2024 Pairwise Predictor

[robertearle]

Thanks for this! Looks like one of my last refreshes didn't go through despite me updating the date. Try it now.

Wisconsin-Ohio State went from 75% OSU to about 60-40 OSU.

(Three wins in four game to three wins in five games. Did I need a whole app to get those percents?

 

[TonyTheTiger20]

it’s funny because “the math says it’s true” is being used to defend the bracket but in 2021, “the math is wrong,” justified Penn State’s exclusion. Is it possible the math is wrong again and OSU should at least be in a bracket with only 2 wcha teams?

Eh, 2021 was a pretty unique situation with no non-conference games. The math really just had no meaning there -- while now we at least have limited non-conference games to at least give us some idea of the differences between conferences, we had none back in 2021. So it was treating the CHA as being exactly as good as the WCHA. It would have done the same with the NEWHA if they were eligible at the time.

 

[robertearle]

Good catch. I started off by looking for KRACH on USCHO but couldn't find any there, although that may just be a personal shortcoming.

Regardless of whether Grant's KRACH is current or outdated, or any particular ordering, I would prefer a mathematical model that didn't require as many adjustments over the years as the PairWise does, or at least, a better formula to use as its backbone. When you have to ignore results because they are somehow misleading, it suggests that the basic model is flawed.

Except for a more-or-less tie between Clarkson and Colgate, KRACH is now pretty much in agreement with the Pairwise, FWIW.

 

[Lindsay]

Eh, 2021 was a pretty unique situation with no non-conference games. The math really just had no meaning there -- while now we at least have limited non-conference games to at least give us some idea of the differences between conferences, we had none back in 2021. So it was treating the CHA as being exactly as good as the WCHA. It would have done the same with the NEWHA if they were eligible at the time.

We’ll never agree about 2021. It’s ok.

But I’m curious what you think of this years bracket? Not just how right they got it but interesting stuff.

 

[BowWowWow]

Sometimes in the past couple years, there is an argument to be made that the rankings should really have the top 4 WCHA teams as the top 4 teams in the country. It's probably more even than that between the WCHA and ECAC, though, at least currently. UMD may deserve a higher Pairwise/NPI ranking or seed, but how high?

Here is the current inter-conference record comparison from this season; of course, it will change as a result of the national tournament. The trends are that CHA and ECAC are each doing somewhat better than last season, while Hockey East and NEWHA have trended down, and WCHA is fairly constant.

Women's Division I Hockey Inter-Conference Records: 2023-2024

vs.	CHA	ECAC	Hockey East	NEWHA	WCHA	OOC Record
CHA	—	9-29-3 (.256)	1-7-2 (.200)	13-0-0 (1.000)	3-12-0 (.200)	26-48-5 (.361)
ECAC	29-9-3 (.744)	—	26-8-2 (.750)	11-1-0 (.917)	5-9-3 (.382)	71-27-8 (.708)
Hockey East	7-1-2 (.800)	8-26-2 (.250)	—	13-0-0 (1.000)	0-8-0 (.000)	28-35-4 (.448)
NEWHA	0-13-0 (.000)	1-11-0 (.083)	0-13-0 (.000)	—	0-6-0 (.000)	1-43-0 (.023)
WCHA	12-3-0 (.800)	9-5-3 (.618)	8-0-0 (1.000)	6-0-0 (1.000)	—	35-8-3 (.793)

Here is the inter-conference record comparison from last season after the national tournament:
Women's Division I Hockey Inter-Conference Records: 2022-2023 (end of year)

vs.	CHA	ECAC	Hockey East	NEWHA	WCHA	OOC Record
CHA	—	8-32-0 (.200)	3-7-2 (.333)	13-2-0 (.867)	1-17-0 (.056)	25-58-2 (.306)
ECAC	32-8-0 (.800)	—	18-18-0 (.500)	13-4-1 (.750)	4-12-2 (.278)	67-42-3 (.612)
Hockey East	7-3-2 (.667)	18-18-0 (.500)	—	10-2-0 (.833)	0-5-0 (.000)	35-28-2 (.554)
NEWHA	2-13-0 (.133)	4-13-1 (.250)	2-10-0 (.167)	—	1-9-0 (.100)	9-45-1 (.173)
WCHA	17-1-0 (.944)	12-4-2 (.722)	5-0-0 (1.000)	9-1-0 (.900)	—	43-6-2 (.863)

Source: https://www.uscho.com/stats/interconference/division-i-women/

 

[ARM]

We’ll never agree about 2021. It’s ok.

Back in 2021, Grant still had more realistic hope that once afforded a chance, BC could do something with it.

No matter what was done in 2021, someone was going to be unhappy, myself included. There was no way that HEA deserved three teams in an eight-team field, but as Grant says, there wasn't any math available to help rank teams from other conferences relative to each other. And HEA had more committee members, so... It was almost impossible to know what to do with the four-team ECAC, with Clarkson and Colgate playing each other ... was it 10 times? No way to prove anything now, but my gut tells me:
Providence wasn't top 10, let alone top 8;
any of Penn State, Clarkson, Quinnipiac, St. Lawrence, and Minnesota might have won a game with a friendly matchup;
I don't consider either Wisconsin nor Ohio State from that season to be friendly matchups.

Regarding Badger Pete's suggestion, it would be fun to have those mini-tourney's the week before the conference tournaments, based off the current season rather than the previous (I get that wouldn't work for reasons like air travel, but given we're only dreaming, why let harsh realities get in the way?)

 

[QuasiCosmos]

Going back to previous comments about moving UMD to #6 and moving Cornell down to #8 in order to avoid the potential "first-game-played" OSU-UMD and Colgate-Cornell intra-conference quarterfinal games, and the impression some gave that UMD was under-rated in the rankings.

Looking over UMD's season record, I don't see why they should get an easier draw as #6 (and, against a team, Colgate, they already played twice this season – and beat and tied).

UMD was a combined 1-11-1 against the top three WCHA teams, and 1-1-2 against ECAC teams. Their loss and tie against Quinnipiac, which UMD just beat out for the last at-large slot, probably had a big impact on their Pairwise ranking below Cornell and St. Lawrence. Cornell was 3-1 and St. Lawrence was 1-0-1 against Quinnipiac.

Having said all this, a Colgate-UMD matchup in the quarterfinals would probably have been an exciting game to watch.

 

[Lindsay]

Going back to previous comments about moving UMD to #6 and moving Cornell down to #8 in order to avoid the potential "first-game-played" OSU-UMD and Colgate-Cornell intra-conference quarterfinal games, and the impression some gave that UMD was under-rated in the rankings.

Looking over UMD's season record, I don't see why they should get an easier draw as #6 (and, against a team, Colgate, they already played twice this season – and beat and tied).

UMD was a combined 1-11-1 against the top three WCHA teams, and 1-1-2 against ECAC teams. Their loss and tie against Quinnipiac, which UMD just beat out for the last at-large slot, probably had a big impact on their Pairwise ranking below Cornell and St. Lawrence. Cornell was 3-1 and St. Lawrence was 1-0-1 against Quinnipiac.

Having said all this, a Colgate-UMD matchup in the quarterfinals would probably have been an exciting game to watch.

good observations. Where I was getting hung up was with how OSU absolutely steamrolled SLU but potentially may face UMD who they’ve had closer games with. I look forward to seeing it play out.

 

[Lindsay]

No matter what was done in 2021, someone was going to be unhappy, myself included.

I don't consider either Wisconsin nor Ohio State from that season to be friendly matchups.
)

True I have no faith Penn State would have won a first round game that year, but I’ve always felt it matters to the players and can positively impact the program by just making it to the tournament for a first time.

considering Penn State started and ended their season by making history last year, only to kick off the off season by getting gutted in the portal, I may need to rethink this theory.

 

[PSUfan]

True I have no faith Penn State would have won a first round game that year, but I’ve always felt it matters to the players and can positively impact the program by just making it to the tournament for a first time.

considering Penn State started and ended their season by making history last year, only to kick off the off season by getting gutted in the portal, I may need to rethink this theory.

I agree with you Lindsay -- making the NCAA tournament for the first time last year absolutely had a positive impact on the program, both giving the current players confidence and influencing future recruiting. I've also seen the influence in the fanbase, home game attendance was up substantially this year after the conference tournament win and making the NCAA tournament last year. It would have made a difference if PSU had made the tournament in 2021 but I also understand the lack of out of conference competition that season made it very difficult to pick Penn State. And I will be holding my breath this year during the transfer portal window!

 

[Reddington]

PSU had Quinnipiac 2-1 with six left last year. If PSU had held onto that lead and won, would those two players have left? Who knows. I'm going to disagree with the tournament being the end all as far as attendance and recruiting . 20 win seasons put fannies in the seats and shows the program in a favorable light. Everyone likes to support or be a part of a winner. I'm just thrilled that the band was there for three home games, it means a lot to the atmosphere. It should be an interesting matchup with two future Olympians going at it with Gosling and Janecke.

The last three games with SLU over the last two years were 1-0 SLU. 1-1 tie. 4-2 PSU. PSU carried the play in all three games, but the scoreboard does not necessarily reflect that and the teams are majorly different this year.

​​​​​​As far as 2021 and the NCAA's, that's the year that a controversial formula was used that factored in prior years. Do you want to talk about unprecedented and of course the rebuttal is the year was unprecedented. It's all water under the bridge now.

 

[dave1381]

Thanks for the updated KRACH rankings, Tony -- that's about as much consensus as I've ever seen between KRACH and the actual rankings. Clarkson/Colgate is the only difference with a razor-thin margin in between.

Now to spoil any feeling of consensus, here are the Massey Ratings that have UMD at 5: https://masseyratings.com/chw/ncaa-d1/ratings
Those take into account margin of victory, so I guess UMD gets some credit for some of the tight losses vs OSU and Wisconsin and that 3-goal win vs Colgate.

Really, the right way to one a selection process is to define the principle you want, and then produce an objective system that delivers on all those principles. Indeed, what we have instead is an objective system for a subset of those principles that's been given all sorts of tweaks over the years, and is largely judged retrospectively by results in past years, and some other unwritten principles (e.g., I suspect, not too many .500-ish West teams ranked above East teams with shinier records) are used to judge which tweaks are accepted or not.

And as someone mentioned, yes, more East-West games would be good of course, but I gather there are budgetary / student-experience reasons why historically the modal number east-west trips for top teams is 1 (leading to 4 east-west games per team with 2 home and 2 road). But hey, if Stanford and Cal can join the ACC maybe the idea of spending like crazy on travel for women's hockey travel will become more acceptable.

 

[QuasiCosmos]

I was thinking about how the schedules for different leagues affect a fan's and a player's perception of a season, and how this might relate to the Pairwise rankings.

In the WCHA, each team plays the other seven league teams four times, leaving only six out-of-conference games during the thirty-four game regular season. Many WCHA teams play a two game series against three different out-of-conference teams; some do two game series against two teams and single games against two teams. Thus a WCHA team can play only ten or eleven different teams during the regular season. (Sadly, Ohio State got to play only nine different teams this year.)

Non-Ivy League ECAC teams play the other eleven league teams two times, leaving a dozen games for out-of-conference play. (Since the Ivy League limits the total number of games played / length of season, Ivy ECAC teams play five fewer out-of-conference games). This year, for example, Colgate played two game series against five out-of-conference teams and single games against two teams. Thus Colgate had the experience of playing eighteen different teams during the regular season. Cornell (Ivy League) played their seven non-conference games against five different out-of-conference teams, so sixteen different teams overall.

For the Colgate fans, this was marvelous. They got to see fifteen different teams in their home arena.

Colgate played zero NEWHA, one HE, three CHA, and three WCHA teams, along with the eleven ECAC teams. Cornell played zero, zero, three and two against these leagues.

This results in a fan of (or player on) an ECAC team having a significantly different "national" experience than a WCHA fan. The combination of playing only two games games against each of the other eleven league teams, and more out-of-league games accounts for the breadth of competition that ECAC fans and players experience. (I leave it to you all to consider how your leagues compare!)

I leave it to the experts on the Pairwise system to figure out how this significant range between the five leagues in the number of different teams played during the season impacts the Pairwise rankings of teams from the various leagues.

 

[Timothy A]

This results in a fan of (or player on) an ECAC team having a significantly different "national" experience than a WCHA fan. The combination of playing only two games games against each of the other eleven league teams, and more out-of-league games accounts for the breadth of competition that ECAC fans and players experience. (I leave it to you all to consider how your leagues compare!)

I wish the wcha could play more n/c games. They did after the Sue left and before the Tommies joined up. That was neat to experience that as a fan. I just don't think there is any way to not play each team twice in the same weekend with the distances involved. The conference schedule has to remain balanced.

 

[pgb-ohio]

I was thinking about how the schedules for different leagues affect a fan's and a player's perception of a season, and how this might relate to the Pairwise rankings.

In the WCHA, each team plays the other seven league teams four times, leaving only six out-of-conference games during the thirty-four game regular season. Many WCHA teams play a two game series against three different out-of-conference teams; some do two game series against two teams and single games against two teams. Thus a WCHA team can play only ten or eleven different teams during the regular season. (Sadly, Ohio State got to play only nine different teams this year.)

Non-Ivy League ECAC teams play the other eleven league teams two times, leaving a dozen games for out-of-conference play. (Since the Ivy League limits the total number of games played / length of season, Ivy ECAC teams play five fewer out-of-conference games). This year, for example, Colgate played two game series against five out-of-conference teams and single games against two teams. Thus Colgate had the experience of playing eighteen different teams during the regular season. Cornell (Ivy League) played their seven non-conference games against five different out-of-conference teams, so sixteen different teams overall.

For the Colgate fans, this was marvelous. They got to see fifteen different teams in their home arena.

Colgate played zero NEWHA, one HE, three CHA, and three WCHA teams, along with the eleven ECAC teams. Cornell played zero, zero, three and two against these leagues.

This results in a fan of (or player on) an ECAC team having a significantly different "national" experience than a WCHA fan. The combination of playing only two games games against each of the other eleven league teams, and more out-of-league games accounts for the breadth of competition that ECAC fans and players experience. (I leave it to you all to consider how your leagues compare!)

I leave it to the experts on the Pairwise system to figure out how this significant range between the five leagues in the number of different teams played during the season impacts the Pairwise rankings of teams from the various leagues.

Very insightful post. And yes, this does help explain why we in the WCHA crave more inter-conference play in the post-season. And of course frown upon conference tournament rehashes.

 

[BackyardIce]

And OSU played only 4 out of conference games. They played 32 games in their regular season ... 2 less than they are permitted to play.

 

[Hockeybuckeye]

And OSU played only 4 out of conference games. They played 32 games in their regular season ... 2 less than they are permitted to play.

I think a weekend series against either BC or BU would have been fun.

 

[Timothy A]

I think a weekend series against either BC or BU would have been fun.

I can attest to the fact that playing BC was a lot of fun.

 

[Hockeybuckeye]

I can attest to the fact that playing BC was a lot of fun.

Grant might think otherwise! LOL!

 

[TonyTheTiger20]

Lol hey we led in one of those games

BC has Wisconsin and Clarkson at home next year, pretty excited for that