2015 Pairwise Discussion & Predictions

[dave1381]

Re: 2015 Pairwise Discussion & Predictions

I hate so much the argument "there is no perfect system." Why not use a better system if it exists?

If a system discriminates against the same team year after year, a team that's consistently third and fourth in the same conference, that's simply not fair. Sure that team can always do better, but they deserve better.

And I agree there's some margin at which it's nice to have tournament rather than relying on selection criterion. At the same time you want to earn your way into the tournament. College football clearly does better to have a four-team tournament then to have a selection process pick two teams. For people (including me) who thought OSU didn't belong, it was about them having the worst performance of any of the top 5 teams. It had nothing to do with whether they were capable of winning the tournament. Surely there was a case they were one of the 4 best teams. But that committee has been quite inconsistent about taking the 4 best teams vs. who deserves to go. But anyway... clearly it was good for women's hockey to expand to 8 back in 2005. Maybe we're almost at the stage where more teams would be nice but NCAA tends to keep the tournament share in the 15-30% range and 12 would give us more than 1/3 of the major conference teams. So be it.

 

[ARM]

Re: 2015 Pairwise Discussion & Predictions

Good points. I also think there can be two different discussions. One is making sure that the tournament includes the teams best suited to winning the tournament, while the other is including the teams whose performance over the season is most deserving of a tournament berth just for the sake of competing in the tournament. While OSU proved that a fourth (or lower) team by the criteria was capable of winning the football tournament, I haven't really seen any indication that a team that ranks around the bottom of the top eight or lower is likely to win the women's hockey NCAAs. Had UND gotten by Minnesota in 2013, I think it had as good a chance as any team that I remember, but that's all hypothetical.

 

[ne7minder]

Re: 2015 Pairwise Discussion & Predictions

[I know somebody is going to finish my sentence by saying "and then they both got smashed to a pulp by Minnesota and Wisconsin" but I've beaten you to the punch!]

There is that other part of me that thinks not having a champion crowned might be a better deal. A lot of teams could look back on a season and say it was a great season if it were not of the last game they played. I know that is not how humans are put together & fans least of all but it would have been nice so many times to just be happy with a really really good season.

 

[Eeyore]

Re: 2015 Pairwise Discussion & Predictions

I hate so much the argument "there is no perfect system." Why not use a better system if it exists?

Maybe I should explain what I mean. I absolutely agree that if there is a system that would better accomplish the goals of the tournament, use it. It would be much better to use a system based upon KRACH than the one we have based on RPI. I can state that unequivocally because those two methods are attempting to measure the same thing and KRACH measures it better.

When I say that there is no perfect system, I mean two things. The first is that the criteria for deciding who makes the field aren't objective. Different people are going to have different priorities. Some will think that games at the end pf the season should be weighted more heavily than those at the beginning. Some people will think that the field should be large enough to encompass any team that would have a half decent shot should be included; some will think that it should be small and only give teams that had an outstanding regular season entry; some will think that it should only include teams that won their conference because anyone else has already demonstrated that they are not the best team (and this would split between those who think winning a conference should be determined by the regular season and those who think it should be the playoff winner).

Despite the fact that I have strong feelings about some of those decisions, none of them are objectively right or wrong. There is no system that is going to satisfy everyone and no system that really ought to be thought of as perfect.

The second thing is that no system of measuring is going to be perfect. There are going to be times, especially when teams are tightly bunched by the criteria like they are this year, when the result doesn't come out like everyone thinks it should even if they have the same priorities for the criteria. It won't work just right all the time. We should minimize the number of times that it is wrong and the magnitude of error, but it's worth acknowledging that it's going to fail once in a while and not let an error push you into making changes that might fix that one specific occurrence but cause bigger problems elsewhere.

Cue pokechecker complaining about nitpickiness.

 

[ARM]

Re: 2015 Pairwise Discussion & Predictions

There is that other part of me that thinks not having a champion crowned might be a better deal. A lot of teams could look back on a season and say it was a great season if it were not of the last game they played. I know that is not how humans are put together & fans least of all but it would have been nice so many times to just be happy with a really really good season.

Feel free to quit following the sport for the rest of the season, because over the next month, we'll see a lot of teams being eliminated from contention.

I see no problem with those teams that are eliminated still viewing their season as successful. The real point of the sport is the competition, not the winning. So for that reason, I don't like your idea, because it reduces the competition rather than enhancing it.

 

[KTDC]

Re: 2015 Pairwise Discussion & Predictions

Feel free to quit following the sport for the rest of the season, because over the next month, we'll see a lot of teams being eliminated from contention.

I see no problem with those teams that are eliminated still viewing their season as successful. The real point of the sport is the competition, not the winning. So for that reason, I don't like your idea, because it reduces the competition rather than enhancing it.

Yes, I feel it is the responsibility of the fan and player and coach to maintain perspective on how to judge the season as successful even when losing your last game. Obviously, Frost has addressed this directly. I think it would be crazy making for a program that has a bunch of good years in a row like Minnesota to only think a season has been a success if they win a national championship, despite the temptation of feeling that way and the strong desire to go all the way. Wisconsin, similarly. I bet St Lawrence is feeling their season this year was pretty successful.

A great case is last year's Minnesota team. Was it or wasn't it a success? Would BC's season this year be a success if they don't win out the national playoff? From my distance (and with my bias) obviously it would be.

I will say this though, if your record is 40-0 going into the championship game, it would sure be tough to take losing the final game!

 

[dave1381]

Re: 2015 Pairwise Discussion & Predictions

Maybe I should explain what I mean. I absolutely agree that if there is a system that would better accomplish the goals of the tournament, use it. It would be much better to use a system based upon KRACH than the one we have based on RPI. I can state that unequivocally because those two methods are attempting to measure the same thing and KRACH measures it better.

Ah, I agree with everything you said. Yes, I agree "no perfect system" is a good rationale for having a tournament rather than having a computer pick the winner at the end of the conference tournaments. I agree it's a bad rationale for maintaining the status quo in the selection process, which is how I've more often heard it used, including by a recent NCAA women's hockey selection committee chair.

 

[Numbers]

Re: 2015 Pairwise Discussion & Predictions

Maybe I should explain what I mean. I absolutely agree that if there is a system that would better accomplish the goals of the tournament, use it. It would be much better to use a system based upon KRACH than the one we have based on RPI. I can state that unequivocally because those two methods are attempting to measure the same thing and KRACH measures it better.

When I say that there is no perfect system, I mean two things. The first is that the criteria for deciding who makes the field aren't objective. Different people are going to have different priorities. Some will think that games at the end pf the season should be weighted more heavily than those at the beginning. Some people will think that the field should be large enough to encompass any team that would have a half decent shot should be included; some will think that it should be small and only give teams that had an outstanding regular season entry; some will think that it should only include teams that won their conference because anyone else has already demonstrated that they are not the best team (and this would split between those who think winning a conference should be determined by the regular season and those who think it should be the playoff winner).

Despite the fact that I have strong feelings about some of those decisions, none of them are objectively right or wrong. There is no system that is going to satisfy everyone and no system that really ought to be thought of as perfect.

The second thing is that no system of measuring is going to be perfect. There are going to be times, especially when teams are tightly bunched by the criteria like they are this year, when the result doesn't come out like everyone thinks it should even if they have the same priorities for the criteria. It won't work just right all the time. We should minimize the number of times that it is wrong and the magnitude of error, but it's worth acknowledging that it's going to fail once in a while and not let an error push you into making changes that might fix that one specific occurrence but cause bigger problems elsewhere.

Cue pokechecker complaining about nitpickiness.

This is a great post from a non-emotional angle.

To it I would like to add that even KRACH is not perfect. I agree it's better than RPI (and you can tell that from all the tweaks there are in the RPI formula). It is my understanding that both metrics attempt to measure some sort of a "How good was your record adjusted for the strength of your competition" quantity. I believe that this particular quantity, although it is often sought, is ghostlike in that it is impossible to capture. As a way to understand that comment, consider the Men's this year. Both metrics have 6 of the 8 NCHC teams within in the top 14 schools in the country. And, the reason for that is that the particular Math involved here in both cases makes the non-conference schedule of utmost importance. I know that mathematically, KRACH is perfect.... It will post predict the records of all the teams. But I actually think that the small set of games (1/3 to 1/4 of the year, typically) which actually creates the rankings, is too small to be significant. I would be unable to guess at how to arrange the error bars, but given that to be true, I say there is no perfect to way to adjust record for strength of schedule.

Of course there are other factors as well. In the women's, Wisconsin and Minnesota commonly have very strong teams. Since UND and UMD may play them each 4 or 5 times a year, KRACH seems to lift them (again, the math of KRACH is perfect if you want to post predict - the issue here is error bars in that a small number of games determines the relative rankings). The Women's RPI seems to punish them in a way, because the SOS factors can't over come the 3 or 4 or 5 losses on their own record. There simply is not perfect way to do that. And, so, we have a system. It's not perfect. It is perhaps improvable, although I am not sure how, and if so, it should be improved. But, no improvement will be perfect. And, you know something....? That's ok. In the end this really only games and entertainment.

 

[zoofer]

Re: 2015 Pairwise Discussion & Predictions

Guys, we all know that it's all about the journey, not just the end result....from the other end of the spectrum, after last year, did UNH have an unsuccessful season? For me, it's been one of the more interesting! For the first time in a long time, we had a new Coach to lead the ship....10 new players, new strategies, new everything! Some thought that UNH wouldn't win a single game! If you go back on our (UNH) thread, I predicted that we'd make home ice....while I agree it was aiming high, any athlete will tell you they're ready to compete no matter what! So yea I was hoping for a lot of puck luck per se......
At season's end, UNH finished 6th in league, 5 points from home ice! We played OSU, Dartmouth, Quinnipiac, Wisconsin, BC and BU (x3).....in league opponents, all of which have improved....my point here is every game this season was worth watching a team form from scratch....finding lines that worked, systems that gave us more SOG, goaltenders stepping up to hold the fort with 40 save averages.......you get my drift....
So when you debate whether or not winning a championship is the only way to define a successful season, take a step back, then get a life!

 

[Eeyore]

Re: 2015 Pairwise Discussion & Predictions

I agree. The limitation is on the limited number of non-conference games. It's too small a sample for any statistical system to capture well, even before you get into the problem that teams aren't really identical from one game to the next due to injuries or other factors.

 

[TonyTheTiger20]

Re: 2015 Pairwise Discussion & Predictions

Interesting thing I noticed looking at the Pairwise -- Clarkson can't flip the comparison with BU, because BU has locked up the common opponent piece and was 2-0 against Clarkson this year. So for BU to drop out, they would have to lose an additional comparison. And yet the Terriers are pretty darn close to that bubble.

 

[ARM]

Re: 2015 Pairwise Discussion & Predictions

And yet the Terriers are pretty darn close to that bubble.

Doesn't really matter; they always win the Hockey East tourney.

 

[robertearle]

Re: 2015 Pairwise Discussion & Predictions

Interesting thing I noticed looking at the Pairwise -- Clarkson can't flip the comparison with BU, because BU has locked up the common opponent piece and was 2-0 against Clarkson this year. So for BU to drop out, they would have to lose an additional comparison. And yet the Terriers are pretty darn close to that bubble.

Can things get any closer? Your observation means the answer is 'yes'.

That sets up the possibility of a three-way tie (at least as far as pairwise pair wins is concerned) at 5,6,7 or at 6,7,8.

Say BU's RPI drops below Clarkson and Duluth; Clarkson wins their pair over Duluth, Duluth wins their pair over BU, but BU still wins over Clarkson. Three-way tie.

 

[vicb]

Re: 2015 Pairwise Discussion & Predictions

When Clarkson beats Dartmouth twice this weekend, I think Dartmouth drops out as a TUC. Clarksons TUC instead of being 11-9-3 = .595 drops to 7-9-3 = .447 . If Minn-Dul beats Bemidji 2x they are at 10-10-4 = .500. This flips the TUC comparison and Duluth takes over 7th.

 

[pokechecker]

Re: 2015 Pairwise Discussion & Predictions

C) I'm not convinced that RPI is better or worse than any other system, does not matter if NCAA chose some other system as there is no perfect system

ah, but it does matter grasshopper
if the NCAA started this weekend:
according to RPI
CHA entry plays at BC
Clarkson plays at MN
BU plays at WI
Quinippiac plays at Harvard
according to KRACH BU & Clarkson are out
CHA entry plays at MN
UND plays at BC
UMD plays at WI
Q plays at Harvard
according to Rutter, BU is out
CHA plays at MN
either Quinie or Clarkson plays at BC
UND plays at WI
UMD plays at Harvard


so you see, it does matter, some teams are in, some out
and who plays who is quite different
the RPI seems to give an advantage to the ECAC

 

[CrazyDave]

Re: 2015 Pairwise Discussion & Predictions

so you see, it does matter, some teams are in, some out
and who plays who is quite different
the RPI seems to give an advantage to the ECAC

You assume (very incorrectly) that the selection committee would maintain bracket integrity. We all know how likely that is (less than a snowball's chance in hell).

 

[TonyTheTiger20]

Re: 2015 Pairwise Discussion & Predictions

ah, but it does matter grasshopper
if the NCAA started this weekend:
according to RPI
CHA entry plays at BC
Clarkson plays at MN
BU plays at WI
Quinippiac plays at Harvard
according to KRACH BU & Clarkson are out
CHA entry plays at MN
UND plays at BC
UMD plays at WI
Q plays at Harvard
according to Rutter, BU is out
CHA plays at MN
either Quinie or Clarkson plays at BC
UND plays at WI
UMD plays at Harvard

That is some haiku...

 

[TonyTheTiger20]

Re: 2015 Pairwise Discussion & Predictions

I put together a bracketology post on BCI -- it, like the world in general, is BC-centric, but still some interesting stuff.

http://www.bcinterruption.com/bosto...es-minnesota-golden-gophers-wisconsin-badgers

 

[Numbers]

ah, but it does matter grasshopper
if the NCAA started this weekend:
according to RPI
CHA entry plays at BC
Clarkson plays at MN
BU plays at WI
Quinippiac plays at Harvard
according to KRACH BU & Clarkson are out
CHA entry plays at MN
UND plays at BC
UMD plays at WI
Q plays at Harvard
according to Rutter, BU is out
CHA plays at MN
either Quinie or Clarkson plays at BC
UND plays at WI
UMD plays at Harvard


so you see, it does matter, some teams are in, some out
and who plays who is quite different
the RPI seems to give an advantage to the ECAC

To correct...
RPI...
Clarkson/BC; CHA/Minnesota;BU/Wisconsin; QU/Harvard

KRACH...
CHA/BC; UND/Minnesota; UMD/Wisconsin; QU/Harvard

Rutter...
UND/Minnesota; QU or Clark or CHA/BC; UMD/WI; OtherEast/Harvard

And the initial point still applies. Your ranking system makes a difference. Interestingly, it makes no difference in the seeded teams.

 

[pokechecker]

Re: 2015 Pairwise Discussion & Predictions

Interestingly, it makes no difference in the seeded teams.

my main point was that it does make a difference to 5-8, especially in which team is #7 & #8 and which teams sit home

the rule about travel is what screws up the tournament, how many times does the WCHA champ have to prove their superiority to finally get a crack at the top eastern teams?
prove you are the best thru a 28 game season
then prove it again in a tournament
only to have to prove it again in the NCAA before you get to play an eastern team

it is almost an advantage to be the 2nd best WCHA team, you get to host the first round, and get to play an eastern team which has been fattened on cupcakes

another thing that is really unfair is the HE autobid if the HE champ doesn't win their tourney
all BU has to do is beat one good team to get into the NCAA

OSU-BSU-UND-UMD first have to prevail over one another
then beat WI & MN (IOW, beat 3 top teams )
no wonder the WCHA tourney champ has gone on to win the NCAA 13 of 15 years