Page 18 of 26

Re: EvenMatchOmni

Posted: Wed Apr 20, 2022 7:03 pm
by McLovin
Something is F'd up.

1. Blue team is losing it badly.
2. I call a balance and red has a higher number than blue.
3. I join, and I'm put on red team. Come on, that aint right.

Image
Image

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 12:55 am
by captainsnarf
What is ConfigPPHDiff set to? Teams are considered even if the PPH difference is less than that. The default is 200 but I think it was set in the ini file also. I thought we made it bigger to account for enyo :roll: If it's something like 700 then there's your problem.

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 11:43 am
by pooty

Code: Select all

[EvenMatchOmni.MutTeamBalance]
bCustomScoring=True
ActivationDelay=25
MinDesiredFirstRoundDuration=0
bShuffleTeamsAtMatchStart=True
;bShuffleTeamsAtMatchStart=False
bShuffleTeamsAtRoundStart=False
bRandomlyStartWithSidesSwapped=True
bAssignConnectingPlayerTeam=True
bIgnoreConnectingPlayerTeamPreference=True
bAnnounceTeamChange=True
bIgnoreBotsForTeamSize=True
bBalanceTeamsBetweenRounds=False
bBalanceTeamsWhilePlaying=False
bBalanceTeamsDuringOvertime=False
bBalanceTeamsOnPlayerRequest=False
bBalanceTeamsOnAdminRequest=True
bDisplayRoundProgressIndicator=False
SmallTeamProgressThreshold=0.500000
SoftRebalanceDelay=10
ForcedRebalanceDelay=30
SwitchToWinnerProgressLimit=0.600000
ValuablePlayerRankingPct=50
RecentBalancingPlayerTime=140
NoBalancingUndoTime=30
MinPlayerCount=4
TeamsCallString=
DeletePlayerPPHAfterDaysNotSeen=90
bDebug=True
PlayerGameSecondsBeforeStoringPPH=150
PlayerMinScoreBeforeStoringPPH=4
TeamImbalancePPHThreshold=400
CustomRegulationPoints=1
CustomOvertimePoints=1
Is it the TeamImbalancePPHThreshold? 400? Maybe should be 300 or even 250? I've seen on most maps 250 is an above average score generally

Does it leverage KnownPlayer multipliers?

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 11:54 am
by pooty
Is it the TeamImbalancePPHThreshold? 400? Maybe should be 300 or even 250? I've seen on most maps 250 is an above average score generally
Just lowered it to 250. Its still not going to be perfect. The main flaw we fight is PPH imbalances. Some players rack up huge points (levi/mino spammers) vs. others with high "skill", node aggressiveness etc. Hopefully this makes it better.

Looking at the code, the mutlipliers happen in team score for balance so that answers my own question.

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 12:06 pm
by captainsnarf
Hmm, that's still too low to explain what happened to McLovin.

1107 > 667+400


The onscreen values don't match the total either. unless we have some negative multipliers something else is going on.

853+372=1225 (even match says 1107)
529+248=777 (even match says 667)

I think there is still a problem with the net code for the balance() call. The numbers it shows aren't right and you need to call it twice to see the real value. It's an actor that gets replicated. If you saw the dinosaur icon thingy in old versions of evenmatchomni then you saw 'balance score actor'. Not sure why it needs to be called twice to replicate once.

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 2:34 pm
by pooty
Saw that before too, where add was just wrong based on math...

Is there a way during balance command to call it twice and only display the last one?
I kind of want to see balance displayed all the time (could you do it like stat net and stat fps on demand).

EvenMatch though doesn't use straight 1 game PPH, its the game, plus overall player average I think..there was a formula for it, I think to avoid skew based on one game super performance...

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 3:13 pm
by Enyo
So, was just wondering last night... now that the scoring seems to be much more consistent match to match than before, is it necessary or even wise to continue using a multiplier for the balancer?

BTW, the scoring really is better now, not just in that it's more accurate. It's much more obvious which players actually attack nodes now. Regardless of some being mino/MAS spam whores, their scores now reflect whether they're hitting nodes or not. Only killing and not attacking nodes now yields a lower score, whereas linking of all sorts (nodes and vehicles) now rewards all linkers.

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 5:16 pm
by pooty
Interesting...I think the whole point of the multiplier was to make sure that Leon, Enyo and Xexx didn't end up all on the same team. Which I think that part is working. The top tier multiplier is 1.4, maybe back that down a bit to 1.3 or 1.25, as it might be throwing off the balance.

So either lower the top multiplier (only really applies to six players), the "good" tier players...
OR just make everything back to just 1.0 (default) and see how it goes

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 5:40 pm
by Enyo
pooty wrote: Thu Apr 21, 2022 5:16 pm Interesting...I think the whole point of the multiplier was to make sure that Leon, Enyo and Xexx didn't end up all on the same team. Which I think that part is working. The top tier multiplier is 1.4, maybe back that down a bit to 1.3 or 1.25, as it might be throwing off the balance.

So either lower the top multiplier (only really applies to six players), the "good" tier players...
OR just make everything back to just 1.0 (default) and see how it goes
I say reset everyone back to 1.0 for now and see. Now I think it's skewing the balance of top players so that leon and I are NEVER on the same team anymore. I don't know if removing the multiplier will change that, but seems like it's making it far less likely that we're ever on the same team, even when other "top tier" players are on. Kinda sucks when you never get the benefit of being on the same team as another good player that uses teamwork.

Re: EvenMatchOmni

Posted: Thu Apr 21, 2022 6:02 pm
by McLovin
pooty wrote: Thu Apr 21, 2022 5:16 pm ...
(only really applies to six players), the "good" tier players...
...
This is the problem right here. Nothing against AnKeeDo, but our 2nd best player, above Enyo, come on?

Image