The 2015 Pairwise, Bracketology and History Thread

[bigblue_dl]

Re: The 2015 Pairwise, Bracketology and History Thread

Given the current RPI, I think it would be safe to say that:

NoDak and Mankato will end 1/2.
Duluth has a large edge for #3, and will therefore likely end up with a #1 seed.

The other #1 seed is a huge fight between Miami, BU, Denver and Omaha.

BC's RPI seems strong enough that, barring strange results the next 2 weeks, they should be in.

Everyone else better win, and no other seedings are really clear. The #2 band could fall to #3s, and the 3s could move either up to 2s, or even fall out.

Given travel concerns, it appears also that we can say that:
Since Duluth and NoDak appear to be #1s, if Minnesota qualifies, they will go to Fargo. (Only exception being if they are in the same seed band as SCSU, and SCSU is the 'normal' seed for the Fargo region.)

Miami would appear to be headed for South Bend, as well as Bowling Green if they qualify.

Comments from others?

Tech is in that fight for the 4th #1 as well. This weekend will go a long way in determining that, but they are sitting very well with a sweep, and should be right in it with a split. I've run scenarios when Tech loses the next four and they drop to around 10th. So barring a major collapse, they should be in.

 

[Priceless]

Re: The 2015 Pairwise, Bracketology and History Thread

Given the current RPI, I think it would be safe to say that:

NoDak and Mankato will end 1/2.
Duluth has a large edge for #3, and will therefore likely end up with a #1 seed.

The other #1 seed is a huge fight between Miami, BU, Denver and Omaha.

BC's RPI seems strong enough that, barring strange results the next 2 weeks, they should be in.

Everyone else better win, and no other seedings are really clear. The #2 band could fall to #3s, and the 3s could move either up to 2s, or even fall out.

Given travel concerns, it appears also that we can say that:
Since Duluth and NoDak appear to be #1s, if Minnesota qualifies, they will go to Fargo. (Only exception being if they are in the same seed band as SCSU, and SCSU is the 'normal' seed for the Fargo region.)

Miami would appear to be headed for South Bend, as well as Bowling Green if they qualify.

Comments from others?

I haven't tried to simulate future games because the tool at CHN only goes to the end of the season and resets after every single game, which means you have to go back and simulate every game to change one result. No thanks. I loved the simulator at slack.net because you could simulate as many games as you wanted, create and simulate the playoffs, change various results...it's a real shame that site is no longer with us.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

I haven't tried to simulate future games because the tool at CHN only goes to the end of the season and resets after every single game, which means you have to go back and simulate every game to change one result. No thanks. I loved the simulator at slack.net because you could simulate as many games as you wanted, create and simulate the playoffs, change various results...it's a real shame that site is no longer with us.

yes, slack.net was much better...you could easily change the result of a holiday tournament and add your own results for what you thought would happen in conference tournaments. NOt to mention, it was a lot easier to tweak your adjustments after the first run like you said.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

Playoffstatus.com has Minnesota State (94%), North Dakota (90%), Minnesota-Duluth (63%), Michigan Tech (65%) as the likely #1 seeds. The next highest probabilities are Miami (31%), Nebraska-Omaha (22%), Boston U (18%), Denver U (15%)... everyone below Denver is less than 4% chance of being a #1 seed...That is extremely intriguing with all but BU representing a western conference.

 

[Priceless]

Re: The 2015 Pairwise, Bracketology and History Thread

Playoffstatus.com has Minnesota State (94%), North Dakota (90%), Minnesota-Duluth (63%), Michigan Tech (65%) as the likely #1 seeds. The next highest probabilities are Miami (31%), Nebraska-Omaha (22%), Boston U (18%), Denver U (15%)... everyone below Denver is less than 4% chance of being a #1 seed...That is extremely intriguing with all but BU representing a western conference.

Are these based on pairwise?

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

Are these based on pairwise?

Im not really sure:
"The NCAA Hockey Tournament Seedings Probabilities table present the probability that a team obtains different seedings. All future unplayed games are assumed won/lost with a probability based upon relative team strengths."
http://playoffstatus.com/ncaahockey/ncaahockeytournseedprob.html

EDIT:I'm confident the rankings are based on PWR, but I'm not sure they use PWR as a basis for determining relative team strength. Most predictive things like Jim Dahl's site use KRACH to determine relative team strength.

 

[Fighting Sioux 23]

Re: The 2015 Pairwise, Bracketology and History Thread

Are these based on pairwise?

No. These come up every year...

Everybody else has their own National Rankings. Now PlayoffStatus.com does as well. Teams are ranked by PlayoffStatus.com's Meaningful Win Percentage (MWP) calculations, which are based upon a team's win percentage modified by strength of schedule and home/away play. It will take several weeks of play (¼ to ½ of a season) before meaningful results start to appear. This ranking system is new and still under development.

This is their own calculations and their "percentages" relate to that team finishing in the top 4 of the sites "MWP"...it has nothing to do with pairwise.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

No. These come up every year...



This is their own calculations and their "percentages" relate to that team finishing in the top 4 of the sites "MWP"...it has nothing to do with pairwise.

But when they say a team has a 34% probability of being #1, they're talking about being #1 in the pair wise right?

 

[Fighting Sioux 23]

Re: The 2015 Pairwise, Bracketology and History Thread

But when they say a team has a 34% probability of being #1, they're talking about being #1 in the pair wise right?

Well, they are saying that the team has a 34% probability of getting a top 4 MWP ranking. That doesn't mean that they will have a 34% probability of getting a top 4 ranking based on pairwise. The site is useful for some things, but I wouldn't trust it for it predicting pairwise rankings...they use a different system.

 

[Numbers]

Well, they are saying that the team has a 34% probability of getting a top 4 MWP ranking. That doesn't mean that they will have a 34% probability of getting a top 4 ranking based on pairwise. The site is useful for some things, but I wouldn't trust it for it predicting pairwise rankings...they use a different system.

I have actually emailed these guys twice in the past. They make some comment about attempting to use PWR in the future, but it never happens.

 

[Slap Shot]

Re: The 2015 Pairwise, Bracketology and History Thread

FS23 why do you label the conference leaders as AQ when that won't be determined until after the Playoffs? Perhaps I'm picking nits.

 

[mcfarljd]

FS23 why do you label the conference leaders as AQ when that won't be determined until after the Playoffs? Perhaps I'm picking nits.

A place holder I'm assuming for the leagues that may not qualify a team in the top 16 (Atlantic/Big ten)

 

[HarleyMC]

Re: The 2015 Pairwise, Bracketology and History Thread

Well, they are saying that the team has a 34% probability of getting a top 4 MWP ranking. That doesn't mean that they will have a 34% probability of getting a top 4 ranking based on pairwise. The site is useful for some things, but I wouldn't trust it for it predicting pairwise rankings...they use a different system.

It's most likely a simple multiple regression based upon several predictor variables. Agreed, it's not reliable and has absolutely nothing to do with pairwise procedures.

 

[BUPhD]

Re: The 2015 Pairwise, Bracketology and History Thread

A place holder I'm assuming for the leagues that may not qualify a team in the top 16 (Atlantic/Big ten)

I'd just like to register my {choose one: Shock, amusement, concern} that these two conferences are mentioned in the same breath for anything.

 

[mnstate0fhockey]

I'd just like to register my {choose one: Shock, amusement, concern} that these two conferences are mentioned in the same breath for anything.

Big Ten only has 6 teams. Much easier to have a down season where all teams are down.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

So playoffstatus.com is far more like someone basing everything on KRACH and then ranking teams by KRACH vs what Jim Dahl does where he uses KRACH to assign probability of outcomes but then ranks teams after those results by PWR?

 

[Numbers]

So playoffstatus.com is far more like someone basing everything on KRACH and then ranking teams by KRACH vs what Jim Dahl who uses KRACH to assign probability of outcomes but then ranks teams after those results by PWR?

This is a good description. Although we really don't know what their metric is. My guess is that it's not as good as KRACH

 

[bigblue_dl]

Re: The 2015 Pairwise, Bracketology and History Thread

This is a good description. Although we really don't know what their metric is. My guess is that it's not as good as KRACH

When comparing their rankings to Krach, it is similar, but not exact. There are several teams that are swapped with others, for example, UND and MSU are flipped between the 2 rankings, as are UMD and MTU. In the 9-12 range it is a little more different, but not extremely so. Krach is probably better, but the rankings this method comes up with are at least comparable. Like I told Shirtless Guy this morning, you can probably look at playoffstatus.com's percentages and apply a +/-, of maybe 5%, to each one and it would fall in that range if the same thing was done with PWR. Don't know what that tolerance would have to be without more analysis of how accurate these predictions were from past years.

 

[Numbers]

When comparing their rankings to Krach, it is similar, but not exact. There are several teams that are swapped with others, for example, UND and MSU are flipped between the 2 rankings, as are UMD and MTU. In the 9-12 range it is a little more different, but not extremely so. Krach is probably better, but the rankings this method comes up with are at least comparable. Like I told Shirtless Guy this morning, you can probably look at playoffstatus.com's percentages and apply a +/-, of maybe 5%, to each one and it would fall in that range if the same thing was done with PWR. Don't know what that tolerance would have to be without more analysis of how accurate these predictions were from past years.

Mostly true. What i am not sure about is if these guys really understand things like auto bids.

By the time we are done, the tournament bubble could look vastly different, and since these guys don't have the proper system, I don't think it's worth my time to consider their predictions.

Now, where are Jim Dahl and Reilly Hamilton? I want their predictors.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

Mostly true. What i am not sure about is if these guys really understand things like auto bids.

By the time we are done, the tournament bubble could look vastly different, and since these guys don't have the proper system, I don't think it's worth my time to consider their predictions.

Now, where are Jim Dahl and Reilly Hamilton? I want their predictors.

They obviously have some clue about autoqualifiers or Robert Morris wouldn't have a 36% probability of being 16th with the other highest probability for #16 are all in the AHA with Bentley at 18%, Mercyhurst at 12%.

The B1G schools that likely need to win their tournament to get in (PSU/MSU) both have relatively high probabilities of being a 15th PSU - 15%, MSU - 9% so it sure seems like they understand that aspect of things.