TonyTheTiger20
#SOAR
Pretty sure last year BC was like top 5 or so. We played every single team in the top 7 other than ourselves.
I know that doesn't automatically make us have a highly ranked SOS but I'm fairly sure I remember that being the case.
USCHO Fan Forum
This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
-
The USCHO Fan Forum has migrated to a new plaform, xenForo. Most of the function of the forum should work in familiar ways. Please note that you can switch between light and dark modes by clicking on the gear icon in the upper right of the main menu bar. We are hoping that this new platform will prove to be faster and more reliable. Please feel free to explore its features.
2013-2014 Women's D-I PairWise Contentions and Affirmations
- Thread starter brookyone
- Start date
I know that doesn't automatically make us have a highly ranked SOS but I'm fairly sure I remember that being the case.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Also -- the SOS ranking in KRACH just makes no sense. ALL 8 WCHA teams in the top 9?? The only thing I can figure is that Minnesota and Wisconsin's wildly extreme KRACH ratings are skewing the SOS...
I freely admit that I have no basis for complaint here other than "there's just no way that can be right."
Also -- the SOS ranking in KRACH just makes no sense. ALL 8 WCHA teams in the top 9?? The only thing I can figure is that Minnesota and Wisconsin's wildly extreme KRACH ratings are skewing the SOS...
I freely admit that I have no basis for complaint here other than "there's just no way that can be right."
Last edited:
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
So do you then agree with RPI SOS which only has 4 WCHA teams in the top ten (well, 6 if we discount SH and SA)?
So do you then agree with RPI SOS which only has 4 WCHA teams in the top ten (well, 6 if we discount SH and SA)?
ManBehindTheCurtain
Great and powerful wizard
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Even you in your rankings for your posters poll have three WCHA teams in the top 7. YOU, yes, YOU rank Minnesota and Wisconsin 1 and 2. That means that BY YOUR RANKING every single team in the WCHA has played 4 games against the top 2, 6 games against teams from the top 7.
So what exactly is it that you think SOS is measuring if it is not whether or not you have played games against the strongest teams?
Geez Grant. Do the math for crying out loud. Or maybe just think a little bit before you reflex post.
Even you in your rankings for your posters poll have three WCHA teams in the top 7. YOU, yes, YOU rank Minnesota and Wisconsin 1 and 2. That means that BY YOUR RANKING every single team in the WCHA has played 4 games against the top 2, 6 games against teams from the top 7.
So what exactly is it that you think SOS is measuring if it is not whether or not you have played games against the strongest teams?
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
To get to 6 you have to discard Sacred Heart, St. Anselm, and Holy Cross as Wisconsin is #13. However, to make a meaningful comparison to KRACH you have to exclude them, so that's valid. Also note that with the exclusions the WCHA holds positions 1-4 in RPI strength of schedule. In addition, 2 of the other 4 teams in the top ten are Yale and Princeton, who played nonconference series against Minnesota. Cornell and St. Lawrence are the only teams among the top ten SOS who have not played the Gophers.
So the basic answer is that Minnesota is skewing strength of schedule pretty heavily in all of the ratings systems. Minnesota State and Bemidji State are #11 and #12.
To get to 6 you have to discard Sacred Heart, St. Anselm, and Holy Cross as Wisconsin is #13. However, to make a meaningful comparison to KRACH you have to exclude them, so that's valid. Also note that with the exclusions the WCHA holds positions 1-4 in RPI strength of schedule. In addition, 2 of the other 4 teams in the top ten are Yale and Princeton, who played nonconference series against Minnesota. Cornell and St. Lawrence are the only teams among the top ten SOS who have not played the Gophers.
So the basic answer is that Minnesota is skewing strength of schedule pretty heavily in all of the ratings systems. Minnesota State and Bemidji State are #11 and #12.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
As far as my comment on Wisconsin and Minnesota skewing it more than others -- the teams out west all play, for example, Lindenwood a couple times a year.
The median KRACH rating is 80 or so.
Lindenwood's KRACH is 60 below the median.
Minnesota's is 2,764 above the median.
So, playing Minnesota is going to dramatically improve your SOS more than playing Lindenwood is going to hurt it. That's what I meant.
Can't you guys see why I make comments about western fans? Whenever one of you perceives some tiny, tiny slight against the west you freak out about it.
You have been dominating this thread with quality posts, by the way.
Mostly I agree with whichever is right -- which leads me to point to KRACH. I don't have a problem with it, it's just so extreme that it seems bizarre.
Whoa hey, settle down. My point was mostly just that it was such a high concentration that it was worthy of note. I did say that I had no rational reason to complain about it.
As far as my comment on Wisconsin and Minnesota skewing it more than others -- the teams out west all play, for example, Lindenwood a couple times a year.
The median KRACH rating is 80 or so.
Lindenwood's KRACH is 60 below the median.
Minnesota's is 2,764 above the median.
So, playing Minnesota is going to dramatically improve your SOS more than playing Lindenwood is going to hurt it. That's what I meant.
Can't you guys see why I make comments about western fans? Whenever one of you perceives some tiny, tiny slight against the west you freak out about it.
Well, that really tells the story right there.
You have been dominating this thread with quality posts, by the way.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
One other thing I noticed:
One other thing I noticed:
It would appear, then, that KRACH *can* be distorted by teams with weak records (that is, winless) against strong competition. No?
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Yes, but not by as much as you are implying. KRACH is based upon ratios, not differences. So the relevant figure is that Minnesota is about 35 times the median while Lindenwood is about 1/4 the median. And this shouldn't be surprising, because Lindenwood is much closer to a .500 round robin winning percentage than Minnesota is; they aren't as bad as Minnesota is good. Note that the mean KRACH rating is higher than the median.
Yes, but not by as much as you are implying. KRACH is based upon ratios, not differences. So the relevant figure is that Minnesota is about 35 times the median while Lindenwood is about 1/4 the median. And this shouldn't be surprising, because Lindenwood is much closer to a .500 round robin winning percentage than Minnesota is; they aren't as bad as Minnesota is good. Note that the mean KRACH rating is higher than the median.
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
What do you mean by distorted? If you mean that a team that has no wins and no ties at all will skew the ratings, yes, this is true. That is the flip side of Minnesota skewing the ratings last year. As I said, KRACH does not handle either infinity or zero well. However, since it is very rare for you to find a team that has neither won nor tied a game by the time you reach the second half of the season it isn't usually an issue.
If this is meant as an extension about the effect that Lindenwood is having right now, then, no, it doesn't get distorted. Since KRACH works by the ratios of teams' ratings rather than the differences between those ratings, the effect isn't what you seemed to think it is. What the current ratings are telling us is that Minnesota would beat the median team (using Minnesota State as that team) 35 times as often as it loses, while the median team would beat Lindenwood about 4 times as often as it loses. You can argue about whether those figures are right but they don't look obviously crazy.
What do you mean by distorted? If you mean that a team that has no wins and no ties at all will skew the ratings, yes, this is true. That is the flip side of Minnesota skewing the ratings last year. As I said, KRACH does not handle either infinity or zero well. However, since it is very rare for you to find a team that has neither won nor tied a game by the time you reach the second half of the season it isn't usually an issue.
If this is meant as an extension about the effect that Lindenwood is having right now, then, no, it doesn't get distorted. Since KRACH works by the ratios of teams' ratings rather than the differences between those ratings, the effect isn't what you seemed to think it is. What the current ratings are telling us is that Minnesota would beat the median team (using Minnesota State as that team) 35 times as often as it loses, while the median team would beat Lindenwood about 4 times as often as it loses. You can argue about whether those figures are right but they don't look obviously crazy.
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Just having fun with you, Grant.
Sounds like a complaint to me. Just saying.
Well, you obviously fooled MBTC.
(emphasis added by me) So you do have a complaint.
Sounds like you're backtracking now. Tsk, tsk, tsk.
Just having fun with you, Grant.
OnMAA
Vort Doen.....Heija Heija Heija
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Everything appears to be tilted into extremes these days, both the numbers and the discussions.
Everything appears to be tilted into extremes these days, both the numbers and the discussions.Welcome to my experience.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Heh... let me clarify -- I was complaining, but acknowledging that I had no logical reason to.Just having fun with you, Grant.
ARM
Fan of chipmunk-like mascots.
You poor dears.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
That is, Minnesota's KRACH is 35x better than that of the median KRACH, and Brown's is only 6x worse. So it would appear to me, then, that losing to Minnesota helps you out in the rankings a lot more than actually winning against Brown would. And disproportionately so.
It was even more fun last year when Minnesota's KRACH was in scientific notation because it was so high But as someone mentioned earlier, KRACH doesn't handle 0/infinity well.
I think if you see Minnesota get 1 or 2 more losses before the season is out that those WCHA SOSs will drop substantially.
EDIT: And I guess by extension I would say that the effect is reversed (if someone muted) in men's with Army being 14x worse than the median and Minnesota (Jesus, enough Minnesota!!) just being 5x better than the median.
Well that's not really what I meant, I meant that it appears that a team that goes winless against Minnesota will get a disproportionate boost in the rankings because their SOS would go through the roof -- that is, a loss against the best team would help a ton more than a win against the worst team would hurt.
That is, Minnesota's KRACH is 35x better than that of the median KRACH, and Brown's is only 6x worse. So it would appear to me, then, that losing to Minnesota helps you out in the rankings a lot more than actually winning against Brown would. And disproportionately so.
It was even more fun last year when Minnesota's KRACH was in scientific notation because it was so high
But as someone mentioned earlier, KRACH doesn't handle 0/infinity well.I think if you see Minnesota get 1 or 2 more losses before the season is out that those WCHA SOSs will drop substantially.
EDIT: And I guess by extension I would say that the effect is reversed (if someone muted) in men's with Army being 14x worse than the median and Minnesota (Jesus, enough Minnesota!!) just being 5x better than the median.
Last edited:
ARM
Fan of chipmunk-like mascots.
Losing to anyone, even a very strong team, doesn't help. It just hurts much less than a loss usually does.
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
No. This is not how KRACH works. A loss will NEVER help your rating. A win will NEVER hurt it. For god's sake, read the explanation of it; it's not that hard.
No. This is not how KRACH works. A loss will NEVER help your rating. A win will NEVER hurt it. For god's sake, read the explanation of it; it's not that hard.
. The earlier comment was not intended for you. Quite the opposite. Enjoy our dialog's.
TonyTheTiger20
#SOAR
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
Seriously. It doesn't say "wins always help and losses always hurt" in the explanation. But if that's something I didn't know about the KRACH that you do (you're obviously very knowledgeable about these things and your contributions here have actually been really informative), then thanks for sharing it. I just don't think you need to be so insulting about it to someone who is less knowledgeable than you.
As someone who clearly understands KRACH better than me, I want to thank for you for helping me improve my knowledge of it it in such a friendly, kind, not-at-all self-important and arrogant way. *Thumbs up*
Seriously. It doesn't say "wins always help and losses always hurt" in the explanation. But if that's something I didn't know about the KRACH that you do (you're obviously very knowledgeable about these things and your contributions here have actually been really informative), then thanks for sharing it. I just don't think you need to be so insulting about it to someone who is less knowledgeable than you.
Like this. This was nice.
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
I apologize. The explanation on USCHO used to say this and it also had a link to a more detailed explanation. Apparently USCHO decided that having people understand things was a bad idea. Try College Hockey News' explanation or Ken Butler's more mathematical take.
I apologize. The explanation on USCHO used to say this and it also had a link to a more detailed explanation. Apparently USCHO decided that having people understand things was a bad idea. Try College Hockey News' explanation or Ken Butler's more mathematical take.
Re: 2013-2014 Women's D-I PairWise Contentions and Affirmations
I'll try to explain some of the basics of KRACH. I'm going to do so in a way that I think gets the conceptual point across though it may end up somewhat misleading as to the actual mathematical calculations used, so don't quote it if you want to talk about the math.
The underlying principle is that every team has a rating and that you can calculate the implied probability of each team winning. (I'm going to ignore ties because essentially that's what KRACH does; a tie is considered half of a win and half of a loss and just added to those totals.)
You start out giving each team the same rating; I think USCHO's calculation uses 100 as the opening rating but it doesn't really matter what value you use; what is important is the ratios of the ratings and you'll end up with the same ratios no matter what starting value you use.
You then look at every game that has been played to date; this process of evaluating games rather than season records is one of the things that makes KRACH very different from RPI. You calculate the probability of either team winning each game; obviously in the first calculation every team will have a 0.5 probability of winning each game since the ratings are the same. Conceptually what happens is that each win increases a team's rating by an amount proportional to the difference between 1 and their probability to win the game. Conversely, every loss decreases a team's rating by the difference between their probability of winning and 0. (This is where I'm bending the math out of recognition; this isn't the actual process but the effect is pretty much the same.)
Once you've run every game played through this process you have a new set of ratings for every team. These will not be correct, or at least the likelihood that they are correct is vanishingly small. So you go through the whole process again to get a new set of ratings. And so on. You stop when the new set of ratings is identical to the previous set to whatever degree of precision you desire; USCHO uses four decimal places.
So a win never hurts your rating and a loss never helps because you don't have a "strength of schedule" component in the calculations; the rankings of SoS are derived after the ratings are determined, not before. If you beat a bad team, it just means that the difference between 1 and your probability of winning is very small, so there is little advantage in the ratings. Conversely, the system expected you to lose to Minnesota so it doesn't lower your rating very much when you do.
I'll try to explain some of the basics of KRACH. I'm going to do so in a way that I think gets the conceptual point across though it may end up somewhat misleading as to the actual mathematical calculations used, so don't quote it if you want to talk about the math.
The underlying principle is that every team has a rating and that you can calculate the implied probability of each team winning. (I'm going to ignore ties because essentially that's what KRACH does; a tie is considered half of a win and half of a loss and just added to those totals.)
You start out giving each team the same rating; I think USCHO's calculation uses 100 as the opening rating but it doesn't really matter what value you use; what is important is the ratios of the ratings and you'll end up with the same ratios no matter what starting value you use.
You then look at every game that has been played to date; this process of evaluating games rather than season records is one of the things that makes KRACH very different from RPI. You calculate the probability of either team winning each game; obviously in the first calculation every team will have a 0.5 probability of winning each game since the ratings are the same. Conceptually what happens is that each win increases a team's rating by an amount proportional to the difference between 1 and their probability to win the game. Conversely, every loss decreases a team's rating by the difference between their probability of winning and 0. (This is where I'm bending the math out of recognition; this isn't the actual process but the effect is pretty much the same.)
Once you've run every game played through this process you have a new set of ratings for every team. These will not be correct, or at least the likelihood that they are correct is vanishingly small. So you go through the whole process again to get a new set of ratings. And so on. You stop when the new set of ratings is identical to the previous set to whatever degree of precision you desire; USCHO uses four decimal places.
So a win never hurts your rating and a loss never helps because you don't have a "strength of schedule" component in the calculations; the rankings of SoS are derived after the ratings are determined, not before. If you beat a bad team, it just means that the difference between 1 and your probability of winning is very small, so there is little advantage in the ratings. Conversely, the system expected you to lose to Minnesota so it doesn't lower your rating very much when you do.