2022-01-21 Princeton

[Robb]

We were thinking this could be a 1W 1L weekend. Better the L is today.

That isn't true for PWR.

Could you explain that please - is it because of the clause that prevents a team's RPI suffering if it beats a bad team?

Otherwise, I would not think it would matter much, since RPI has three criteria:

A team's own winning percentage (25%) - Cornell's winning percentage would be unaffected
The average of the team's opponents' winning percentages (21%) - I guess the boost to Princeton's winning% is less that the hit to Q's - is this where it matters?
The average of the team's opponents opponents' winning percentages (54%) - I can't believe this is significantly impacted by swapping the wins.

And in this case, a loss to Q will flip the NoDak pairwise comparison.

I am slavishly repeating what I have heard.  My intuition is they should be exactly the same impact provided we play both teams the same number of times.

I guess it is arguably more likely that by the end of Placid we will have played Q more.

Yes, but the number of times you play is still irrelevant to *your* record.  If you end the season 18-2, then you are .900, whether your two losses are to Yale or Quinnipiac.  If they're to Yale, well, then that means you beat Q every time you played them - good on ya!  But the inconsistency of losing to Yale is still reflected in your .900.  Beating the best team and throwing clunkers to bad teams is seen as exactly as valuable as losing to the best teams and then beating everyone else.  It doesn't matter if your 18-2 record included other games against Q or Y - your record is still .900.

Where playing someone more times does matter is how your opponents do.  If the records are what they are, and you ask the question, is it better for Y or Q to take an extra loss, *then* the answer would be "whichever one we played fewer times."  When calculating our opponents' record, each opponent's record get included each time we play them.  So you always want the teams you've played to do well in their other games, and that goes even more strongly for the teams you play more times.

[billhoward]

Very uneven night.  Some guys looked tired.  I hope everyone is healthy.  Glad to see two EA goals.

Trying to recall how often, how recently we iced a game with a 175-foot empty netter. I'm pretty sure it wasn't at the Clarkson game.

[Trotsky]

We were thinking this could be a 1W 1L weekend. Better the L is today.

That isn't true for PWR.

Could you explain that please - is it because of the clause that prevents a team's RPI suffering if it beats a bad team?

Otherwise, I would not think it would matter much, since RPI has three criteria:

A team's own winning percentage (25%) - Cornell's winning percentage would be unaffected
The average of the team's opponents' winning percentages (21%) - I guess the boost to Princeton's winning% is less that the hit to Q's - is this where it matters?
The average of the team's opponents opponents' winning percentages (54%) - I can't believe this is significantly impacted by swapping the wins.

And in this case, a loss to Q will flip the NoDak pairwise comparison.

I am slavishly repeating what I have heard.  My intuition is they should be exactly the same impact provided we play both teams the same number of times.

I guess it is arguably more likely that by the end of Placid we will have played Q more.

Yes, but the number of times you play is still irrelevant to *your* record.  If you end the season 18-2, then you are .900, whether your two losses are to Yale or Quinnipiac.  If they're to Yale, well, then that means you beat Q every time you played them - good on ya!  But the inconsistency of losing to Yale is still reflected in your .900.  Beating the best team and throwing clunkers to bad teams is seen as exactly as valuable as losing to the best teams and then beating everyone else.  It doesn't matter if your 18-2 record included other games against Q or Y - your record is still .900.

Where playing someone more times does matter is how your opponents do.  If the records are what they are, and you ask the question, is it better for Y or Q to take an extra loss, *then* the answer would be "whichever one we played fewer times."  When calculating our opponents' record, each opponent's record get included each time we play them.  So you always want the teams you've played to do well in their other games, and that goes even more strongly for the teams you play more times.

How about this: since a loss brings Q's winning percentage down farther than Princeton's it is better to beat Princeton?  Any takers?

[BearLover]

Isn't there a bonus for beating teams in the top 20? In which case, beating Q and losing to P is more valuable than vice versa. I imagine the bonus is quite small, however.

[Dafatone]

Isn't there a bonus for beating teams in the top 20? In which case, beating Q and losing to P is more valuable than vice versa. I imagine the bonus is quite small, however.

I think they got rid of that? But I'm very uncertain.

[osorojo]

Consider the possibility that Princeton is simply a better hockey team. It appears it was last night, at least.

[nshapiro]

We were thinking this could be a 1W 1L weekend. Better the L is today.

That isn't true for PWR.

Could you explain that please - is it because of the clause that prevents a team's RPI suffering if it beats a bad team?

Otherwise, I would not think it would matter much, since RPI has three criteria:

A team's own winning percentage (25%) - Cornell's winning percentage would be unaffected
The average of the team's opponents' winning percentages (21%) - I guess the boost to Princeton's winning% is less that the hit to Q's - is this where it matters?
The average of the team's opponents opponents' winning percentages (54%) - I can't believe this is significantly impacted by swapping the wins.

And in this case, a loss to Q will flip the NoDak pairwise comparison.

I am slavishly repeating what I have heard.  My intuition is they should be exactly the same impact provided we play both teams the same number of times.

I guess it is arguably more likely that by the end of Placid we will have played Q more.

Yes, but the number of times you play is still irrelevant to *your* record.  If you end the season 18-2, then you are .900, whether your two losses are to Yale or Quinnipiac.  If they're to Yale, well, then that means you beat Q every time you played them - good on ya!  But the inconsistency of losing to Yale is still reflected in your .900.  Beating the best team and throwing clunkers to bad teams is seen as exactly as valuable as losing to the best teams and then beating everyone else.  It doesn't matter if your 18-2 record included other games against Q or Y - your record is still .900.

Where playing someone more times does matter is how your opponents do.  If the records are what they are, and you ask the question, is it better for Y or Q to take an extra loss, *then* the answer would be "whichever one we played fewer times."  When calculating our opponents' record, each opponent's record get included each time we play them.  So you always want the teams you've played to do well in their other games, and that goes even more strongly for the teams you play more times.

How about this: since a loss brings Q's winning percentage down farther than Princeton's it is better to beat Princeton?  Any takers?

I think so, given that a loss to Princeton and Win over Q impacts both those percentages more than the reverse.
Where I get completely lost is the *, which means that an RPI is adjusted to make sure that winning over a team with a bad record does not negatively affect your RPI.  Does this mean that if at the moment you beat a team, and your RPI would fall, that game is removed from your schedule?  Is that a permanent decision? If the team you beat goes on a winning streak, and adding that win back in would later help your RPI, does that happen?

[Robb]

We were thinking this could be a 1W 1L weekend. Better the L is today.

That isn't true for PWR.

Could you explain that please - is it because of the clause that prevents a team's RPI suffering if it beats a bad team?

Otherwise, I would not think it would matter much, since RPI has three criteria:

A team's own winning percentage (25%) - Cornell's winning percentage would be unaffected
The average of the team's opponents' winning percentages (21%) - I guess the boost to Princeton's winning% is less that the hit to Q's - is this where it matters?
The average of the team's opponents opponents' winning percentages (54%) - I can't believe this is significantly impacted by swapping the wins.

And in this case, a loss to Q will flip the NoDak pairwise comparison.

I am slavishly repeating what I have heard.  My intuition is they should be exactly the same impact provided we play both teams the same number of times.

I guess it is arguably more likely that by the end of Placid we will have played Q more.

Yes, but the number of times you play is still irrelevant to *your* record.  If you end the season 18-2, then you are .900, whether your two losses are to Yale or Quinnipiac.  If they're to Yale, well, then that means you beat Q every time you played them - good on ya!  But the inconsistency of losing to Yale is still reflected in your .900.  Beating the best team and throwing clunkers to bad teams is seen as exactly as valuable as losing to the best teams and then beating everyone else.  It doesn't matter if your 18-2 record included other games against Q or Y - your record is still .900.

Where playing someone more times does matter is how your opponents do.  If the records are what they are, and you ask the question, is it better for Y or Q to take an extra loss, *then* the answer would be "whichever one we played fewer times."  When calculating our opponents' record, each opponent's record get included each time we play them.  So you always want the teams you've played to do well in their other games, and that goes even more strongly for the teams you play more times.

How about this: since a loss brings Q's winning percentage down farther than Princeton's it is better to beat Princeton?  Any takers?

No, because games against you don't count when tallying your opponents' record in RPI.  If Q is undefeated against everyone else and you beat them like a borrowed mule 8 times, then you get credit for beating an undefeated team in each of those 8 games.