pls change elo system

[galadion1]

its very hard to hold the elo when i get + 6 for win and -12 for lose.
i Play mostly with Feeder and i think this is not fair
can you pls Balance the elo System, playing with Feeder and noobs is hard enough
Thx

my last game with galadion2: 2 hour game, 74:30 stats and only +5.94 elo

[Max]

If you win more ELO point each victory than you would loose with a defeat you would have an ELO even higher, something like 1700+ after some games. Then what would happen? You will play with more noobs again and again and it will get even worse and harder for you to keep it that high.

Another problem is that league is dying slowly and good players don't play anymore, which makes games more difficult to balance...

I don't think that that kind of change can be done.

[eSVau]

Well, i've thought about the league system. Actually there are two major problems, mass multi accounts and few player base since no one is willing to wait a couple of minutes -
One guy thinks: "ranked takes to long to start, so i join qualify" which causes ranked to start long since this player hasn't joined, so the next guy does the same - <irony>but hey, having leavers all the time must be fun</irony>

About ELO:
I dislike the fact i've created it with a base of 1500. It did not worked that well as i thought it should, most people just oscillate around 1500 with +-50 points, if the have a winning streak to will lose soon a couple of game and being back where they started.
And the fact of so many multi account (mostly of strong players, sometimes top20 were occupied by about 5 distinct people..) caused that the median has dropped below 1500 due multi account grabbing ELO of single accounts and changing because the probably can't bear fair teams.

Now i would make a start value of 0 but with the current way of calculation just a few dozen player would have more than an elo of 0 therefore it requires some way to account experience (read: number of games) and for this i haven't got a suitable (easy) solution without having some awkward ranking (e.g. new player with a couple of wins getting on top, which i would like to avoid).

some conditions a rating should have

• player strength prevails "number of games"
  
• possible to catch up top player
  
• some kind of diminishing return for streaks
  
• nice distribution of ALL player between initial rating and top rating
  
• some points i might have missing
  

But i kinda busy right now so that wont happen fast.

Besides ELO calculation which is meant to be less the higher someone rating is maybe balance might be overhauled again.

• The current one: four worst rating are split between team and the other six distributed most equally possible
  
• We also had: four best ratings are split up
  
• and before that: two best and two worst are split up
  

P.S.: i respect you for playing with such a high ELO and trying to lead a team by advising them

[Flygplan]

Could not the system be changed to be based more on how each player do as indivduals instead of being based on how the team do.

I can see a big problem with ego playing but if supports add as much or neraly as much to your performance score as kills then this will probably reduce this problem.

The advantage with this system would be that players who do play well will get good score even if they lose and the feeders will take most of the hits on both teams.

Elo is made for induvidual playing as far as I know and it will not work as well in a team game like BT even if it do work well.

[Prog]

This was brought up very frequently and I can just reiterate that a high individual influence leads to egoistic play. I certainly would steal every creep and kill.

[GodStrongArms]

My 2 cents on this : for team shuffle, not only the ELO should be taken in consideration, but also the player average K/D ratio. I saw too much games with a way unbalanced team average K/D ratio between the 2 teams pool of players, and an obvious game outcome as a result of it.

[Flygplan]

My 2 cents on this : for team shuffle, not only the ELO should be taken in consideration, but also the player average K/D ratio. I saw too much games with a way unbalanced team average K/D ratio between the 2 teams pool of players, and an obvious game outcome as a result of it.

The problem is that good K/D ratio don't have to mean that this player is better then one with low K/D.
Some people like to play support and don't have good K/D ratio but may still be much better then the 30-0 cowardly killstealer ego player.

[eSVau]

K/D ratio is a bad value - support/death would be better but also does not covering everything

[GEN_Schwarzkopf]

I agree the game is very often determined by the weakest link.

The weakest link has 2 common issues:
1. New player buys lots of long range weapons, gets lots of farm but cannot defend. I've seen the worst players get the highest individual score in the final game screen (not tank performance stats) but having high tank value, most creeps killed and or most towers killed. They end up with high individual scores and think they are good. Some may even have good k:d.

2. The new player has a 1500 ELO but their play style as described above actually ruins the economy for the rest of the team as they lose gold to them continually hoarding 1300 range weapons, and they continue to die from lack of team defense.

I support your idea of moving the ELO score away from being based on these types of things. ELO should be based on winning. Winning is determined by defending allies and CPs and taking CPs over.

- Rob

[Flygplan]

Maybe we should try a very very simple but
probably very robust system of just basing ranking on a win lose ratio like if I win 5 games and lose 5 games I have a winratio of 50%.
To make the system less about creating new accounts I would suggest a game played only count for one month.

This system should work because really the only thing that counts is how many games you can win, by removing the extra variables ranking becomes all about winning and encourage all players to win and only win, egoplay and such will not get you extra points.

Ranking should in this system be based not only on winratio acctualy but on how many games the player play to avoid that people win one game and then get top rating.
By also basing it on how many games people play maybe the multi account useage can be put on an end and also encourage active playing at ranked.

[galadion1]

what you want to do with 55:18 win rate ?

[Prog]

what you want to do with 55:18 win rate ?

Should be 54:19, but the disconnects are too strong :/

[El_Polacco]

3. Player need at least 1400 elo for Ranked games .....

Typical game in which balance system is unable to do it in a most proper way . Any other idea to fix such imbalanced games? http://league.btanks.net/game.php?gameid=40132

[Althend]

This was brought up very frequently and I can just reiterate that a high individual influence leads to egoistic play. I certainly would steal every creep and kill.

Stealing all creep and kills make u get rank 1 faster but is not mandatory


(*said the one who did not play since many years...)