Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of letters

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Forgive my ignorance as I'm new to how the ncaa playoff format works. I'm wondering as to how these teams are picked and the playoffs haven't even started yet. If a Princeton or Clarkson or Cornell in the ECAC lose in the playoffs, am I understanding right that it doesn't mean squat. They would go to the final 8 and the team that upset them wouldn't. If this is the case, why have playoffs? Im confused!! I would appreciate any information on this please and thank you.

Hello!

So, 8 teams make the tournament. Those 8 teams consist of:

--The WCHA tournament champion
--The Hockey East tournament champion
--The ECAC tournament champion
--The CHA tournament champion
--The four teams ranked highest in the Pairwise Rankings that did not win their conference championship.

So, to answer this hypothetical:

If a Princeton or Clarkson or Cornell in the ECAC lose in the playoffs, am I understanding right that it doesn't mean squat.

If Princeton or Clarkson or Cornell lose in the playoffs, but they are still one of the 4 highest ranked teams that didn't win their conference tournament, they would be in. But if they aren't, they would be out. And if someone like, say, St. Lawrence, wins the ECAC, then St. Lawrence would be in the tournament.

So right now the Pairwise is:

1 Wisconsin
2 Minnesota
3 Northeastern
4 Clarkson
5 Princeton
6 Cornell
7 Boston College
8 Ohio State

Ohio State wouldn't make the tournament as it stands today, because someone has to win the CHA championship, and they would knock Ohio State out.

If St. Lawrence wins the ECAC tournament, then SLU would be in and that would knock out Boston College.
If Minnesota-Duluth also wins the WCHA tournament, then SLU and UMD would be in and BC and Cornell would be out.
If Vermont also wins the Hockey East tournament, then SLU and UMD and UVM would be in and BC and Cornell and Princeton would be out.

Think of it this way -- The top 7 teams in the Pairwise are expected to make the tournament, plus the CHA winner. Every "upset champion" in the other three conferences reduces that "7" by one. In order to absolutely lock yourself into the NCAA tournament, you need to either win your conference tournament or be in the top 4. Otherwise you're at risk of being knocked out.

That's a quick and dirty crash course -- hopefully that wasn't too much!

 

[robertearle]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

It's a little wild this year. BC is locked into losing the comparison to UMD and there are a couple other flipped comparisons too. More so than usual anyway.

I miss the "TUC cliff"... NOT!!

Good job on corrected RPIs and 'bracketology'.

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

I miss the "TUC cliff"... NOT!!

Good job on corrected RPIs and 'bracketology'.

My least favorite thing about this year was that since BC and UMD don't share any common opponents, BC was locked into losing the Duluth comparison after week 1 (!!!!), even if BC went undefeated the rest of the way and UMD went winless every game the rest of the way. lol

 

[99hockey]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Thank you Tony and Robert. Makes sense now. I appreciate it

 

[99hockey]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Thank you Tony and Robert. Makes sense now. I appreciate it

 

[ARM]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

... if someone random wins the WCHA, which, again, lol nope...

This was always "the way things are" in the WCHA, but IMO, the times they are a changing. From memory, last year's Gophers were the first team seeded No. 3 or lower to win win the WCHA Tournament. The challenge in the past was that it meant beating two teams ranked in the top 8 or higher on back-to-back days. Often those were top-five teams, and even when the third team was also a power team, it was tough to do.

Last year, Minnesota wasn't that good, but it played one very good weekend of hockey. A lower seed winning the tourney was long overdue. Back in 2001, Wisconsin took UMD to OT in one semi and top-seeded Minnesota got blown out in the other.

As good as Wisconsin is this year, we know that the Buckeyes can beat them. Can they at a neutral site? Maybe. I think that OSU, UMD, and BSU are all capable of defeating the Gophers at home in a one-and-done format. It could be that both favored teams get upset in the semis, and nobody has to knock off both No. 1 and No. 2. So I'd say that there is at least a 10-percent chance that a "random" team wins the WCHA autobid.

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

So I'd say that there is at least a 10-percent chance that a "random" team wins the WCHA autobid.

That's not TOO far off, you're pretty much within an order of magnitude here -- according to KRACH (and I'm not really sure I know what the WCHA is doing with 7 teams... first round bye for #1? With everyone else playing a 3 game series? That's what I input anyway), as it stands, these are the WCHA tournament odds:

Wisconsin 52.01%
Minnesota 43.97%
Ohio State 2.32%
Minnesota Duluth 1.47%
Bemidji State 0.12%
Minnesota State 0.11%
BYE: 0.00%
St. Cloud State 0.00% (lol -- mostly due to the fact that they have to take 2 out of 3 from Minnesota)

That feels about right. Minnesota and Wisconsin are just so far ahead of everyone else...

 

[robertearle]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

(and I'm not really sure I know what the WCHA is doing with 7 teams... first round bye for #1? With everyone else playing a 3 game series? That's what I input anyway)

Correct

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Correct

Thanks!

 

[ARM]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

That's not TOO far off, you're pretty much within an order of magnitude here -- according to KRACH (and I'm not really sure I know what the WCHA is doing with 7 teams... first round bye for #1? With everyone else playing a 3 game series? That's what I input anyway), as it stands, these are the WCHA tournament odds:

Wisconsin 52.01%
Minnesota 43.97%
Ohio State 2.32%
Minnesota Duluth 1.47%
Bemidji State 0.12%
Minnesota State 0.11%
BYE: 0.00%
St. Cloud State 0.00% (lol -- mostly due to the fact that they have to take 2 out of 3 from Minnesota)

That feels about right. Minnesota and Wisconsin are just so far ahead of everyone else...

The one piece that we're missing right now is that once OSU, UMD, BSU, or whomever advances, its KRACH rating will likely rise more than that of UW/UM in advancing over BYE and No. 7. But yeah, I mostly pulled a number out of thin air.

In other news, Lock now shows Minnesota over Wisconsin, in part because Minnesota is 3-1 and UW is 1-1 against the common opponent that both lost to, and also because his system looks at O & D separately, so margin matters:
http://it.stlawu.edu/~chodr/wchodr/current.html

 

[ChickHicks86]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Unlike Grant's Bracketology, which is based on what's actually going on in women's college hockey, I decided to imagine what the postseason selection would look like in an early 21st Century college football world. For no other reason that I thought it would be interesting (and I have no desire to do anything outside until the temperature goes above 0), I put together a BCS rankings system for women's college hockey.

For anybody not familiar with the BCS (or have already erased it from your memory), it is not a statistical calculation, but a ranking system based on both human and computer polls. Without getting too far off track (you can Google "BCS ranking formula" for much more detail), the BCS system assigns a rating out of 1 based on the percentage of the voting points a team received. These ratings are then averaged for 3 polls (Coaches, Harris, computer for football | USCHO, USA Today/Hockey, computer for hockey) to determine the BCS rating. As an example, in the USA Today poll this week, Wisconsin earned all 19 1st place votes for a total of 190/190, or a rating of 1, and Boston College earned 75/190 for a rating of 0.3947, and so on. The computer "poll" is taken as if the "voters" were unique computer ratings system voting for their Top 10 teams. For my calculations, I used RPI (begrudgingly), KRACH, WCHODR, and GRaNT. If that explanation makes no sense (as things I write have a tendency to do), please let me know if you have questions!

So without further ado, I present the BCS rankings for women's college hockey:

Rank | USCHO | USA | Computer | BCS Rating | Team Name
1 | 0.9667 | 1.0000 | 0.9750 | 0.9806 | Wisconsin
2 | 0.9333 | 0.9000 | 0.9250 | 0.9194 | Minnesota
3 | 0.8000 | 0.8000 | 0.8000 | 0.8000 | Northeastern
4 | 0.6733 | 0.6684 | 0.6750 | 0.6723 | Princeton
5 | 0.6000 | 0.6000 | 0.5750 | 0.5917 | Clarkson
6 | 0.4867 | 0.5263 | 0.4000 | 0.4710 | Cornell
7 | 0.3733 | 0.3947 | 0.4750 | 0.4144 | Boston College
8 | 0.2467 | 0.2579 | 0.3250 | 0.2765 | Boston University
9 | 0.2000 | 0.2000 | 0.1750 | 0.1917 | Ohio State
10 | 0.1800 | 0.1316 | 0.0250 | 0.1122 | Colgate
11 | 0.0133 | 0.1263 | 0.0250 | 0.0549 | Merrimack
12 | 0.0133 | 0.0053 | 0.0750 | 0.0312 | Minnesota Duluth
13 | 0.0133 | 0.0053 | 0.0500 | 0.0229 | Providence
(All other teams have a BCS rating of 0)

 

[pokechecker]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Unlike Grant's Bracketology, which is based on what's actually going on in women's college hockey, I decided to imagine what the postseason selection would look like in an early 21st Century college football world. For no other reason that I thought it would be interesting (and I have no desire to do anything outside until the temperature goes above 0), I put together a BCS rankings system for women's college hockey.

For anybody not familiar with the BCS (or have already erased it from your memory), it is not a statistical calculation, but a ranking system based on both human and computer polls. Without getting too far off track (you can Google "BCS ranking formula" for much more detail), the BCS system assigns a rating out of 1 based on the percentage of the voting points a team received. These ratings are then averaged for 3 polls (Coaches, Harris, computer for football | USCHO, USA Today/Hockey, computer for hockey) to determine the BCS rating. As an example, in the USA Today poll this week, Wisconsin earned all 19 1st place votes for a total of 190/190, or a rating of 1, and Boston College earned 75/190 for a rating of 0.3947, and so on. The computer "poll" is taken as if the "voters" were unique computer ratings system voting for their Top 10 teams. For my calculations, I used RPI (begrudgingly), KRACH, WCHODR, and GRaNT. If that explanation makes no sense (as things I write have a tendency to do), please let me know if you have questions!

So without further ado, I present the BCS rankings for women's college hockey:

Rank | USCHO | USA | Computer | BCS Rating | Team Name
1 | 0.9667 | 1.0000 | 0.9750 | 0.9806 | Wisconsin
2 | 0.9333 | 0.9000 | 0.9250 | 0.9194 | Minnesota
3 | 0.8000 | 0.8000 | 0.8000 | 0.8000 | Northeastern
4 | 0.6733 | 0.6684 | 0.6750 | 0.6723 | Princeton
5 | 0.6000 | 0.6000 | 0.5750 | 0.5917 | Clarkson
6 | 0.4867 | 0.5263 | 0.4000 | 0.4710 | Cornell
7 | 0.3733 | 0.3947 | 0.4750 | 0.4144 | Boston College
8 | 0.2467 | 0.2579 | 0.3250 | 0.2765 | Boston University
9 | 0.2000 | 0.2000 | 0.1750 | 0.1917 | Ohio State
10 | 0.1800 | 0.1316 | 0.0250 | 0.1122 | Colgate
11 | 0.0133 | 0.1263 | 0.0250 | 0.0549 | Merrimack
12 | 0.0133 | 0.0053 | 0.0750 | 0.0312 | Minnesota Duluth
13 | 0.0133 | 0.0053 | 0.0500 | 0.0229 | Providence
(All other teams have a BCS rating of 0)

I did something similar a few years back to determine ranking I would submit for the USCHO Pollsters Poll.
Back then I used KRACH, RPI, USA Today, Coaches Poll, and I think USCHO poll. WCHODR, and GRaNT did not exist.
Just like the more people voting in a poll, the more ranking systems you use to arrive at the ranking, the more accurate it becomes. (ignore eeyores objection to this which will surely be wrong)

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

SB Nation has a poll as well: https://www.theicegarden.com/platfo...e-hockey-division-i-poll-continuous-consensus

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

So BC fell back to 9th despite winning last night, and it was because BU flipped the BC/BU comparison back into their favor:

https://www.bcinterruption.com/bost...ey-nukes-holy-cross-crusaders-final-score-6-0

<img src="https://i.imgur.com/5lGrRvV.png" width="600"

Before last night, BC was -1 on H2H, +1 on RPI, and tied on CoOpp, leaving the comparison tied and putting the tiebreaker on RPI, giving BC the comparison and jumping OSU and BU, where they "should be" (so to speak) based on RPI.

BC beat Holy Cross, who they've already beaten, so their common opponent's record with BU was unchanged (their 1.000 winning percentage against Holy Cross stayed at 1.000).

BU, on the other hand, beat Vermont, a team they tied earlier in the season, improving their winning percentage against Vermont from 0.750 to 0.833. That un-tied Common Opponents with BC, giving BU the comparison and dropping BC back below BU and Ohio State.

Common Opponents is on a razor's edge between these two, but the only reason it's close right now is thanks to BC's 0-1 (0.000) record against UNH, and BU's 1-0 (1.000) record against UNH. Both teams still have two against UNH coming up, where BC should improve their 0.000 and BU cannot improve upon their 1.000 (and in fact can only hurt it).

BC doesn't have any scheduled with BU the rest of the way, but if the two teams meet in the Beanpot finals (or the WHEA tournament), BC can force the comparison back to "RPI Only" by evening out the Head To Head comparison 2-2. At that point, Common Opponents won't matter, because at best it would tie RPI, and RPI itself is the tiebreaker.

Anyway, there's no real point to this post except to give an interesting example of how the Pairwise can be weird sometimes.

 

[TonyTheTiger20]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

If NU hadn't lost to Holy Cross, they'd be #1 in the Pairwise. Could you imagine?

 

[robertearle]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

If NU hadn't lost to Holy Cross, they'd be #1 in the Pairwise. Could you imagine?

Grant, I was looking at how close Wisconsin and Minnesota are in RPI - USCHO has then tied beyond four digits; you have them tied through three digits - and it brought to mind a fairly idle question (now, maybe not in a month). How much of a 'quality win bonus cliff' is there?

Let's say that a win by Minnesota over Duluth tomorrow were to drop Duluth below the 'QWB' range. Obviously, Minn gets a boost to their RPI from the win. But how much of that boost would be negated by the loss of QWB for the win as well as for their earlier win over Duluth?

No, you don't need to do any calculations; a gut-feel guess will suffice.

Thanks.

 

[ARM]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

but how much of that boost would be negated by the loss of qwb for the win as well as for their earlier wins over duluth?

Fixed, as the two teams are 1-1-1 in OT, and UM won the first meeting in regulation.

 

[TonyTheTiger20]

Grant, I was looking at how close Wisconsin and Minnesota are in RPI - USCHO has then tied beyond four digits; you have them tied through three digits - and it brought to mind a fairly idle question (now, maybe not in a month). How much of a 'quality win bonus cliff' is there?

Let's say that a win by Minnesota over Duluth tomorrow were to drop Duluth below the 'QWB' range. Obviously, Minn gets a boost to their RPI from the win. But how much of that boost would be negated by the loss of QWB for the win as well as for their earlier win over Duluth?

No, you don't need to do any calculations; a gut-feel guess will suffice.

Thanks.

It isn't really much of a cliff, because unlike TUC, the value of a QWB is relative to how good that team is. So beating the team in 13th is worth 0.0000 extra, 12th is worth 0.0050, 11th is worth 0.0100, 10th is worth 0.0150 ... Etc. (Note that's not added to your actual RPI but averaged out over your full schedule.) In the end, the value of the win over that team is worth significantly more that the value lost by that team dropping a couple spots (or out) of the QWB.

However, if in the end Wisconsin has more wins than Minnesota over UMD, and the RPI is close enough, USCHO's rankings could incorrectly have Minnesota #1 and Wisconsin #2. Because with St. Anselm in the rankings, those wins over UMD would be undercounted. USCHO has UMD as 12th in the RPI (with St. Anselm 1st) when they are really 11th. St. Anselm is ineligible this year and shouldn't be counted.

So while it's unlikely to fall into place like that, just keep in mind that USCHO's Pairwise is wrong, and BCI's is right, for that one reason.

Sorry, that was probably nonsense to most of you!

EDIT: Forgot to take into account the fact that UM has the 1.5 fewer wins over UMD. So yes, USCHO's PWR is undercounting those UMD QWB points for Wisconsin and it could conceivably flip the two teams. Made some clarifying edits above.

 

[robertearle]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Fixed, as the two teams are 1-1-1 in OT, and UM won the first meeting in regulation.

Yes, I did forget that the tie brings with it a "half-win" for QWB purposes.

Really not that important for the purpose of the question, is it?

 

[robertearle]

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Re: Fun With Numbers 2019: PWR, RPI, KRACH, GRaNT, WCHODR, & other assortments of let

Thanks, Grant. I sort-of figured that, but thought asking was worth it, anyway.