Corvidius
Joined: Feb 15, 2011
|
  Posted:
Apr 19, 2011 - 00:54 |
|
Just for reference what is the problem with mirror matches? Isn't that by definition the fairest type of game? |
|
|
Were_M_Eye
Joined: Sep 24, 2007
|
  Posted:
Apr 19, 2011 - 00:59 |
|
|
Oli_B
Joined: Apr 24, 2006
|
  Posted:
Apr 19, 2011 - 01:07 |
|
Corvidius wrote: | Just for reference what is the problem with mirror matches? Isn't that by definition the fairest type of game? |
I gather the problem is that the scheduler actively avoids (statistically) setting up games with one race playing themselves. |
|
|
The_Murker
Joined: Jan 30, 2011
|
  Posted:
Apr 19, 2011 - 01:19 |
|
floppedit.. a like mind. And poker player too! ANY official's opinion would be appreciated. And I'd go so far as to say Chaos teams would start to lose ALOT and move way down in thew "relative" rankings if they had to play each other. They'd lose, get beaten on, and whine like the rest of us. Give me a few matches where claw piles aren't a factor and I could use those SPP to thump a brainless chaos player. Skills on a catcher. WOo, crazy talk. |
_________________
Join the wait-list. Watch the action. Leave the Empire. Come to Bretonnia! |
|
Carnis
Joined: Feb 03, 2009
|
  Posted:
Apr 19, 2011 - 01:26 |
|
The_Murker wrote: | floppedit.. a like mind. And poker player too! ANY official's opinion would be appreciated. And I'd go so far as to say Chaos teams would start to lose ALOT and move way down in thew "relative" rankings if they had to play each other. They'd lose, get beaten on, and whine like the rest of us. Give me a few matches where claw piles aren't a factor and I could use those SPP to thump a brainless chaos player. Skills on a catcher. WOo, crazy talk. |
You cant go below 50% by playing mirror matches though, as a race.. You can only go towards 50%.. So doubt the relative rankings would change that much really.. |
|
|
Carnis
Joined: Feb 03, 2009
|
  Posted:
Apr 19, 2011 - 01:28 |
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Apr 19, 2011 - 01:29 |
|
Too much is made of the mirror match modifier I think, does anyone know what it is or how many match-ups it affects?.
Claw teams get beat up plenty.
I do think it's a good idea to get rid of the mirror modifier, encouraging mirror matches is an even worse idea than discouraging them however. |
_________________ Pull down the veil - actively bad for the hobby! |
|
Carnis
Joined: Feb 03, 2009
|
  Posted:
Apr 19, 2011 - 01:30 |
|
JimmyFantastic wrote: | I do think it's a good idea to get rid of the mirror modifier, encouraging mirror matches is an even worse idea than discouraging them however. |
Getting rid of the modifier makes mirror matches encouraged by the bot though, as they are always 50% against each other.. So one factor in the table is always optimal.. That is why the "discouraging factor" is in. |
|
|
RobRoyDuncan
Joined: Apr 15, 2011
|
  Posted:
Apr 19, 2011 - 01:34 |
|
I just played a Chaos Dwarf mirror. I would recommend getting confirmation that this bias exists before breaking out the pitchforks and torches. |
|
|
Sutherlands
Joined: Aug 01, 2009
|
  Posted:
Apr 19, 2011 - 01:50 |
|
The "bias" exists only because there is inherent bias TOWARDS mirror matches. The scheduler takes into accounts historical records of the races against each other. If DEs win 80% of the games against orcs, that's not a very good match-up. Since DEs always win 50% of the matches against other DEs, the scheduler sees that as "fair". The bias is inserted so to even things out, so that it sees other matches at about the same fairness as mirrors. Of course, this is all moot, since what matches you get are MUCH MORE determined by what matches are AVAILABLE when you queue. |
|
|
Woodstock
Joined: Dec 11, 2004
|
  Posted:
Apr 19, 2011 - 02:14 |
|
Afaik the 'mirror'-penalty was removed. Ill ask Christer when I see him.
Btw, if it is still there... I think the influence is far less then any one that complains about wants it to be.. |
|
|
pythrr
Joined: Mar 07, 2006
|
  Posted:
Apr 19, 2011 - 02:34 |
|
yes, it was a tiny modifier in the first place, and i think it got dumped too. |
_________________
|
|
The_Murker
Joined: Jan 30, 2011
|
  Posted:
Apr 19, 2011 - 02:39 |
|
what are you people talking about? Look and ANY Chaos team's last 20 games. They don't play Chaos. Look at any elf team's last 20 games... they play lots and lots of other elves, but not their own kind. Look at any human team, they get stuck with Chaos based teams all the time, and NEVER see another human team. Look at the actual data right in front of you. The lack of CHAOS on CHAOS screams volumes. |
_________________
Join the wait-list. Watch the action. Leave the Empire. Come to Bretonnia! |
|
Woodstock
Joined: Dec 11, 2004
|
  Posted:
Apr 19, 2011 - 02:47 |
|
What you are saying actually proves...... Nothing. Do you have data on what teams were activated during the times those teams got scheduled? How did the other combination of matches score on suitability?
How many teams are actually active in the box? How is that spread over the TV range?... You should stop focusing on what you want to see, and accept that it could actually works with the input that is given. |
|
|
licker
Joined: Jul 10, 2009
|
  Posted:
Apr 19, 2011 - 02:49 |
|
Sutherlands wrote: | The "bias" exists only because there is inherent bias TOWARDS mirror matches. The scheduler takes into accounts historical records of the races against each other. If DEs win 80% of the games against orcs, that's not a very good match-up. Since DEs always win 50% of the matches against other DEs, the scheduler sees that as "fair". The bias is inserted so to even things out, so that it sees other matches at about the same fairness as mirrors. Of course, this is all moot, since what matches you get are MUCH MORE determined by what matches are AVAILABLE when you queue. |
Well this explains it, but what is the rational for historical records? Unless it's each coaches historical record individually. Granted, some matchups are just murder for one team, but isn't that the entire point of blind draws? You don't know what you are going to get?
Is part of the issue small sample size (not in games played, in games available at any scheduling time)? Is a solution to go to 20m starts rather than 15? |
|
|
|