2015-2016 USCHO Posters Poll

[pokechecker]

Re: 2015-2016 USCHO Posters Poll

I would be kind of surprised to see either UND or BSU make it,

it would be pointless if they did, they'd be going to Madison for possibly the 6th meeting between the two teams.

you should put your math to work to show how the ECAC gets an advantage with the current way teams are selected
the math is pretty easy to understand why it's not gonna change
ECAC has more teams, they are not going to give up their advantage

I do like the NE - MN matchup though, that would be an interesting game

 

[TonyTheTiger20]

Re: 2015-2016 USCHO Posters Poll

it would be pointless if they did, they'd be going to Madison for possibly the 6th meeting between the two teams.

I actually think if UND made it they would go to UM not matter what. UW is a flight for them -- which yeah basically supports your point.

I do like the NE - MN matchup though, that would be an interesting game

Yeah I've been really hoping to see that one.

 

[TonyTheTiger20]

Re: 2015-2016 USCHO Posters Poll

you should put your math to work to show how the ECAC gets an advantage with the current way teams are selected
the math is pretty easy to understand why it's not gonna change
ECAC has more teams, they are not going to give up their advantage

EDIT: This is long and mathy. Sorry. Pokey asked me for math...

The PWR puts more of an emphasis than, say, KRACH, for not losing games, regardless of who you lose to.

Take a look at the ratios that RPI (which for all intents and purposes sets 95% of the Pairwise Rankings):

30% for winning percentage
24% for opponents' winning percentage
46% for opponents' opponents' percentages.

Now look at the range of possible values in the RPI calculations for those three factors (columns J, K, and L here):

Range of winning percentages: 0.000 to 1.000 (because you will either win, lose, or tie the game)
Difference: 1.000

Range of opp. winning percentages: 0.375 to 0.628
Difference: 0.254

Range of opp opp winning percentages: 0.467 to 0.540
Difference: 0.074 (!!)

The .3/.24/.46 ratios have the right idea by putting more emphasis on the one with the smallest range, but one could argue that it might not be doing enough, honestly. If you're taking those differences, the most each value can affect your RPI is

30% x 1.000 = 0.300
24% x 0.254 = 0.061
46% x 0.074 = 0.034

And that's the most it can affect the RPI (max minus min).

So even though the RPI is trying to give some weighting to strength of schedule, it's not giving very much. Winning percentage has a much, much bigger effect on the ranking than the other two do, even combined.

There are two ways you could fix this problem by adjusting the weighting factors. Either (1) making all three affect the RPI the same amount, or (2) by making winning percentage affect it half and then make the "strength of schedule" portion affect it half.

Let's pick better factors for option #1 first:

(skipping the math part for brevity)

Win% = 5.4% x 1.000 = 0.054
OW% = 21.3% x 0.254 = 0.054
OOW% = 73.3% x 0.074 = 0.054

So, with those three factors, all three will affect RPI the same amount. Plugging it into my handy dandy spreadsheet gives the following RPI (did not include QWB):

1) UW .6190
2) UM .5977
3) BC .5958
3) BSU .5831
5) QU .5573
6) ND .5701
7) PU .5586
8) CU .5525
9) UMD .5578
10) HU .5508

Looking at option two, making the SOS portion count as much as winning percentage:

Win% = 10.3% x 1.000 = 0.103
OW% = 20.2% x 0.254 = 0.0515
OOW% = 69.5% x 0.074 = 0.0515

Using those factors, the resulting RPI would be:

1) UW .6715
2) BC .6415
3) UM .6342
4) QU .5998
5) BSU .6131
6) ND .5986
7) CU .5926
8) PU .5923
9) NU .5801
10) HU .5791

Are either of those "more right" than one another? I don't know. Are they more right than the factors used right now? I still don't know. The fact that there are so many ways to justify using different seemingly arbitrary combinations of weighting factors kind of speaks to why RPI is broken in the first place.

But to summarize -- as it stands right now, because winning percentage is such an important part of your RPI, it really penalizes you for losing.

The factors used in women's hockey aren't even the same as men's hockey, let alone the "standard" factors. The factors used for other sports like basketball are 25%/50%/25%. Needless to say, the rankings would be much different with those factors as well.

 

[Timothy A]

Re: 2015-2016 USCHO Posters Poll

Yep. North Dakota actually has a bit of ground to make up.

UND 2 MSU 1 last night......I think I'm over-valuing UND due to their success against UM and UW, that has to mean something, right? Statistically, they are just another BSU, great goaltending, great team D, little scoring talent. My eyes told me last weekend they are speed-wise equal to UW. They are just lacking scoring talent, though all teams but UW and UM have that issue in the WCHA. I may have to rethink my 4-6 ranking, it's just so hard to discount what I see compared to the eastern box scores. Maybe there's some truth to this whole math thing afterall. I still don't like math.

 

[pokechecker]

Re: 2015-2016 USCHO Posters Poll

EDIT: This is long and mathy.....

you are lacking a conclusion

It looks to me like your math supports my thesis, but it would be nice to hear you say it

 

[pokechecker]

Re: 2015-2016 USCHO Posters Poll

since WI has completed the schedule against BSU & UND, that probably accounts for them have the tougher SOS at this points, at least I hope that is why because when you look at the non conference schedule MN has played the tougher non conference schedule
PSU vs. Lindenwood
Yale vs. Dartmouth
SCSU vs. Providence
assuming there is not a change in these teams fortunes, I would expect MN to have the higher SOS at the end of the season

 

[TonyTheTiger20]

Re: 2015-2016 USCHO Posters Poll

you are lacking a conclusion

It looks to me like your math supports my thesis, but it would be nice to hear you say it

Your thesis is on the right track, but it's not really that the ECAC is "bigger"; It's more to do with the fact that RPI/PWR doesn't really reward a tough strength of schedule because losing (regardless of who you lose to) hurts so much.

I guess if you're trying to say that because the WCHA is smaller, and the other teams have to play UM and UW a total of 8 times, then that hurts those teams -- then in that case, yeah, your point is valid, your point is just more of a corollary to the "it doesn't reward strength of schedule" reason rather than the reason itself.

 

[pokechecker]

Re: 2015-2016 USCHO Posters Poll

SOS is a result of the teams you play, by having a big conference you don't play the top teams as often as if the conference is smaller
this is a big advantage for the ECAC
in the past there were 3 top teams in the WCHA, and not top 10, top 4, 5, or 6
now there are 4, what other conference can say that? that's 12 games, against top teams, the 3rd & 4th place team are doing well if they win half those games, but it totally destroys them from the criteria for NCAA tourney selection
I'm saying that the large number of teams in the ECAC pretty much guarantees them 3 teams in the tournament while pretty much guaranteeing that the other three conferences can have at most 2.


and if that weren't enough, the reward for making the tournament for a third WCHA team is yet another game against MN

 

[pokechecker]

Re: 2015-2016 USCHO Posters Poll

in addition, people look at UND with a suspecting eye for losing to Syracuse
but how about Clarkson & Quinnipiac?
and to top it off, the best teams they play all year is each other
twice
hell, that's a weekend in the WCHA

 

[Eeyore]

Re: 2015-2016 USCHO Posters Poll

SOS is a result of the teams you play, by having a big conference you don't play the top teams as often as if the conference is smaller this is a big advantage for the ECAC

You also don't play the bottom teams as often. There is no inherent bias to strength of schedule based upon the size of a conference. Your in-conference strength of schedule will be the average of the quality of the other teams in your league, so long as you play a balanced schedule. It doesn't matter how many of them there are. The only way that the size of the conference matters in terms of how many teams it gets into the NCAA tournament is that the ECAC has one third of the total number of teams in NCAA Div 1, while the WCHA has less than a quarter, so the former has more chances to have teams that meet the qualification criteria.

The difference for strength of schedule is that the average quality of the teams in the WCHA is higher than in the ECAC. Adding more teams wouldn't necessarily make the WCHA SoS any lower; adding more bad teams would.

Edit: Let me amend this a bit. If you have two conferences with equal average team strengths, you will get some difference in each team's in-conference SoS, but it isn't consistent and, for teams like UND and BSU, cuts the other way from your claim. The third place team in the smaller conference will have a lower SoS than the third place team in the larger conference. That derives entirely from the fact that teams don't play themselves, so the team in the smaller conference removes a higher percentage of games against above average competition than the one in the larger conference. For below average teams, this is reversed and the ones in the smaller conference will have a higher SoS.

 

[vicb]

Re: 2015-2016 USCHO Posters Poll

SOS is a result of the teams you play, by having a big conference you don't play the top teams as often as if the conference is smaller
this is a big advantage for the ECAC
in the past there were 3 top teams in the WCHA, and not top 10, top 4, 5, or 6
now there are 4, what other conference can say that? that's 12 games, against top teams, the 3rd & 4th place team are doing well if they win half those games, but it totally destroys them from the criteria for NCAA tourney selection
I'm saying that the large number of teams in the ECAC pretty much guarantees them 3 teams in the tournament while pretty much guaranteeing that the other three conferences can have at most 2.


and if that weren't enough, the reward for making the tournament for a third WCHA team is yet another game against MN

May be the WCHA should expand to a 12 team league and then the whining from you all will stop . It is not the ECAC's fault that 2 of the top teams in the nation are in the WCHA and the 3rd and 4th best teams have to play them 4 or 6 times per season. I am wondering when the 2 Michigan schools will finally wake up and have women's teams. Then the B1G can ruin the WCHA like they did to the men by having their own league. As a matter of fact, instead then North Dakota and Bemidji will have much better shots at making the Tourney with the WCHA auto bid not residing in Minny or Wisco. Then the ECAC and HE can whine about 3 or 4 Western teams plus the CHA champ leaving only 2 or 3 spots for their teams.

 

[ne7minder]

UND 2 MSU 1 last night......I think I'm over-valuing UND due to their success against UM and UW, that has to mean something, right? Statistically, they are just another BSU, great goaltending, great team D, little scoring talent. My eyes told me last weekend they are speed-wise equal to UW. They are just lacking scoring talent, though all teams but UW and UM have that issue in the WCHA. I may have to rethink my 4-6 ranking, it's just so hard to discount what I see compared to the eastern box scores. Maybe there's some truth to this whole math thing afterall. I still don't like math.

We saw UND v BSU this year and they must have thought they were playing a mirror. It looked like UND played less physically than against MN and BSU kept up. Both need snipers

 

[PuckRush]

Re: 2015-2016 USCHO Posters Poll

in addition, people look at UND with a suspecting eye for losing to Syracuse
but how about Clarkson & Quinnipiac?
and to top it off, the best teams they play all year is each other
twice
hell, that's a weekend in the WCHA

I'm not buying - there is a ton of depth in the ECAC. Quinnipiac and Clarkson have the best records, and nipping at their heels are some real tough teams: Princeton, Colgate and Harvard. Then you've got perennially tough out St Lawrence, who can beat any of these teams on any given night. RPI, Dartmouth, Yale and Cornell - play your best or you'll tie or lose. Summary: You've got 10 teams in the ECAC that would give BSU, ND, MD, Ohio St, St Cloud and Minnesota State fits every weekend.

If I were buying, I'd look to Hockey East. BC is the cream, Northeastern the challenger, and everyone else is CHA-like right now.

 

[Eeyore]

Re: 2015-2016 USCHO Posters Poll

RPI, Dartmouth, Yale and Cornell - play your best or you'll tie or lose.

Those teams are a combined 0-6 against the WCHA this year.

 

[PuckRush]

Re: 2015-2016 USCHO Posters Poll

Those teams are a combined 0-6 against the WCHA this year.

Oh, the beauty of statistics. These 4 teams are NOT 0-6 vs the WCHA. They are 0-6 vs #3 Minnesota, #2 Wisconsin and #8 North Dakota.

Let's do the equivalent. Send St Cloud, Ohio State, and Minn St against #4 Quinny, #5 Clarkson and #9 Princeton. The WCHA will be 0-6.

The WCHA is the deepest conference in the country, and I'm the only fool who had UMD in the top 10 over the past two weeks. But please don't tell me the ECAC is weak. Anyone who says that isn't watching the games.

 

[ARM]

Re: 2015-2016 USCHO Posters Poll

Let's do the equivalent. Send St Cloud, Ohio State, and Minn St against #4 Quinny, #5 Clarkson and #9 Princeton. The WCHA will be 0-6.

I agree with your point that the ECAC is a deep conference, but SCSU, OSU, and MSU aren't the equivalent of RPI, Dartmouth, Yale, and Cornell. MSU is the equivalent of Union and/or Brown, in terms of being at the very bottom, so you'd have to include those results in the mix. MSU may not be as good as the top 10 of the ECAC, but it is better than the bottom two. The ECAC is deeper, in part because it has more teams. It also has more teams at the same level. As good as Clarkson is, there is a wide range of ECAC opponents against which the Golden Knights are vulnerable. They are like a UND equivalent, with a bit more skill.

I think a person could craft an argument that the ECAC is better than the WCHA. That could be true or not based on how one rates certain factors. However, it would be tough to put forth a convincing argument that the CHA or Hockey East is better than either the ECAC or the WCHA. I think the WCHA/ECAC are nearly equal, and I'd give a slight edge to the WCHA based on its top quarter (two teams) being stronger than the collective top three in the ECAC (also a quarter). Others could look at a different slice of teams and reach a different conclusion. Hockey East is third, and CHA is last, primarily because it lacks a power team entirely.

 

[PuckRush]

Re: 2015-2016 USCHO Posters Poll

I agree with your point that the ECAC is a deep conference, but SCSU, OSU, and MSU aren't the equivalent of RPI, Dartmouth, Yale, and Cornell. MSU is the equivalent of Union and/or Brown, in terms of being at the very bottom, so you'd have to include those results in the mix. MSU may not be as good as the top 10 of the ECAC, but it is better than the bottom two. The ECAC is deeper, in part because it has more teams. It also has more teams at the same level. As good as Clarkson is, there is a wide range of ECAC opponents against which the Golden Knights are vulnerable. They are like a UND equivalent, with a bit more skill.

I think a person could craft an argument that the ECAC is better than the WCHA. That could be true or not based on how one rates certain factors. However, it would be tough to put forth a convincing argument that the CHA or Hockey East is better than either the ECAC or the WCHA. I think the WCHA/ECAC are nearly equal, and I'd give a slight edge to the WCHA based on its top quarter (two teams) being stronger than the collective top three in the ECAC (also a quarter). Others could look at a different slice of teams and reach a different conclusion. Hockey East is third, and CHA is last, primarily because it lacks a power team entirely.

ARM... this is a perfect post. You have a real good sense of how all the teams stack up. I think what is most impressive is your analysis of MSU v Brown v Union, because you really have to be paying attention to women's hockey to speak to that (and yes, MSU is better than those two).

Just for fun, here's how I think a combined WCHA and ECAC conference might look like in the standings right now:

1. Wisconsin
2. Minnesota
3. Quinnipiac
4. Bemidji
5. North Dakota
6. Clarkson
7. Princeton
8. Harvard
9. Colgate
10. Minn-Duluth
11. St. Lawrence
12. St. Cloud
13. Yale
14. Ohio State
15. RPI
16. Cornell
17. Dartmouth
18. Minn State
19. Brown
20. Union

 

[brookyone]

Re: 2015-2016 USCHO Posters Poll

ARM... this is a perfect post. You have a real good sense of how all the teams stack up. I think what is most impressive is your analysis of MSU v Brown v Union, because you really have to be paying attention to women's hockey to speak to that (and yes, MSU is better than those two).

Just for fun, here's how I think a combined WCHA and ECAC conference might look like in the standings right now:

1. Wisconsin
2. Minnesota
3. Quinnipiac
4. Bemidji
5. North Dakota
6. Clarkson
7. Princeton
8. Harvard
9. Colgate
10. Minn-Duluth
11. St. Lawrence
12. St. Cloud
13. Yale
14. Ohio State
15. RPI
16. Cornell
17. Dartmouth
18. Minn State
19. Brown
20. Union

This looks good. I might flip Clarkson & NoDak.

 

[TonyTheTiger20]

Re: 2015-2016 USCHO Posters Poll

Nice posts ... Yeah that doesn't look too far off

 

[vicb]

Re: 2015-2016 USCHO Posters Poll

1. Boston College
2. Wisconsin
3. Minnesota
4. Quinnipiac
5. Bemidji
6. Northeastern
7. Clarkson
8. North Dakota
9. Princeton
10. Harvard (Only because they have Maschmeyer in goal. Only Clarkson and Minn Dul have scored more than 2 goals in a game on Harvard)
10.1 Colgate