An early look at 2019-2020

[adamw]

Yes, I know it doesn't necessarily help PWR (that's been made very clear to me), but wouldn't playing a better team help you be a better team, even if your Pairwise drops or stays flat with a loss or tie?  If the standard is winning uber alles than why not just play Mercyhurst, Army, and Canisius to make wins more likely.

Because that's not the only standard. Sure, as you said, playing better teams helps you get better. That's two separate discussions. The Pairwise part is simply factual. And just playing Mercyhurst, et al only helps your Pairwise IF you win them ALL. You darn well better.

Basically, there is an inverse ratio between Strength of Schedule and Expected Win Percentage.  Seems obvious, but people are constantly trying to game the system.  There is no way to game it, really. Not in hockey.  You play harder teams, you may win less - you play weaker teams, you may win more.  Either way, Pairwise is the same.  But - if you win a couple more vs. those harder teams - or you lose a couple vs. those weaker teams - it all blows up.  There's just no way to know in advance - so scheduling to game the Pairwise is useless.

In theory, you're right that the PWR perfectly accounts for the greater difficulty of harder opponents by weighting wins proportionally more and losses proportionally less, and vice versa for weaker opponents. But in practice, do we really have a sense of whether that weighting is correct? PWR is basically RPI, which CHN says is computed as follows:
(1) A team's own winning percentage (25%)
(2) The average of the team's opponents' winning percentages (21%)
(3) The average of the team's opponents opponents' winning percentages (54%)

Do we actually have a good sense for whether these numbers are calibrated properly? I.e., is this truly the ideal ratio of (1), (2), and (3) for determining how good a team is? If the breakdown were 35/16/49, for instance, it would be more beneficial than it is now to schedule easier opponents. And the opposite would be true if the breakdown were 15/26/59. These are just hypothetical numbers; the the point is: what makes the 25/21/54 breakdown "right"? How did we arrive there? It's highly possible the numbers are off, and the ideal ratio of (1), (2) and (3) for evaluating teams is something else, such that under the current RPI it is beneficial to schedule easier opponents, or harder opponents.

The same goes for the home/away weighting, as Dafatone alluded to. I recall seeing somewhere that teams receive a 1.2x boost for an away win, and likewise suffer only .8x as much of an RPI hit for an away loss. Are these 1.2/.8 numbers based on a real study of the benefit of home-ice advantage, or just conjecture? Before we treat the PWR/RPI as the be-all-end-all, we should probably confirm the formulas on which they are based.

I agree that these numbers are somewhat arbitrary - which is why KRACH is better. But I can tell you that they are tested to find a reasonable equilibrium. It's basically fudging by hand until it makes sense. And they've been changed over the years, for a variety of reasons.

[jtwcornell91]

Yes, I know it doesn't necessarily help PWR (that's been made very clear to me), but wouldn't playing a better team help you be a better team, even if your Pairwise drops or stays flat with a loss or tie?  If the standard is winning uber alles than why not just play Mercyhurst, Army, and Canisius to make wins more likely.

Because that's not the only standard. Sure, as you said, playing better teams helps you get better. That's two separate discussions. The Pairwise part is simply factual. And just playing Mercyhurst, et al only helps your Pairwise IF you win them ALL. You darn well better.

Basically, there is an inverse ratio between Strength of Schedule and Expected Win Percentage.  Seems obvious, but people are constantly trying to game the system.  There is no way to game it, really. Not in hockey.  You play harder teams, you may win less - you play weaker teams, you may win more.  Either way, Pairwise is the same.  But - if you win a couple more vs. those harder teams - or you lose a couple vs. those weaker teams - it all blows up.  There's just no way to know in advance - so scheduling to game the Pairwise is useless.

In theory, you're right that the PWR perfectly accounts for the greater difficulty of harder opponents by weighting wins proportionally more and losses proportionally less, and vice versa for weaker opponents. But in practice, do we really have a sense of whether that weighting is correct? PWR is basically RPI, which CHN says is computed as follows:
(1) A team's own winning percentage (25%)
(2) The average of the team's opponents' winning percentages (21%)
(3) The average of the team's opponents opponents' winning percentages (54%)

Do we actually have a good sense for whether these numbers are calibrated properly? I.e., is this truly the ideal ratio of (1), (2), and (3) for determining how good a team is? If the breakdown were 35/16/49, for instance, it would be more beneficial than it is now to schedule easier opponents. And the opposite would be true if the breakdown were 15/26/59. These are just hypothetical numbers; the the point is: what makes the 25/21/54 breakdown "right"? How did we arrive there? It's highly possible the numbers are off, and the ideal ratio of (1), (2) and (3) for evaluating teams is something else, such that under the current RPI it is beneficial to schedule easier opponents, or harder opponents.

The same goes for the home/away weighting, as Dafatone alluded to. I recall seeing somewhere that teams receive a 1.2x boost for an away win, and likewise suffer only .8x as much of an RPI hit for an away loss. Are these 1.2/.8 numbers based on a real study of the benefit of home-ice advantage, or just conjecture? Before we treat the PWR/RPI as the be-all-end-all, we should probably confirm the formulas on which they are based.

I agree that these numbers are somewhat arbitrary - which is why KRACH is better. But I can tell you that they are tested to find a reasonable equilibrium. It's basically fudging by hand until it makes sense. And they've been changed over the years, for a variety of reasons.

Back when the weightings were 25-50-25 or 35-50-15, a good way to game the system (perhaps unintentionally) was to play a bunch of opponents who themselves had good records against weak schedules.  That way, your opponents' winning percentage is high, even though your opponents themselves may not be so good.  This is basically how Quinnipiac and Niagara got overrated in RPI, and therefore PWR, back in the early 00s.  The problem is that while a team's RPI has a big strength of schedule component, RPI's assessment of that strength of schedule was almost all down to the winning percentage of opponents, not considering *their* strength of schedule very much (especially when the weights were changed to 35-50-15 to reduce the chance of dropping in RPI by beating a good team).  The strength of an opponent was something like 67-33-00 or 77-23-00 in the old system.  (50/(50+25)=.67 and 50/(50+15)=.77).  With the new weights, however, the breakdown of record and schedule in evaluating an opponent is 28-72-00 (21/(21+54)=.28), which is a log closer to the 25% record and 75% strength of schedule used in RPI itself.

See the discussion/definition of "RPI Strength" at http://elynah.com/tbrw/tbrw.cgi?rpi versus http://elynah.com/tbrw/tbrw.cgi?2006/rpi and http://elynah.com/tbrw/tbrw.cgi?2002/rpi

[billhoward]

If Cornell didn't have a crazy home tie to QU, a crazy home OT loss to Colgate and RPI, a crazy tie at Brown, and two bad home losses to Michigan State - they'd have been a No. 1 seed.

And then we might have run into Providence in the first game.

The other purpose of playing good teams is maybe you'll beat some of them. Doing that mid-fall when your opponent has played a half-dozen games, is a challenge.

[upprdeck]

and maybe play them in game 1 before so many got hurt and the ice sucked

[Trotsky]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    Angello 13, Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


[Jim Hyla]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

[Trotsky]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

[arugula]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

Great as Niuewendyk was, those were such radically different times for scoring.  Just apples and oranges.  Not uncommon then to see two points a game players, as Joe was the following two years.

[nshapiro]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

Great as Niuewendyk was, those were such radically different times for scoring.  Just apples and oranges.  Not uncommon then to see two points a game players, as Joe was the following two years.

I am !

[BearLover]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

Well, all those other freshmen in double-digits had many more assists than Regush (3).

[Trotsky]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

Great as Niuewendyk was, those were such radically different times for scoring.  Just apples and oranges.  Not uncommon then to see two points a game players, as Joe was the following two years.

Fair enough.  In 1978 as a junior, Lance Nethery had 23 goals in 26 games.

Oh, and 60 assists.

And then there's this.

[arugula]

Wow, that's nuts.  Interesting too is the difference between NC$$ in the 70s versus the 80s as an NHL feeder.  Nethery was even more dominant than Niuewendyk, yet had a very minor NHL career.

[Trotsky]

Wow, that's nuts.  Interesting too is the difference between NC$$ in the 70s versus the 80s as an NHL feeder.  Nethery was even more dominant than Niuewendyk, yet had a very minor NHL career.

There was very much a stigma about COLLEGE BOY! in the 70s in the NHL.  I mean, that was almost as bad as being a homosexual.

Don Cherry was making personnel decisions in the 70s.  Christ, Eddie Shore wasn't dead yet.

[redice]

Wow, that's nuts.  Interesting too is the difference between NC$$ in the 70s versus the 80s as an NHL feeder.  Nethery was even more dominant than Niuewendyk, yet had a very minor NHL career.

There was very much a stigma about COLLEGE BOY! in the 70s in the NHL.  I mean, that was almost as bad as being a homosexual.

Don Cherry was making personnel decisions in the 70s.  Christ, Eddie Shore wasn't dead yet.

Lance had everything that an NHL'er needed except great skating skills.  If he could have refined that skill, he would have had agood NHL career.

[ugarte]

Progression of Cornell GF and 10+ scorers:

Code Select Expand


2015    57    Hilbrich 10
2016    79    Angello 11, Kubiak 10
2017    99    Vanderlaan 15, Angello 12, Yates 12
2018   102    [b]Angello 13[/b], Yates 13, Rauter 11
2019   108    Barron 15, Donaldson 12, Regush 12, Vanderlaan 11


If only.........

Speaking of... most goals by a freshman:

Code Select Expand

2000    8    Matt McRae
2001    7    Vesce
2002    4    Knoepfli
[b]2003   13    Moulson[/b]
2004    6    Carefoot
2005    5    Scott
2006    3    Kennedy
[b]2007   11    Greening
2008   12    R. Nash[/b]
2009    3    Collins, Ross
2010    4    D'Agostino, Esposito
2011    3    Mowrey
2012    8    Ferlin
2013    2    Hilbrich, Knisley
2014    4    Buckles
2015    2    Bliss, Tschantz
[b]2016   11    Angello[/b]
2017    6    Malott
2018    5    Barron, Betts
[b]2019   12    Regush[/b]

On a team without Barron we'd all be insanely high on Regush.

For fun comparison:

Code Select Expand

[b]1985   21    Nieuwendyk[/b]
in 29 games

Great as Niuewendyk was, those were such radically different times for scoring.  Just apples and oranges.  Not uncommon then to see two points a game players, as Joe was the following two years.

yeah nieuwy was probably overrated