Great Lakes Region

KnightSlappy · January 30, 2014, 09:37:02 AM

Here's what I have for the GL as of this morning:

RG   ##   WP     wSOS    wRPI    NAT   D3     RRO   CONFER   TEAM
GL   01   0.875   0.559   0.638   009   14-2    0-0   NCAC    Wooster
GL   02   0.833   0.565   0.632   011   15-3    0-0   NCAC    Ohio Wesleyan
GL   03   0.765   0.563   0.614   019   13-4    0-0   NCAC    DePauw
GL   04   0.765   0.545   0.600   026   13-4    0-0   OAC     Marietta
GL   05   0.667   0.574   0.597   032   10-5    0-0   MIAA    Hope
GL   06   0.929   0.481   0.593   038   13-1    0-0   PrAC    St. Vincent
GL   07   0.824   0.505   0.584   048   14-3    0-0   OAC     Mount Union
gl   08   0.778   0.504   0.572   063   14-4    0-0   PrAC    Bethany
gl   09   0.625   0.551   0.569   067   10-6    0-0   OAC     Wilmington
gl   10   0.833   0.478   0.567   070   15-3    0-0   NCAC    Wittenberg
gl   11   0.833   0.466   0.558   084   15-3    0-0   AMCC    Penn State-Behrend
gl   12   0.786   0.467   0.546   106   11-3    0-0   MIAA    Calvin
gl   13   0.611   0.521   0.544   114   11-7    0-0   OAC     John Carroll
gl   14   0.556   0.539   0.543   117   10-8    0-0   OAC     Ohio Northern
gl   15   0.500   0.557   0.543   120   9-9    0-0   OAC     Baldwin Wallace
gl   16   0.500   0.547   0.535   132   9-9    0-0   OAC     Capital
gl   17   0.538   0.532   0.534   135   7-6    0-0   PrAC    Grove City
gl   18   0.533   0.529   0.530   143   8-7    0-0   UAA     Carnegie Mellon
gl   19   0.625   0.493   0.526   150   10-6    0-0   UAA     Case Western Reserve
gl   20   0.444   0.549   0.523   155   8-10    0-0   NCAC    Denison

sac · January 30, 2014, 12:26:05 PM

St. Vincent 13-1
Wooster 14-2
Ohio Wesleyan 15-3
Wittenberg 15-3
PSU-Behrend 15-3
Hilbert 15-3
Mt. Union 14-3
Calvin 10-3**
Bethany 14-4
DePauw 13-4
Marietta 13-4
LaRoche 12-4
-------------------------------- .700
Hope 10-5
Wilmington 10-6
John Carroll 11-7
Trine 9-7
Albion 8-7

A week ago Wooster had a huge cushion I thought, now that's pretty much gone. Its doubtful St. Vincent ends up with more than 2 losses.

No head-to-head games this weekend.

Fifth and Putnam · January 30, 2014, 12:40:49 PM

Does anyone know what most consider a safe national ranking (RPI based) heading into Selection Sunday to be considered a solid lock for a pool C bid should one need it?

KnightSlappy · January 30, 2014, 12:54:24 PM

Quote from: Fifth and Putnam on January 30, 2014, 12:40:49 PM
Does anyone know what most consider a safe national ranking (RPI based) heading into Selection Sunday to be considered a solid lock for a pool C bid should one need it?

The committee doesn't actually use RPI -- and there are other criteria involved -- but I usually think of .600 as a general in/out bubble cutoff.

I'm not sure there is such a thing as a "safe" Pool C, but maybe an RPI in the .630+ range would be that.

sac · January 31, 2014, 02:34:27 PM

First official ranking is February 12.

KnightSlappy · February 02, 2014, 11:59:41 AM

The first set of real rankings will use data through next Sunday. Here are the numbers I have through yesterday.

RG   ##   WP     wSOS    wRPI    NAT   D3     RRO   CONFER   TEAM
GL   01   0.882   0.541   0.626   011   15-2    0-0   NCAC    Wooster
GL   02   0.842   0.544   0.618   016   16-3    0-0   NCAC    Ohio Wesleyan
GL   03   0.688   0.580   0.607   024   11-5    0-0   MIAA    Hope
GL   04   0.722   0.559   0.600   029   13-5    0-0   NCAC    DePauw
GL   05   0.778   0.536   0.597   032   14-4    0-0   OAC     Marietta
GL   06   0.833   0.494   0.579   049   15-3    0-0   OAC     Mount Union
GL   07   0.789   0.508   0.578   050   15-4    0-0   PrAC    Bethany
gl   08   0.933   0.460   0.578   051   14-1    0-0   PrAC    St. Vincent
gl   09   0.647   0.531   0.560   074   11-6    0-0   OAC     Wilmington
gl   10   0.842   0.457   0.553   090   16-3    0-0   AMCC    Penn State-Behrend
gl   11   0.789   0.470   0.550   095   15-4    0-0   NCAC    Wittenberg
gl   12   0.526   0.552   0.545   107   10-9    0-0   OAC     Ohio Northern
gl   13   0.474   0.569   0.545   109   9-10    0-0   NCAC    Denison
gl   14   0.526   0.549   0.543   113   10-9    0-0   OAC     Baldwin Wallace
gl   15   0.800   0.457   0.543   115   12-3    0-0   MIAA    Calvin
gl   16   0.632   0.512   0.542   117   12-7    0-0   OAC     John Carroll
gl   17   0.474   0.563   0.541   120   9-10    0-0   OAC     Capital
gl   18   0.647   0.487   0.527   145   11-6    0-0   UAA     Case Western Reserve
gl   19   0.500   0.527   0.521   157   8-8    0-0   UAA     Carnegie Mellon
gl   20   0.467   0.532   0.516   164   7-8    0-0   PrAC    Grove City

sac · February 02, 2014, 05:24:35 PM

I still haven't seen anything official that says the entire AMCC is in the Great Lakes Region for this season.

However the AMCC and Heartland are both approved for the Great Lakes Region for the 2014/2015 season. That takes us to 59 schools and we should get 9 ranking slots.

Is there a link the 2014 Tournament manual floating around yet?

Dave 'd-mac' McHugh · February 02, 2014, 10:50:00 PM

Page 24 of the Pre-Championship handbook lists the entire AMCC in the Great Lakes Region as noted it would nearly a year ago: www.ncaa.org/sites/default/files/PreChamps_DIII_MBasketball_2014_Revised_12-5-13_0.pdf

That is per what we released nearly a year ago: www.d3sports.com/notables/2013/02/regional-realignment-coming

KnightSlappy · February 03, 2014, 09:07:48 AM

The tricky thing, especially for these early rankings, is knowing how the RAC treat results versus regionally ranked opponents. If I understand correctly, they sort of do a double-iteration for the final rankings, but do the RAC get any indication of who is ranked in other regions?

Dave 'd-mac' McHugh · February 03, 2014, 12:11:54 PM

I have been meaning to ask that... keep forgetting... but I think initial rankings get down and then a double-iteration is done to fine tune them. OR they do the rankings and the national committee then uses the vRRO to adjust accordingly. I will try and get an answer.

sac · February 03, 2014, 02:54:24 PM

I did Hope's OWP with the home/away modifiers yesterday. The modifiers are just crazy.

Top to bottom most "valuable games"

@ Stevens Point 1.250
@ Calvin 1.071
@ Whitewater 1.053
Centre (N) .917
Carthage (N) .765
@ Trine .750
Wheaton (N) .684
@ Thomas More .662
Illinois Wesleyan .625
Trine .450
Edgewood .397
@ Albion .350
Lake Forest .264
@ Alma .221
Adrian .115
Olivet .114

OWP .606
without modifiers .604

I find it very hard to stomach that playing at 9-6 Trine and 9-8 Thomas More is more valuable to your OWP than playing 15-3 Illinois Wesleyan at home. Playing on the road does not make Thomas More and Trine better than Illinois Wesleyan, nor does playing Illinois Wesleyan at home suddenly make them worse than a 15-3 team.

KnightSlappy · February 03, 2014, 03:31:28 PM

Quote from: sac on February 03, 2014, 02:54:24 PM
I did Hope's OWP with the home/away modifiers yesterday. The modifiers are just crazy.

Top to bottom most "valuable games"

@ Stevens Point 1.250
@ Calvin 1.071
@ Whitewater 1.053
Centre (N) .917
Carthage (N) .765
@ Trine .750
Wheaton (N) .684
@ Thomas More .662
Illinois Wesleyan .625
Trine .450
Edgewood .397
@ Albion .350
Lake Forest .264
@ Alma .221
Adrian .115
Olivet .114

OWP .606
without modifiers .604

I find it very hard to stomach that playing at 9-6 Trine and 9-8 Thomas More is more valuable to your OWP than playing 15-3 Illinois Wesleyan at home. Playing on the road does not make Thomas More and Trine better than Illinois Wesleyan, nor does playing Illinois Wesleyan at home suddenly make them worse than a 15-3 team.

Looks like maybe you flipped the multiplier on Albion and Olivet. I have Hope's OWP at .618 with the multiplier.

You're right that 25% is still too high. Probably 10% is closer to where it should be.

And thanks for doing this, it helped me identify an error in my calculations that incorrectly adjusted for teams playing opponents multiple times.

sac · February 03, 2014, 03:39:21 PM

Quote from: KnightSlappy on February 03, 2014, 03:31:28 PM
Quote from: sac on February 03, 2014, 02:54:24 PM
I did Hope's OWP with the home/away modifiers yesterday. The modifiers are just crazy.

Top to bottom most "valuable games"

@ Stevens Point 1.250
@ Calvin 1.071
@ Whitewater 1.053
Centre (N) .917
Carthage (N) .765
@ Trine .750
Wheaton (N) .684
@ Thomas More .662
Illinois Wesleyan .625
Trine .450
Edgewood .397
@ Albion .350
Lake Forest .264
@ Alma .221
Adrian .115
Olivet .114

OWP .606
without modifiers .604

I find it very hard to stomach that playing at 9-6 Trine and 9-8 Thomas More is more valuable to your OWP than playing 15-3 Illinois Wesleyan at home. Playing on the road does not make Thomas More and Trine better than Illinois Wesleyan, nor does playing Illinois Wesleyan at home suddenly make them worse than a 15-3 team.

Looks like maybe you flipped the multiplier on Albion and Olivet. I have Hope's OWP at .618 with the multiplier.

You're right that 25% is still too high. Probably 10% is closer to where it should be.

And thanks for doing this, it helped me identify an error in my calculations that incorrectly adjusted for teams playing opponents multiple times.

I might even do away with the home game penalty altogether.

It seems like we do a lot of math to arrive at a similar answer, and just present more opportunities to make mistakes. I sort of agree there should be some kind of "bonus" for an away game, but the 1.25 and .75 modifiers seem extreme.

KnightSlappy · February 03, 2014, 03:49:04 PM

Quote from: sac on February 03, 2014, 03:39:21 PM
I might even do away with the home game penalty altogether.

It seems like we do a lot of math to arrive at a similar answer, and just present more opportunities to make mistakes. I sort of agree there should be some kind of "bonus" for an away game, but the 1.25 and .75 modifiers seem extreme.

This would be the same as removing the "bonus" for playing a neutral site game, cutting the multiplier in half, and shifting the average to something like .600.

I think you would still want the average/median to be .500, so I think you need to do the opposite to the home games as whatever you're doing to the road games.

ziggy · February 03, 2014, 04:02:14 PM

It makes sense to discount a home game if you're going to "reward" an away game, it just has to be done in a way that makes sense. The multiplier should be such that it accurately represents the difference in playing at home and on the road.

Consider the following win probabilities per Massey:
IWU (75%) at Hope (25%)
IWU (81%) vs (neutral court) Hope (19%)
IWU (88% hosting Hope (12%)

Thomas More (17%) at Hope (83%)
Thomas More (22%) vs (neutral court) Hope (78%)
Thomas More (31%) hosting Hope (70%)

Proves sac's point, which probably everyone would agree with anyway. Any multiplier that makes Thomas More look like a tougher opponent than Illinois Wesleyan is totally out of whack. Hope is a solid favorite over Thomas More regardless of location and a solid underdog to Illinois Wesleyan regardless of location.

It's a limited set of two Hope opponents but we see similar differences in the win probability based on where the game is played. A better real-world set of multipliers looks to be 0.94/1.0/1.06 but a 0.9/1.0/1.1 would probably generally pass the sniff test while allowing for a little more penalty/reward for how a schedule is put together.