Forged Alliance Forever Forged Alliance Forever Forums 2012-10-09T15:01:25+02:00 /feed.php?f=2&t=2023 2012-10-09T15:01:25+02:00 2012-10-09T15:01:25+02:00 /viewtopic.php?t=2023&p=21511#p21511 <![CDATA[Re: random setting is not always balanced]]>

Statistics: Posted by uberge3k — 09 Oct 2012, 15:01


]]>
2012-10-08T11:46:24+02:00 2012-10-08T11:46:24+02:00 /viewtopic.php?t=2023&p=21486#p21486 <![CDATA[Re: random setting is not always balanced]]> Statistics: Posted by rootbeer23 — 08 Oct 2012, 11:46


]]>
2012-10-08T11:31:08+02:00 2012-10-08T11:31:08+02:00 /viewtopic.php?t=2023&p=21485#p21485 <![CDATA[Re: random setting is not always balanced]]>
I'm pretty sure it's not in the trueskill code (it come from python and I've a testsuite to find errors),
most probably here :
https://bitbucket.org/thepilot/forged-a ... /lobby.lua
in the function AssignRandomStartSpots.

Feel free to debug and fix it, I will implement the patch when it's done.

EDIT : I think I found the error.
ratingTable is a table that stock all the ratings.
a is the sorted ratingTable.

But later in the code, it's using ratingTable and not the sorted one. So it's kind of random.

I will try to find some time tonight to push the fix.

Statistics: Posted by Ze_PilOt — 08 Oct 2012, 11:31


]]>
2012-10-08T11:28:58+02:00 2012-10-08T11:28:58+02:00 /viewtopic.php?t=2023&p=21484#p21484 <![CDATA[Re: random setting is not always balanced]]>
Ze_PilOt wrote:
Auto-team was enabled ?


we did not pick our teams manually. it was left vs right.

Statistics: Posted by rootbeer23 — 08 Oct 2012, 11:28


]]>
2012-10-08T11:08:47+02:00 2012-10-08T11:08:47+02:00 /viewtopic.php?t=2023&p=21483#p21483 <![CDATA[Re: random setting is not always balanced]]> Statistics: Posted by Ze_PilOt — 08 Oct 2012, 11:08


]]>
2012-10-08T11:07:17+02:00 2012-10-08T11:07:17+02:00 /viewtopic.php?t=2023&p=21482#p21482 <![CDATA[Re: random setting is not always balanced]]>
BRS_Destructor(1700), legendlore(1700), root(1400)
vs
steldoforce(1400), griefer(1400),black_phoenix(1100)

Statistics: Posted by rootbeer23 — 08 Oct 2012, 11:07


]]>
2012-10-08T11:00:24+02:00 2012-10-08T11:00:24+02:00 /viewtopic.php?t=2023&p=21481#p21481 <![CDATA[Re: random setting is not always balanced]]> Statistics: Posted by Ze_PilOt — 08 Oct 2012, 11:00


]]>
2012-10-08T10:56:11+02:00 2012-10-08T10:56:11+02:00 /viewtopic.php?t=2023&p=21480#p21480 <![CDATA[Re: random setting is not always balanced]]> Statistics: Posted by rootbeer23 — 08 Oct 2012, 10:56


]]>
2012-10-08T10:51:42+02:00 2012-10-08T10:51:42+02:00 /viewtopic.php?t=2023&p=21479#p21479 <![CDATA[Re: random setting is not always balanced]]>
I rank the player with their rating (not the gaussian curve).
Then I count the number of 1 in their binary rank.

ie

0 A = 00 = even = team 1
1 B = 01 = odd = team 2
2 C = 10 = odd = team 2
3 D = 11 = even = team 1
....

Unless there is new good players in the team (high deviation = low rating but potentially high skill), it will give you a good combination. Not always the best, but something close to it.

And you can't hardly do better without really computing the game quality for each combination possible.
I guess it's possible to make an algo that dismiss immediately certain combinations from previous result to lower the number of computation, but I don't think it's worth the hassle.

Statistics: Posted by Ze_PilOt — 08 Oct 2012, 10:51


]]>
2012-10-08T10:47:15+02:00 2012-10-08T10:47:15+02:00 /viewtopic.php?t=2023&p=21478#p21478 <![CDATA[Re: random setting is not always balanced]]>
rootbeer23 wrote:
are you trying to arrange the teams in a way that the uncertainty of the rating has approximately the same magnitude
in both teams? If i look at my curve it still gives me a single number that is the best guess for my rating.
can you not compute the curves and then use the best guess as an integer to bruteforce the team composition?


No. I won't be better that the current solution, but waaaaaaay longer and complicated to compute.

Statistics: Posted by Ze_PilOt — 08 Oct 2012, 10:47


]]>
2012-10-08T10:45:52+02:00 2012-10-08T10:45:52+02:00 /viewtopic.php?t=2023&p=21477#p21477 <![CDATA[Re: random setting is not always balanced]]> in both teams? If i look at my curve it still gives me a single number that is the best guess for my rating.
can you not compute the curves and then use the best guess as an integer to bruteforce the team composition?

Statistics: Posted by rootbeer23 — 08 Oct 2012, 10:45


]]>
2012-10-08T10:33:09+02:00 2012-10-08T10:33:09+02:00 /viewtopic.php?t=2023&p=21476#p21476 <![CDATA[Re: random setting is not always balanced]]>
The current solution is simple, fast, and work most of the time (unless you have new players with high deviations but high mean).

Statistics: Posted by Ze_PilOt — 08 Oct 2012, 10:33


]]>
2012-10-08T10:31:37+02:00 2012-10-08T10:31:37+02:00 /viewtopic.php?t=2023&p=21475#p21475 <![CDATA[Re: random setting is not always balanced]]> Statistics: Posted by Gowerly — 08 Oct 2012, 10:31


]]>
2012-10-08T10:27:40+02:00 2012-10-08T10:27:40+02:00 /viewtopic.php?t=2023&p=21474#p21474 <![CDATA[Re: random setting is not always balanced]]>
Gowerly wrote:
12!/6!(6!) = 479001600

That's how many ways there are of arranging 6 people from 12. Will take a while to choose the best one there.


Actually 12! / (12-2) 2, divided / 2 as team A vs team B is the same as team B vs team A.

And i'm wrong in the previous post (written before mu coffee:). But that's not important : Lua can't handle it easily.

Statistics: Posted by Ze_PilOt — 08 Oct 2012, 10:27


]]>
2012-10-08T10:21:15+02:00 2012-10-08T10:21:15+02:00 /viewtopic.php?t=2023&p=21473#p21473 <![CDATA[Re: random setting is not always balanced]]>
That's how many ways there are of arranging 6 people from 12. Will take a while to choose the best one there.

Statistics: Posted by Gowerly — 08 Oct 2012, 10:21


]]>