John t whelan ranking simulator

[The Exiled One]

Re: John t whelan ranking simulator

And the tournament field:

North Dakota

Did they get rid of the Wisconsin rule? If not, North Dakota doesn't currently qualify for an at-large bid.

[Numbers]

Re: John t whelan ranking simulator

GoBlue,

I have another question about this KRACH with ties and home/away. The mathematical formulation uses exp(k1+h) for example. I assume that is because it makes the math come out - because you can take the logs and then it's addition rather than multiplication. However, it seems to me that the k(i) that comes out of that calculation is not the k(i) that comes out of Whelan's example of how to calculate KRACH here and on CHN.
It seems to me that the Ki all need to re-exponentiated to get the numbers that are equivalent to the KRACH ratings in Whelan's formulation.

Is that correct?

Thanks again.

[FlagDUDE08]

Re: John t whelan ranking simulator

Did they get rid of the Wisconsin rule? If not, North Dakota doesn't currently qualify for an at-large bid.

The what rule?

[Numbers]

Re: John t whelan ranking simulator

The what rule?

Wisconsin rule. That's the one prohibiting teams with less that .500 records from an at-large place in the tournament.

[FlagDUDE08]

Re: John t whelan ranking simulator

Wisconsin rule. That's the one prohibiting teams with less that .500 records from an at-large place in the tournament.

Does autobids have any impact on that? Alabama Huntsville made the tournament a few years ago with a sub-500 record, but they had the CHA autobid. Also, I seem to remember about 6 or 7 years ago UVM had a team that almost didn't make the Hockey East playoffs but ended making some noise in the national tournament.

[The Exiled One]

Re: John t whelan ranking simulator

Does autobids have any impact on that? Alabama Huntsville made the tournament a few years ago with a sub-500 record, but they had the CHA autobid. Also, I seem to remember about 6 or 7 years ago UVM had a team that almost didn't make the Hockey East playoffs but ended making some noise in the national tournament.

No, it doesn't affect autobids, just at-large bids. Wisconsin earned an at-large bid in 07-08 with a record of 15-16-7. They did end up beating Denver 6-2 in their first NCAA tournament game, but lost to North Dakota in overtime the next game. The "Wisconsin rule" was implemented the year after.

[FlagDUDE08]

Re: John t whelan ranking simulator

No, it doesn't affect autobids, just at-large bids. Wisconsin earned an at-large bid in 07-08 with a record of 15-16-7. They did end up beating Denver 6-2 in their first NCAA tournament game, but lost to North Dakota in overtime the next game. The "Wisconsin rule" was implemented the year after.

If that's still the case, New Hampshire would be the replacement. It does show that it's very early in the season, though.

[goblue78]

Re: John t whelan ranking simulator

exp(ki) is the KRACH rating, ie, ki for team i is a number which needs to exponentiated to be in Whelan's form. You don't need to exponentiate again. h, the home ice advantage, also needs to be exponentiated to make patman's formula pr(home win) = ca/(ca+b) correct. It's generally easier to estimate ki instead of Ki=exp(ki) because the constraint that Ki>0 is a pain. when working with ki, ie ln(Ki) the coefficient is unconstrained.

Note all of the Ks are still only determined up to a multiplicative constant, so you can define the absolute value of the KRACH ratings any way you want. Some common methods are to make the best team 100, to make the sum 1, or, in an understandable if somewhat idiosyncratic view at siouxsports.com, to make North Dakota = 100.

(Neither h nor theta can be scaled. They have to be takes as is.)

[Numbers]

Re: John t whelan ranking simulator

exp(ki) is the KRACH rating, ie, ki for team i is a number which needs to exponentiated to be in Whelan's form. You don't need to exponentiate again. h, the home ice advantage, also needs to be exponentiated to make patman's formula pr(home win) = ca/(ca+b) correct. It's generally easier to estimate ki instead of Ki=exp(ki) because the constraint that Ki>0 is a pain. when working with ki, ie ln(Ki) the coefficient is unconstrained.

Note all of the Ks are still only determined up to a multiplicative constant, so you can define the absolute value of the KRACH ratings any way you want. Some common methods are to make the best team 100, to make the sum 1, or, in an understandable if somewhat idiosyncratic view at siouxsports.com, to make North Dakota = 100.

(Neither h nor theta can be scaled. They have to be takes as is.)

Thanks Go,

That's what I meant. I appreciate all the explaining.

[FlagDUDE08]

Re: John t whelan ranking simulator

As of games completed 22 November 2013, 11:59 Aleutian Time:

5.00 Quinnipiac
4.75 Minnesota
4.50 Michigan
4.25 Ferris State
4.00 St. Cloud State
3.75 Boston College
3.50 LSSU
3.25 Providence
3.00 Yale
2.75 Cornell
2.50 Wisconsin
2.25 Bowling Green
2.00 Notre Dame
1.75 New Hampshire
1.50 Clarkson
1.25 UMASS Lowell
1.00 Miami
0.75 Northeastern
0.50 Northern Michigan
0.25 Union

And the tournament field:

Minnesota
Quinnipiac
Michigan
Ferris State

Boston College
St. Cloud State
LSSU
Providence

Wisconsin
Yale
New Hampshire
Cornell

Notre Dame
Bowling Green
UMASS Lowell
AHA Champ (36 - Mercyhurst)

[FlagDUDE08]

Re: John t whelan ranking simulator

As of games completed 23 November 2013, 11:59 Aleutian Time:

5.00 Minnesota
4.75 Michigan
4.50 Ferris State
4.25 Providence
4.00 St. Cloud State
3.75 Quinnipiac
3.50 Boston College
3.25 Cornell
3.00 LSSU
2.75 Minnesota State Mankato
2.50 Clarkson
2.25 UMASS Lowell
2.00 Yale
1.75 Bowling Green
1.50 Wisconsin
1.25 Notre Dame
1.00 New Hampshire
0.75 Miami
0.50 Union
0.25 Northeastern

And the tournament field:

Minnesota
Michigan
Providence
Ferris State

Quinnipiac
St. Cloud State
Boston College
Cornell

LSSU
UMASS Lowell
Clarkson
Bowling Green

New Hampshire
Wisconsin
Yale
AHA Champ (37 - Air Force)

NOTE: Minnesota State Mankato was removed from 10th overall due to their sub-500 record.

[Patman]

Re: John t whelan ranking simulator

I would take that regional in a heart beat.

[FlagDUDE08]

Re: John t whelan ranking simulator

I would take that regional in a heart beat.

You mean the 9-12 band regional?

[QUAlum2004]

Re: John t whelan ranking simulator

The killer of a loss against Princeton loses 1.25 points. Hopefully a win against Providence Wednesday makes that difference up.

[FlagDUDE08]

Re: John t whelan ranking simulator

The killer of a loss against Princeton loses 1.25 points. Hopefully a win against Providence Wednesday makes that difference up.

That's just Quality Wins Bonus, not your RatingsPI. Your RatingsPI is currently at 59.60 with removals of Bentley, both Holy Cross games, Dartmouth, and the Princeton game from Friday. Beating Providence alone would take you to 60.58, but that may change based upon other happenings today and Tuesday.