ELynah Forum

General Category => John Spencer Is Dead => Topic started by: Trotsky on January 08, 2023, 11:31:12 PM

Title: Advice
Post by: Trotsky on January 08, 2023, 11:31:12 PM
I have been agonizing (well...) about how to deal with the ECAC going to the Derp Point System for games.  I think I have found a way to record results in the manner that is most informative and least destructive of continuity with prior results, but y'all are smarter and maybe have a better way.

TBRW had over 4000 legacy records of team seasons before the rot set in.  I would like to not upset those as much as possible.  I propose to keep those records with simple fields W, L, and T, with the definitions we all learned before the (temporary) triumph of Creeping Meatballism.  The new records will be listed as W-L-T (DW,DL,DT) where the Dx stats are the sum of points awarded in Derp each instance (regulation win + overtime win + shootout win).  So, the current ECAC records:


36 Qpc 12-0-0 (12,0,0)
22 Cor  7-3-0  (7,1,0)
21 Hvd  8-2-0  (5,0,0)
20 Prn  6-7-0  (6,2,0)
19 Cgt  7-3-0  (5,0,0)
15 SLU  5-3-0  (5,0,0)
11 Clk  3-5-2  (3,0,0)
10 RPI  3-7-0  (3,1,0)
 9 Uni  3-7-0  (3,0,0)
 7 Brn  2-6-2  (0,0,1)
 5 Drt  1-7-1  (1,0,1)
 5 Yal  1-8-1  (1,0,1)


This allows calculation of all 6 possible results:

RW = DW
RL = L - DL
OTW = W - DW
OTL = DL
SHW = DT
SHL = T - DT

I hope this tells you at a glance everything the old records did, but still adjusts for the Derp points.  Whaddya think?
Title: Re: Advice
Post by: jtwcornell91 on January 09, 2023, 09:09:53 AM
Quote from: TrotskyI have been agonizing (well...) about how to deal with the ECAC going to the Derp Point System for games.  I think I have found a way to record results in the manner that is most informative and least destructive of continuity with prior results, but y'all are smarter and maybe have a better way.

TBRW had over 4000 legacy records of team seasons before the rot set in.  I would like to not upset those as much as possible.  I propose to keep those records with simple fields W, L, and T, with the definitions we all learned before the (temporary) triumph of Creeping Meatballism.  The new records will be listed as W-L-T (DW,DL,DT) where the Dx stats are the sum of points awarded in Derp each instance (regulation win + overtime win + shootout win).  So, the current ECAC records:


36 Qpc 12-0-0 (12,0,0)
22 Cor  7-3-0  (7,1,0)
21 Hvd  8-2-0  (5,0,0)
20 Prn  6-7-0  (6,2,0)
19 Cgt  7-3-0  (5,0,0)
15 SLU  5-3-0  (5,0,0)
11 Clk  3-5-2  (3,0,0)
10 RPI  3-7-0  (3,1,0)
 9 Uni  3-7-0  (3,0,0)
 7 Brn  2-6-2  (0,0,1)
 5 Drt  1-7-1  (1,0,1)
 5 Yal  1-8-1  (1,0,1)


This allows calculation of all 6 possible results:

RW = DW
RL = L - DL
OTW = W - DW
OTL = DL
SHW = DT
SHL = T - DT

I hope this tells you at a glance everything the old records did, but still adjusts for the Derp points.  Whaddya think?

I don't follow the asymmetry between wins and losses, so rather than continuing to stare at it, I can tell you the notation I've used: For games with 3x3 OT and a shootout, there are six possible results: RW, OTW, SOW, SOL, OTL, RL.  The ECAC and IIHF points system gives those 3, 2, 2, 1, 1, 0 points out of a possible 3.  I've found it convenient in the context of that point system to define OW = OTW + SOW and OL= SOL + OTL.  (Note that in international and European hockey results they seem to count the result of the shootout as one goal, so you see results like 3-2(SO) or 3-2(GWS) for "game-winning-shots".)  You can then map it onto more sane point choices by making definitions like W = RW + OTW, T = SOW + SOL, L = OTL + RL, or whatever you like.  So I think in that notation you'd be best off giving the standings as W-L-T (RW-OW-OL-RL).  FWIW, I have not had the attention available to follow it closely, but I thought I read somewhere that OT and shootouts were treated differently in the tie-breaking system.

Playoff games still have 5x5 overtime, right?  So another category POW and POL probably needs to be added, not to mention something to deal with ties in regular season games when they don't bother to have a shootout.

See also http://dx.doi.org/10.13164/ma.2021.13
Title: Re: Advice
Post by: Trotsky on January 09, 2023, 11:09:10 AM
Continue to stare at it.  I think it actually makes sense.  You just need to let it grow in your head, like a tumor.
Title: Re: Advice
Post by: nshapiro on January 09, 2023, 07:21:10 PM
Quote from: TrotskyContinue to stare at it.  I think it actually makes sense.  You just need to let it grow in your head, like a tumor.
https://video.search.yahoo.com/search/video?fr=mcafee&ei=UTF-8&p=arnold+schwarzenegger+it%27s+not+a+tumor+meme&type=E211US0G91723#id=2&vid=eb5eec36374a8ab11c4e39a06089ce49&action=click
Title: Re: Advice
Post by: jtwcornell91 on January 10, 2023, 04:34:36 PM
Quote from: TrotskyContinue to stare at it.  I think it actually makes sense.  You just need to let it grow in your head, like a tumor.

Okay, now I get it.  It would be a little easier to follow if you defined the formulas for your six quantities, which are, in matrix form,

/ W \\   / 1 1 0 0 0 0 \\ / RW\\
| L |   | 0 0 0 0 1 1 | |OTW|
| T |   | 0 0 1 1 0 0 | |SOW|
|DW | = | 1 0 0 0 0 0 | |SOL|
|DL |   | 0 0 0 0 1 0 | |OTL|
\\DT /   \\ 0 0 1 0 0 0 / \\ RL/

But inverting non-symmetric matrices costs extra.
Title: Re: Advice
Post by: Trotsky on January 10, 2023, 04:38:39 PM
Jesus Christ if you start talking about Eigenvalues...
Title: Re: Advice
Post by: ugarte on January 10, 2023, 09:35:31 PM
hey man you're the one who came to Nerdsville and asked for advice
Title: Re: Advice
Post by: Trotsky on January 11, 2023, 02:48:34 PM
Quote from: ugartehey man you're the one who came to Nerdsville and asked for advice
"What is your quest?"
Title: Re: Advice
Post by: Jeff Hopkins '82 on January 11, 2023, 10:58:28 PM
Quote from: Trotsky
Quote from: ugartehey man you're the one who came to Nerdsville and asked for advice
"What is your quest?"

"I seek the Holy Grail"