games in hand on us; instead, they're effectively 9 out [8 + 1 to overcome tiebreaks] with 8 to play, and still facing low-probability chances of winning in OT with the extra attacker vs. getting 0 for pulling the goalie and losing there again.

 

[stl76]

On the first part, I'm not suggesting that they're connected, just that over a longer time frame than just a game or two the law of averages will catch up with you and you will do what the Wild did - win one and lose one - essentially neutering the purpose of the strategy when the outcome of two events is worse on an expected basis than if you had left the goalie in.

I understand what you're saying about the utility theory (at least I think I do) but that doesn't change the fact that the underlying mathematics don't lie. If you can't sustain a success rate well in excess of 50% (probably more like 75%) this strategy is a losing one over time because you can never get more than the two points. It would change my opinion (not to mention be an awesome topic for discussion) if pulling the goalie would allow you to capture all three points when you win since you get none when you lose.

As I have explained in other posts, it seems like (to use @Brian39 's term) a Hail Mary that you should only pull out when not getting both points will officially and mathematically eliminate you. Otherwise, if you pull it out too soon and have to repeat the trick, you're almost certainly going to squander valuable points along the way. Like they did.

That’s not really how this works if you want to make an optimal decision. What you’re describing sounds more like the gamblers fallacy.

The law of averages would actually agree with what I’m saying: if the calculations show that the best option is to pull the goalie, then you pull the goalie. Doesn’t matter what happened in a previous game - other than the caveat that part of the calculations may include previous events and/or conditional probabilities.

I think the fundamental disconnect here is that you seem to be focused on maximizing standings points. I’m saying the goal should be on maximizing the likelihood of making the playoffs. The two are similar but different in important ways.

 

[Ted Hoffman]

That’s not really how this works if you want to make an optimal decision. What you’re describing sounds more like the gamblers fallacy.

The law of averages would actually agree with what I’m saying: if the calculations show that the best option is to pull the goalie, then you pull the goalie. Doesn’t matter what happened in a previous game - other than the caveat that part of the calculations may include previous events and/or conditional probabilities.

I think the fundamental disconnect here is that you seem to be focused on maximizing standings points. I’m saying the goal should be on maximizing the likelihood of making the playoffs. The two are similar but different in important ways.

If you're in OT and guaranteed a point, and presuming pulling the goalie results in a decision and doesn't cause the game to go to a shootout, then it's an optimal decision to pull the goalie and go for 2 if that decision will be successful more than half the time. [Which is probably not the case.] In reality, once you pull the goalie in OT there's 4 possible outcomes - win in OT (2 points), lose in OT (0 points), win in the shootout (2 points), lose in the shootout (1 point) - and the question becomes what probabilities should we assign to those scenarios? which is going to vary depending on when you decide to pull the goalie in OT. And to a prior point I made: your strategy with the extra attacker impacts those probabilities, both positively and negatively

To the last sentence: while you're right that maximizing points doesn't maximize playoff chances, it's really "maximize points to get to where you've maximized playoff chances" because at a certain threshold, more points doesn't increase your playoff chances. You improve your seed with the additional points, but now we're into a different discussion that doesn't apply when you're just trying to make the playoffs.