Author Topic: 2016 Bracketology  (Read 11579 times)

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
2016 Bracketology
« on: February 16, 2016, 11:19:22 pm »
Welp. Away we go.

Round 1.
"It is hard to free fools from the chains they revere" - Voltaire

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #1 on: February 16, 2016, 11:30:15 pm »

Great write up on the tourney selection process (as always).  I'm curious if you (or anyone else) has any insight as to why they would dowbgrade L25 to secondary criteria after just two years?

Dyce's answer is top-flight and explains the reasons why L25 shouldn't be a primary criterion, but no one ever thought that deeply about it. The change came about due to fears that too much emphasis was placed on something that was new and shiny and just not as important as the rest. There were significant fears two years ago that all the committees cared about was L25, and I think they sought to alleviate the chance it received disproportionate significance. So ultimately I think they got it right, albeit without correctly processing it in the manner Dyce did.
« Last Edit: February 16, 2016, 11:56:56 pm by Matthew Webb »
"It is hard to free fools from the chains they revere" - Voltaire

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #2 on: February 16, 2016, 11:34:43 pm »
Pardon my ignorance here (and that may take a few minutes), but is there a place where I can look at comparisons, or what the NCAA uses for strength of schedule? Or do I have to wait for Wednesday's rankings?

The NCAA calculates SOS in a manner that is a complete insult to the concept of mathematics, but nonetheless....give me a second here and I'll upload the actual data sheets to the site and link them here. I see the current ones are still up at ncaa.com, which is nice to see as for the past four years they'd update on Tuesday afternoon and the revert to 2011 after about 45 minutes. Nonetheless, I'll upload them just to be safe.
"It is hard to free fools from the chains they revere" - Voltaire

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #3 on: February 16, 2016, 11:54:42 pm »
"It is hard to free fools from the chains they revere" - Voltaire

Offline spwood

  • Second-stringer
  • **
  • Posts: 155
  • Karma: +7/-1
    • View Profile
Re: 2016 Bracketology
« Reply #4 on: February 18, 2016, 08:17:17 am »
I'm just stunned at how low Plattsburgh's SOS is compared to the other teams, most notably the ones in their own league.  Buffalo?  I assume that Utica, Elmira, and St. Norbert is that much better than Williams and Norwich?  I hope Plattsburgh can improve the teams in their own tourney going forward as well...

Thanks for the hard work, Matt.  It's definitely appreciated!

Offline PSUChamps2001

  • Junior Varsity
  • *
  • Posts: 13
  • Karma: +2/-0
    • View Profile
Re: 2016 Bracketology
« Reply #5 on: February 18, 2016, 11:05:25 pm »
Just as I put somewhere else, Middlebury, Norwich, and Oswego (6 games) have a combined 36 loses and ties this year.

Here's another thing to ponder about SOS:

Who are the top of the SOS category?

ECAC West 6 of the Top 10

And look who the "other" SUNYAC teams have played....

Buff State NC Schedule (14th overall)
Nazareth (1)
Canton (wash 39)
Canton (wash 39)
St.Norbert (5)
St.Norbert (5)
Utica (6)
Elmira (4)
Johnson&Wales (57)
Daniel Webster (67)
Avg 24.8

Geneseo Schedule (13th overall)
Canton (wash 39)
Canton (wash 39)
Williams (wash 9)
Nazareth (1)
Hobart  (8)
Elmira (4)
Neumann (11)
Neumann (11)
Franklin Pierce (DII)
Avg 15.3

Plattsburgh NC Schedule
Canton (wash 39)
Middlebury (22)
Norwich (25)
Fitchburg (72)
Colby (15)
Norwich (25)
Canton (wash 39)
Middlebury (22)
Williams (9)
Avg 29.8

Just to break it down even more:
Average Conference SOS Rank
ecac west            6.166666667
nescac            22.8
ncha            25.54545455
wiac            35
sunyac            28.66666667
nehc            36
ecac ne            55.88888889
miac            60.33333333
mascac            72.42857143

6 to 22 is a pretty big jump. Some math genius can come up with a reason for it, but something tells me it has to do (somewhat) with playing each other 3 times in a season and not having that weak link in the league. Hobart is a pretty good team, but outside of them the league is (on the ice) average. Now I'm not saying MASCAC type average, BUT Im not sure Id consider another ECAC West team a contender

A better Cardinal Classic would have helped "some", but we've also now given up games to Skidmore and Castleton for Canton (which is improving IMO but still lower then the prior mentioned). Combine that with the down years of the "big 3" (Norwich, Midd and Oswego) and you have a hole in Plattsburgh's SOS, which at 23rd overall isn't really unimpressive, but its not the Top 10 like we're use to. 36 loses and ties among the Big 3 is just huge this year.




Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #6 on: February 18, 2016, 11:49:19 pm »
6 to 22 is a pretty big jump. Some math genius can come up with a reason for it, but something tells me it has to do (somewhat) with playing each other 3 times in a season and not having that weak link in the league.

This is exactly it. Give me a few here and I'm going to try to figure out how to explain why in a relatively simple manner...
"It is hard to free fools from the chains they revere" - Voltaire

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #7 on: February 19, 2016, 12:42:11 am »
The short version is that the best way to ramp up SOS is to have the entire league beat up on others in non-conference play and then roll up their SOS by playing a bunch of league games against teams with relatively good records. If 80% of the league is over .500 it's going to be a SOS boon for everyone in the league, especially since head-to-head results get dropped by NCAA when calculating SOS with the adjusted records. For example, as it pertains to Hobart's SOS this year, its three wins over Utica are dropped for the NCAA calculations so Utica's contribution to Hobart's SOS is 11-6-4 (.619) times three. Not a bad deal, and thus Hobart sits with a lofty .5390 SOS that has it glued into the Pool B driver's seat.

Now, this comes with a caveat. There's a magic spot where this will have the greatest effect. If you look at the numbers over time, I wager it's somewhere around a 14-16 game league schedule, and if I had to settle on one I'd guess 16 was optimum. Get up near 20 like the NCHA has and you slowly whittle away any advantage racked up in non-conference play, while if you play something like this eight game schedule the WIAC fiddled around with this year any non-conference success is never realized to its full potential.

And in men's DIII, this is historically where we have seen the highest SOS figures. It's come out of ECAC West, old NCHA, and at times the NESCAC, when they have had deep leagues that do very well in non-conference play and then play 14-18 game league schedules.

Back in the day I wrote a massive weekly column for USCHO that focused on the NCHA reducing its league schedule. I ran out an entire fictitious season based on the previous season's results but adjusted the schedule to match the reduced number of league games. SOS numbers crashed. Naturally, this was completely ignored by everyone, which is expected and all good :), but sure as hell, next thing you know the NCHA becomes a one-bid league while the MIAC gets two despite getting absolutely trashed by the NCHA in non-conference play those years.

A more recent comparison is what the WIAC did this year. Dropping to this eight-game schedule has led to significantly lower SOS values this year compared to last, and it very, very well could come back to bite it. Eau Claire and Stevens Point had big SOS values last year and they're now way down, and there's only one reason for it. The distribution of records in the WIAC is quite similar to those of the ECACW this year (might even be a tad better), but the meager league schedule meant the results of non-conference success were never really capitalized on.

The amazing thing is the same person/people that drove the reduced NCHA schedule that prompted that USCHO column from me did so because they thought it was going to jack SOS. Now the same thing has happened with the WIAC, due to the same person/people and they did it for the same reason thinking it'd make the WIAC a "three-bid league." Sure, and your sixth-grade math teacher would like a word with you. It's madness, and they're evidently incapable of learning from the previous lead balloon.

A time will come for me to uncork on this, but we'll see what Selection Sunday brings...

Hopefully that made sense. That's the layman's explanation. Wiitala could probably give a pretty hardcore statistical version...
« Last Edit: February 19, 2016, 01:27:32 am by Matthew Webb »
"It is hard to free fools from the chains they revere" - Voltaire

Offline spwood

  • Second-stringer
  • **
  • Posts: 155
  • Karma: +7/-1
    • View Profile
Re: 2016 Bracketology
« Reply #8 on: February 19, 2016, 08:05:19 am »

Hopefully that made sense. That's the layman's explanation. Wiitala could probably give a pretty hardcore statistical version...

For the love of God, it's way to early to unleash the mad scientist on us!!!!

Offline PSUChamps2001

  • Junior Varsity
  • *
  • Posts: 13
  • Karma: +2/-0
    • View Profile
Re: 2016 Bracketology
« Reply #9 on: February 19, 2016, 11:06:43 am »
Thanks Webb lol That's what I've argued for years on the other site with the ECAC West and their SOS. Was never my intentions to call the league "weak", but 6 to 22.....it just raises a big question of the fairness of SOS. I do think their overall NC record has become more competitive, and as I put years ago, have stopped having the majority of their NC schedule against the lower 1/4 of the leagues bottom teams. Having said that:

(SOS RNK) / Average NC SOS RNK
Nazareth (1) / 22.8 (only 9 with 1 DII game)
Elmira (4) 25.7
Hobart (8) / 28.5
Utica (6) / 32.7
M'Ville (7) /34.8
Neumann (11) / 39.1

Average SOS RNK 6.2
Average NC SOS RNK 30.6

It still boggles my mind how a conference who's Average SOS RNK is 6.2 can have a non-conference RNK of 30.6. Granted I haven't gone through the other divisions yet, but I just might....

Offline spwood

  • Second-stringer
  • **
  • Posts: 155
  • Karma: +7/-1
    • View Profile
Re: 2016 Bracketology
« Reply #10 on: February 24, 2016, 08:50:38 am »
Second Bracketology is up and hard to argue with the field or how the bracket fell.  One question though, the article compares three teams for two spots.  In the actual process, does it work like bouncy ball that only the top team left in each region is up for comparison?  (i.e. that Buffalo doesn't come to the table until Trinity is placed in the field)?

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #11 on: February 24, 2016, 10:05:29 am »
Week 2 Bracketology

East Region numbers

West Region numbers

Note that the RNK records are off for some teams in the East as they are still based off of the Feb. 16 regional rankings.

"It is hard to free fools from the chains they revere" - Voltaire

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #12 on: February 24, 2016, 10:07:48 am »
Second Bracketology is up and hard to argue with the field or how the bracket fell.  One question though, the article compares three teams for two spots.  In the actual process, does it work like bouncy ball that only the top team left in each region is up for comparison?  (i.e. that Buffalo doesn't come to the table until Trinity is placed in the field)?

More than two teams are up at a time. I don't know how many but it's more than two. We speculated that St. Thomas' mysterious meteoric rise in 2012 was evidence of that, and it is indeed the case.

Back in the day where they'd go by straight pairwise comparisons it was done in the manner you refer to, but at some point that fell out of fashion. Probably because there are just two regions in hockey.
"It is hard to free fools from the chains they revere" - Voltaire

Offline PeterLangella

  • Second-stringer
  • **
  • Posts: 69
  • Karma: +2/-0
    • View Profile
Re: 2016 Bracketology
« Reply #13 on: February 25, 2016, 12:46:35 pm »
Have you done any playing around with how the numbers look after the upcoming games? Who can afford another loss? Who can't? Which league other than the NCHA has the best chance to get two teams in when another loss is added to team #2? My gut is telling me that if Geneseo or Buff St. loses to Platts in the SUNYAC final, they will leapfrog the NESCAC final loser.

Offline Matthew Webb

  • Moderator
  • Starter
  • *****
  • Posts: 408
  • Karma: +13/-1
    • View Profile
Re: 2016 Bracketology
« Reply #14 on: February 26, 2016, 02:28:52 am »
Have you done any playing around with how the numbers look after the upcoming games? Who can afford another loss? Who can't? Which league other than the NCHA has the best chance to get two teams in when another loss is added to team #2? My gut is telling me that if Geneseo or Buff St. loses to Platts in the SUNYAC final, they will leapfrog the NESCAC final loser.

Yes. Sort of sure. Sort of sure. Probably the NESCAC. Possible.

Since I'm of the mind that a lot of this doesn't matter until the final weekend I try to not get too far ahead, but last spring I wrote a desktop app that I can use to calculate the expected impact of future games on a team's SOS. Now, it isn't perfect as it's obviously not possible to account for OOWP in advance, but it's good within a few thousandths so it's a solid guide. If we look at who is all left we see most everyone will be getting some sort of tick up in SOS, though some more than others. All four NCHA teams as well as Eau Claire and Stevens Point will see a greater jump than others due to the fact they are all playing twice this weekend.

As far as who can afford another loss, I only thing two, maybe three, are in really good shape right now. I think Adrian and St. Norbert are safe, especially St. Norbert. Haven't bothered looking into if neither win the NCHA but let's assume one of them does -- they're both going to be in the field. Hobart would be the third but if it loses in the semifinals and Point wins out the Statesmen could very well cough up the B. It's about the only way it happens, but it's possible. But let's play the odds here and say Hobart is pretty safe. Beyond that...tough to say. We've seen so many strange things happen in the regional rankings at late stages (hi, Trinity) that we can never be too sure. But anyway, the only teams I'd call locks right now are Adrian and St. Norbert, and I'd consider Hobart to be the next-closest.

Who can't afford another loss. Stevens Point. The Pointers are the linchpin to this entire thing. If they doink it away this weekend they should be finished per all practical analysis. In which case everyone in the country should send them a thank you card. Point is likely the difference between an 8-3 and 7-4 split, a loss solidifies Pool B but also brings a bolstered Eau Claire into the mix and who knows what happens from there. Beyond that, I'd look at the regional rankings. I'd wager Plattsburgh's situation is more precarious than some might suspect (especially should it lose in semis), but we can throw in Eau Claire, Marian, Trinity, Geneseo and Babson. I have no idea what to make of Buff State, but leaving it aside for the moment I suspect its win and in for most of the rest as none could be overly comfortable with their spot after loss.

I guess that's really just a long way of saying there are 2.5 teams right now that I think can definitely handle a loss. Williams is in a solid enough position that I'll leave it out of the absolutely has-to-win category. The Ephs also could be sitting on a trump card as they stole a win from Plattsburgh despite being outshot 100-4 or whatever it was.

As to your final point, considering the way the regional rankings are generated that wouldn't surprise me one bit. I wouldn't say I would expect it, but I'd say it wouldn't surprise me. Besides, isn't it becoming an annual tradition for Trinity get tanked? Or will this be Williams' turn?
« Last Edit: February 26, 2016, 02:37:29 am by Matthew Webb »
"It is hard to free fools from the chains they revere" - Voltaire