PTBx on #3

[stutters]

#3 has been volunteered to test an auto team balance script. please post your feedback here, good and bad. please keep it constructive (i.e. no whining). this may or may not be for a limited time, so i'm counting on all of your feedback.

 

thanks.

[stutters]

i'm leaving all the notices on for the time being. this will give you guys an idea of what's really happening. if it sticks around full time, those notices will probably get turned off.

[mookie]

After several maps played, teams seemed to be pretty good. I'm not sure what things are usually like on #3. 4,000 words:

[bulletmagnet]

it sorta works. gives weird messages sometimes tho. at least it mixes up the teams when its too one sided

[stutters]

alright, so here's how it works.

it finds the avg k:d ratio of the team, multiplied by the number of players on the team. that's team rating. it also looks at the team's score, and factors that in...but only for streaks (atm 2). that sets the stage.

 

now two of the following need to be true:

1. team rating is greater than the minimum rating AND and score difference > 2

2. team rating is greater than the minimum rating AND streak > 2

3. team rating is greater than the minimum rating AND more players

4. team rating is greater than a MAX rating (aka stacked)

5. team rating is a lot greater than a MAX ratin (aka WAY stacked)

if only one of these 5 is true, then it says "obersving advantage." if 2 are true, it switches "someone" that will help balance. i have no idea how that someone is determined.

 

so a team with a low k:d ratio could technically still go 10-0. also, there doesn't seem to be any consideration for favoring ct/t on de/cs maps. purely numbers.

 

at any rate, if you see it NOT balancing, take a screen shot and post. if you have any lag/choke/server issues, please post those too.

 

thanks, again.

 

 

 

 

 

 

if you want a headache, keep reading. otherwise, close your eyes and skip past this:

The most important thing PTB evaluates is team strength. PTB calculates

a single number as an indicator of team strength.

 

In PTB 1.5 the team strength calculation changed. In earlier versions,

team strength was calculated as the sum of total kills per team over the

sum of total deaths per team times the number of players per team

(strength = kills/deaths * players). The method didn't always work well,

since a very good player (20:1) and a very bad player (1:20) made no

difference to 2 mediocre players (5:5 and 6:6).

 

The new method calculates a separate kills/deaths ratio per player, sums

the ratios up per team and multiplies them with team size

(strength = players * (sum of kills/deaths per player)).

 

Example:

CT team: A (6:5), B (2:3)

T team: C (1:1), D (4:3), E (1:4)

CT strength: 2 * (6/5 + 2/3) = 3.73

T strength: 3 * (1/1 + 4/3 + 1/4) = 7.75

 

To compare the team strengths against each other, a rating is calculated

per team by dividing the team's strength by the opposing team's strength

(rating = ownStrength / opposingStrength).

 

Example (continued):

CT rating: 3.73/7.75 = 0.48

T rating: 7.75/3.73 = 2.08

 

Just in case you want to know: the old ratings would have been 0.88 for

the CT team and 1.125 for the T team. Just imagine to add player F (0:10)

to the CT team and look what happens to the old and new ratings. I found

the new ratings to represent the teams much better.

 

But team rating is not all. Even the best team might fail. That's why PTB

doesn't limit itself to those kill/death ratios. PTB also compares the number

of team wins and notes the number of consecutive rounds one by a team as

additional indicators for team efficiency.

 

To decide, if there if there is a winning team situation, PTB then evaluates

the following equations:

 

o team rating >= PTB_MINRATING (stronger team) and score > PTB_MAXDIFF

o team rating >= PTB_MINRATING (stronger team) and streak > PTB_MAXSTREAK

o team rating >= PTB_MINRATING (stronger team) and more players

o team rating >= PTB_MAXRATING (way stronger team)

o team rating >= PTB_SUPERRATING (overwhelming team strength)

 

If any TWO of these criteria hold for one team, it is considered to

be the winning team and actions are taken as appropriate.

 

Note: Teams with a rating below PTB_MINRATING are considered as of equal

or very similar strength. If team strengths are very similar, streaks

and scores are ignored. As soon as a team gets noticibly stronger

than it's opposing team, scores, streaks and team size are checked.

 

PTB balances the teams in one of two ways. If the losing team is smaller,

PTB transfers a player from the winning to the losing team. Else two players

are exchanged (switched) with each other. Valid targets for switches and

transfers are dead players and players, who didn't change teams already

this round. These limitations origin from the fact, that it's not possible

to switch a living player with the help of console commands without killing

him. Additionally CS currently does allow only one team change per round.

Since PTB 1.5 all possible transfers or switches are evaluated for the

switch or tranfer, that balances the teams the most. If no switch or transfer

would improve team balance, switching or transfering is skipped.

 

and here's the current setting (comments were PTB default):

//----------------------

// Team Strength Limits

//----------------------

ptb maxstreak 2

ptb maxscore 2

ptb scorehist 0

ptb minrating 1.3 //1.5

ptb maxrating 1.75 //2.0

ptb superrating 2.5 //3.0

ptb maxincidents 50

ptb scaledown 2

ptb initvalue 5

[witt]

i played for awhile tonight w/ this on, thought it worked pretty well. didn't hear any complaints. What i like is that when someone joins and tries to stack a team, it says something like "Sorry tWc|witt, but the CT team is strong enough" and puts them on the team that needs help. pretty cool.

[bulletmagnet]

is there also a new plugin to reflect team damage onto players? i've been shooting afks ... since they'r afk ... and i got suicided by it, haha

[dEL]

It seemed to help last night when I was playing. Even if ppl stack teams like they do anyways it eventually will balance it out. The only problem might be the time it starts to balance things out the maps halfway over or more.

[Laz.e.rus]

I liked how effective it was at not letting people join a pre-stacked team.

As far as the rest, I switched a guy around ( by his choice) manually becuase he was a single-handed team stacker ( plays for powersgaming).

Its effective in its job of balancing, but I did NOT like playing ct for 2 rounds, learn the approach the Ts are using only to open my eyes at spawn and BE a T.

The number of players was even, but it swapped me as highest map score for a guy who was lowest enemy score.

Its like "congratulations! For the amazing job youve just done, bend over!!"

[stutters]

I liked how effective it was at not letting people join a pre-stacked team.

As far as the rest, I switched a guy around ( by his choice) manually becuase he was a single-handed team stacker ( plays for powersgaming).

Its effective in its job of balancing, but I did NOT like playing ct for 2 rounds, learn the approach the Ts are using only to open my eyes at spawn and BE a T.

The number of players was even, but it swapped me as highest map score for a guy who was lowest enemy score.

Its like "congratulations! For the amazing job youve just done, bend over!!"

<{POST_SNAPBACK}>

roughly, what was the team score and your kd ratio? just curious to see where it's hitting...

 

and keep the feedback coming. this is ONLY a test. it will stay if it's doing the job AND the majority approves.

[Laz.e.rus]

I was 9-1or0 end of second round. Next up was like a 3 or something. Swapped me with a 0-2 I believe.

[Laz.e.rus]

As you saw in that pic in the stacking thread john, it was horribly lopsided when I entered that game.

But what if this was running? Teams were 5-5 or 6-6. Even players but score was 9-0

This prog wouldve balnaced that before I got there, leaving a more balanced score. If I played the same, I would have ( and basically did) unbalance the game badly.

So the prog would have wither kept swapping me all over, OR swapped other players all over as the game went on. Meaning a single person (me) joining leads to peope getting swapped from the team theyve been working on. Bleh.

More time for more experiments.....

[witt]

It may not be perfect, but it's necessary. The sad fact is that most players (not all) are not going to consciously switch over to a team that is getting stomped, or join a team that is getting beaten very badly. Many players only enjoy themselves when they are on the winning side posting a big individual score, and not only will they not switch to the losing team on join the losing team, if presented with an opportunity to switch to the winning team they do so. It's too bad, but people just can't police themselves.

[bulletmagnet]

 

 

 

a couple weird results from today's auto balance

[bulletmagnet]

actually i think the script moved F*E*A*R to T team after i took my screenshot. why would it move an 0-9 and not one of the top people?

Edited December 2, 2005 by bulletmagnet

[stutters]

d2

terrorist rating - .30

ct rating - 3.31

 

logisitic assault

terrorist rating - .07

ct rating - 14.0

 

in this case, two conditions were met. maxrating >= 1.75, super rating >= 2.5. this should trigger a balance to the weaker team. t's should have been helped multiple times. the only thing i can think of is that the t's had more than a 2 man advantage, which prevents PTB from doing anything. i could increase that setting, or decrease the ratings, or scratch my head. for now, i gotta scratch my head. i'll be back from class later tonight, and be able to play with it then.

 

keep the screenshots coming.

[stutters]

ptbx don't work with the new steam update. waiting for a fix, then back to testing.

[stutters]

ptbx is back in action. i'd like to run it for at least a few more days then do a poll to see if more want it to stay or go.

[Vovik]

can u remind me to play on #3 tonite cause I always forget, I don't have it in my favs. I'll be playing for a bit at around 5:30 cst.

 

Yeah, i'm assuming you don't have life.....

 

Meh, forget it, I'll remember it I'm not a complete tard.

[stutters]

you don't have us, one of the gc family, in your favorites?

 

note to self: ask fatty to ban vovik from the #3 forum

[stutters]

settings were just set to be way more aggressive and balance within the 10 rounds. post/aim/pm if it's too much.

 

also, turned the msgs back on for testing...

[Laz.e.rus]

alright, I REALLY dont like this thing.

1- It takes away from the "team" concept; what this game and community are based around by swapping you at any time to either side.

2- It doesnt work worth a HOOT!

 

 

 

 

 

[Magruter]

Agreed with laz, i like to somewhat treat a pub as a team effort, and one map i think i was switch 4 or 5 times. I gets kind of tough to get into a groove. It's a good effort, but i think there are enough regs in the server, that are willing to switch teams to keep them even.

[stutters]

an official (GC only) vote/poll coming. in the meantime...

 

on cs_assault:

ironically, the "TEAM BALANCE IS BLISS" is an advert added a few weeks ago to encourage unstacked teams.

[Vovik]

I didn't have #3 in favs cause of my recent pc troubles. i played yesterday. this thing says teams are fine one round, then on the next one Ts win it says Ts have an advantage, then CTs win it says CTs have advantage (noone moved the team). I mean I don't know, it didn't switch me when I was on and the teams were pretty even so I can't tell for sure. The ping although not horrible still gets me killed in skill-firefights (where reaction is needed). I'll prolly play tonite too for a bit